hidekazu-konishi.com

AI and Machine Learning Glossary for AWS - Knowledge Gained While Studying for AWS Certified AI Practitioner and AWS Certified Machine Learning Engineer - Associate
适用于 AWS 的 AI 和机器学习词汇表 – 在学习 AWS Certified AI Practitioner 和 AWS Certified Machine Learning Engineer – Associate 期间获得的知识

First Published:   首次发布：2024-11-24
Last Updated:   上次更新时间：2024-11-25

This time, I have compiled the knowledge gained during my study process to pass the newly added AWS certifications, AWS Certified AI Practitioner and AWS Certified Machine Learning Engineer - Associate, into a "Glossary of AI and Machine Learning Terms Related to AWS".
这一次，我将学习过程中获得的知识汇编成“与 AWS 相关的 AI 和机器学习术语表”，以通过新添加的 AWS 认证，即 AWS Certified AI Practitioner 和 AWS Certified Machine Learning Engineer - Associate。

The knowledge in this "Glossary of AI and Machine Learning Terms Related to AWS" is also used in the questions and answers of "Learning AWS Functions and History Through Quizzes: Selected 'Machine Learning' Edition" in "Compilation of Thin Books on AWS Vol.01", which I co-authored as an individual publication for Japan's "Technical Book Fair 17".
本“与 AWS 相关的 AI 和机器学习术语词汇表”中的知识也被用于“Compilation of Thin Books Vol.01”中的“通过测验学习 AWS 函数和历史记录：精选'机器学习'版”的问答中，我与人合著了该书作为日本“Technical Book Fair 17”的个人出版物。

I hope this will be helpful for those who are preparing to take the AWS Certified AI Practitioner and AWS Certified Machine Learning Engineer - Associate exams.
我希望这对准备参加 AWS Certified AI Practitioner 和 AWS Certified Machine Learning Engineer – Associate 考试的人有所帮助。

AI/ML AWS Services  AI/ML AWS 服务

Amazon SageMaker  亚马逊 SageMaker

Service Name   服务名称	Description   描述
Amazon SageMaker   亚马逊 SageMaker	A fully managed service for efficiently building, training, and deploying machine learning models. An integrated platform that supports the entire ML lifecycle from development to production operation. 一项完全托管的服务，用于高效构建、训练和部署机器学习模型。一个集成平台，支持从开发到生产运营的整个 ML 生命周期。
SageMaker Studio   SageMaker 工作室	One of SageMaker's components, a browser-based integrated development environment (IDE). Enables one-stop execution from notebook creation to model development, training, and deployment. Achieves centralized management of ML workflows. SageMaker 的组件之一，基于浏览器的集成开发环境 （IDE）。实现从 notebook 创建到模型开发、训练和部署的一站式执行。实现 ML 工作流的集中管理。
SageMaker Canvas   SageMaker 画布	One of SageMaker's components, a visual interface that allows building ML models through drag & drop without writing code. A no-code ML development environment for business analysts. SageMaker 的一个组件，一个可视化界面，允许通过拖放构建 ML 模型，而无需编写代码。面向业务分析师的无代码 ML 开发环境。
SageMaker Ground Truth	One of SageMaker's components, a data labeling service for creating high-quality training datasets. Improves efficiency of human labeling tasks and provides semi-automated labeling workflows. SageMaker 的组件之一，用于创建高质量训练数据集的数据标记服务。提高人工标记任务的效率，并提供半自动化的标记工作流程。
SageMaker Data Wrangler   SageMaker 数据管理员	One of SageMaker's components, a tool for streamlining data preparation and preprocessing. Provides over 200 built-in transformation functions, enabling data cleansing to feature engineering via GUI. SageMaker 的组件之一，用于简化数据准备和预处理的工具。提供 200 多个内置转换函数，可通过 GUI 将数据清理为特征工程。
SageMaker Feature Store   SageMaker 特征存储	One of SageMaker's components, a repository for centrally managing and sharing features. Ensures consistency of features in online/offline scenarios and promotes reuse across teams. SageMaker 的组件之一，用于集中管理和共享功能的存储库。保证线上线下场景下功能的一致性，促进团队复用。
SageMaker JumpStart   SageMaker 快速启动	One of SageMaker's components, an ML hub providing pre-trained models and solutions. Deployable with one click and supports transfer learning and fine-tuning. SageMaker 的组件之一，一个提供预训练模型和解决方案的 ML 中心。一键部署，并支持迁移学习和微调。
SageMaker Model Monitor   SageMaker 模型监控器	One of SageMaker's components, continuously monitors model quality in production environments. Detects data drift and bias, enabling early detection of model performance degradation. SageMaker 的组件之一，持续监控生产环境中的模型质量。检测数据漂移和偏差，从而及早发现模型性能下降。
SageMaker Clarify   SageMaker 澄清	One of SageMaker's components, evaluates bias detection and explainability of models. Ensures fairness and transparency, and analyzes the basis for model decisions. SageMaker 的组件之一，评估模型的偏差检测和可解释性。确保公平性和透明度，并分析模型决策的依据。
SageMaker Debugger   SageMaker 调试器	One of SageMaker's components, for debugging and monitoring the training process. Enables visualization of metrics and setting of alerts. Supports optimization of learning. SageMaker 的组件之一，用于调试和监控训练过程。启用指标可视化和警报设置。支持学习优化。
SageMaker Pipelines   SageMaker 管道	One of SageMaker's components, for orchestration of ML workflows. Builds reproducible ML pipelines and enables automated experiment management. SageMaker 的组件之一，用于编排 ML 工作流。构建可重现的 ML 管道并实现自动化实验管理。
SageMaker Model Cards   SageMaker 模型卡	One of SageMaker's components, for creating and managing model documentation. Centralizes management of detailed model information, ensuring governance and compliance. SageMaker 的组件之一，用于创建和管理模型文档。集中管理详细的模型信息，确保治理和合规性。
SageMaker Role Manager   SageMaker 角色管理器	One of SageMaker's components, manages access permissions for ML activities. Implements security based on the principle of least privilege and provides appropriate access control. SageMaker 的组件之一，管理 ML 活动的访问权限。根据最小权限原则实现安全性，并提供适当的访问控制。
SageMaker Experiments   SageMaker 实验	One of SageMaker's components, a tool for tracking and managing machine learning experiments. Automatically records experiment results such as training runs, parameters, and metrics, enabling comparative analysis. SageMaker 的组件之一，用于跟踪和管理机器学习实验的工具。自动记录实验结果，例如训练运行、参数和指标，从而进行比较分析。
SageMaker Model Registry SageMaker 模型注册表	One of SageMaker's components, a repository for cataloging and versioning ML models. Manages model metadata and approval status. SageMaker 的组件之一，用于对 ML 模型进行编目和版本控制的存储库。管理模型元数据和审批状态。

Amazon Bedrock  亚马逊基岩版

Service Name   服务名称	Description   描述
Amazon Bedrock   亚马逊基岩版	A secure, fully managed generative AI platform service. An integrated platform that allows access to multiple Foundation Models through a single API. Provides comprehensive features including Foundation Models (FMs) as the underlying large language models, Knowledge Bases for RAG construction, Agents for automation, Guardrails for harmful content filtering, and Prompt Flows for workflow execution. 安全、完全托管的生成式 AI 平台服务。一个集成平台，允许通过单个 API 访问多个基础模型。提供全面的功能，包括作为底层大型语言模型的基础模型 （FM）、用于 RAG 构建的知识库、用于自动化的代理、用于有害内容过滤的护栏以及用于工作流执行的提示流。
Foundation Models (FMs)   基础模型 （FM）	Large language models that serve as the foundation for text generation, image generation, etc. The core AI component providing the foundational functionality in Bedrock. 作为文本生成、图像生成等基础的大型语言模型。在 Bedrock 中提供基础功能的核心 AI 组件。
Knowledge Bases   知识库	A Bedrock feature that provides information retrieval from external knowledge bases, enabling the construction of Retrieval Augmented Generation (RAG) architectures. 一种 Bedrock 功能，可提供从外部知识库进行信息检索，从而支持构建检索增强生成 （RAG） 架构。
Agents   代理	A Bedrock feature that orchestrates procedural instructions, custom action execution, and Knowledge base utilization in an integrated manner. 一项 Bedrock 功能，以集成方式编排程序指令、自定义作执行和知识库利用。
Guardrails   护栏	A Bedrock feature that detects and filters harmful content and hallucinations, controlling AI output. 一项 Bedrock 功能，可检测和过滤有害内容和幻觉，从而控制 AI 输出。
Prompt Flows   提示流	A Bedrock feature that systematically workflows prompt execution, S3 data input/output, and Lambda function execution. 一项 Bedrock 功能，可系统地工作流提示执行、S3 数据输入/输出和 Lambda 函数执行。

Amazon Q  亚马逊 Q

Service Name   服务名称	Description   描述
Amazon Q   亚马逊 Q	A generative AI-powered assistant service specifically designed for businesses. It comes in Business and Developer editions, specializing in business productivity improvement and development support respectively. It can be integrated with various AWS services, including Amazon Q in Amazon QuickSight, Amazon Q in Amazon Connect, Amazon Q in AWS Chatbot, Amazon Q network troubleshooting, and Amazon Q Data integration in AWS Glue. 专为企业设计的生成式 AI 驱动的助手服务。它提供 Business 和 Developer 版本，分别专注于业务生产力提升和开发支持。它可以与各种 AWS 服务集成，包括 Amazon QuickSight 中的 Amazon Q、Amazon Connect 中的 Amazon Q、AWS 聊天机器人中的 Amazon Q、Amazon Q 网络故障排除以及 AWS Glue 中的 Amazon Q 数据集成。
Amazon Q Business   亚马逊 Q 企业购	A generative AI assistant designed to improve employee productivity. Supports automation and efficiency of general business tasks. 旨在提高员工工作效率的生成式 AI 助手。支持一般业务任务的自动化和效率。
Amazon Q Developer   Amazon Q 开发人员	A generative AI assistant specialized in coding support for developers. Supports development tasks such as code generation, debugging, and optimization. 专门为开发人员提供编码支持的生成式 AI 助手。支持代码生成、调试和优化等开发任务。

Natural Language Processing Services
自然语言处理服务

Service Name   服务名称	Description   描述
Amazon Comprehend   亚马逊 Comprehend	A natural language processing service that performs sentiment analysis, personal information detection, key phrase extraction, etc. from text. Custom model creation is also possible. 一种自然语言处理服务，可从文本中执行情感分析、个人信息检测、关键短语提取等。也可以创建自定义模型。
Amazon Kendra   亚马逊肯德拉	An advanced search service for enterprises. Provides context-aware search results for natural language queries. Easy integration with RAG. 面向企业的高级搜索服务。为自然语言查询提供上下文感知搜索结果。与 RAG 轻松集成。
Amazon Lex   亚马逊 Lex	A service for building interactive interfaces (chatbots). Provides natural language understanding and dialogue management functions. Supports both voice and text. 用于构建交互式界面（聊天机器人）的服务。提供自然语言理解和对话管理功能。支持语音和文本。
Amazon Textract   亚马逊 Textract	A service for extracting text and structured data from documents. Capable of handwriting recognition, form processing, and table analysis. Provides high-accuracy OCR functionality. 用于从文档中提取文本和结构化数据的服务。能够进行手写识别、表单处理和表格分析。提供高精度的 OCR 功能。
Amazon Translate   亚马逊翻译	An automatic translation service between multiple languages. Provides real-time translation between 74 languages. Supports custom terminology dictionaries. 多种语言之间的自动翻译服务。提供 74 种语言之间的实时翻译。支持自定义术语词典。
Amazon Transcribe   亚马逊转录	A speech-to-text (speech recognition) service. Capable of multiple speaker identification and customization of specialized terminology. Supports real-time transcription. 语音转文本（语音识别）服务。能够识别多个说话人和自定义专业术语。支持实时转录。
Amazon Polly	A text-to-speech (speech synthesis) service. Provides natural pronunciation and Neural text-to-speech. Supports multiple languages and voice types. 文本到语音转换（语音合成）服务。提供自然发音和神经文本转语音。支持多种语言和语音类型。
Amazon CodeWhisperer   亚马逊 CodeWhisperer	An AI coding companion for programming assistance. Provides code completion and suggestions. 用于编程辅助的 AI 编码伴侣。提供代码完成和建议。

Image and Video Processing Services
图像和视频处理服务

Service Name   服务名称	Description   描述
Amazon Rekognition	An image and video analysis service. Provides face recognition, object detection, text extraction, content moderation, celebrity recognition, etc. Supports both real-time analysis and batch processing. 图像和视频分析服务。提供人脸识别、物体检测、文本提取、内容审核、明星识别等功能。支持实时分析和批处理。
Amazon Lookout for Vision	An anomaly detection service using industrial image analysis. Used for product defect detection in manufacturing lines, etc. 使用工业图像分析的异常检测服务。用于生产线等的产品缺陷检测。

Other AI-Related Services
其他 AI 相关服务

Service Name   服务名称	Description   描述
Amazon Personalize   Amazon 个性化	A service that provides personalized recommendations. Enables product recommendations and related content suggestions based on user behavior data. Supports real-time recommendations. 提供个性化推荐的服务。根据用户行为数据进行产品推荐和相关内容推荐。支持实时推荐。
Amazon Pinpoint   亚马逊 Pinpoint	A customer engagement service. Provides ML-powered segmentation, user behavior analysis, and optimal delivery time prediction functions. Enables multi-channel communication through email, SMS, push notifications, etc. 客户互动服务。提供 ML 驱动的细分、用户行为分析和最佳交付时间预测功能。支持通过电子邮件、短信、推送通知等进行多渠道通信。
Amazon Fraud Detector	A machine learning-based fraud detection service. Detects online fraudulent transactions, account takeovers, fake account creation, etc. Can be used in combination with custom rules and ML models. 一种基于机器学习的欺诈检测服务。检测在线欺诈交易、帐户接管、虚假帐户创建等。可以与自定义规则和 ML 模型结合使用。
Amazon Augmented AI (A2I) Amazon 增强 AI （A2I）	A service that manages human review task execution. Enables building workflows for human review of machine learning prediction results. 管理人工审核任务执行的服务。支持构建用于人工审核机器学习预测结果的工作流。
Amazon Mechanical Turk (MTurk) 亚马逊机械土耳其人 （MTurk）	A crowdsourcing marketplace. Enables execution of tasks such as data labeling and content moderation by humans. Can be integrated with Amazon SageMaker Ground Truth and Amazon Augmented AI (A2I). 一个众包市场。支持人工执行数据标记和内容审核等任务。可以与 Amazon SageMaker Ground Truth 和 Amazon Augmented AI （A2I） 集成。
Amazon QuickSight   亚马逊 QuickSight	A BI (Business Intelligence) tool. Equipped with ML predictive analysis capabilities, enabling data visualization and analysis. Supports data analysis in natural language through Q function. BI （Business Intelligence） 工具。配备 ML 预测分析功能，支持数据可视化和分析。通过 Q 函数支持自然语言数据分析。

Data Storage and Database Solutions
数据存储和数据库解决方案

Service   服务	Description   描述
Amazon S3   亚马逊 S3	Scalable object storage. Optimal for building data lakes. High durability and availability. 可扩展的对象存储。最适合构建数据湖。高耐用性和可用性。
Amazon EFS	Fully managed scalable file storage. Shareable across multiple instances. Supports NFS protocol. 完全托管的可扩展文件存储。可在多个实例之间共享。支持 NFS 协议。
Amazon FSx for Lustre 适用于 Lustre 的 Amazon FSx	Amazon FSx for Lustre is a high-performance file system capable of directly processing large-scale datasets. It seamlessly integrates with Amazon S3, automating data loading from and writing back to S3, and accelerates workloads with hundreds of GBps of parallel processing. Amazon FSx for Lustre 是一种高性能文件系统，能够直接处理大规模数据集。它与 Amazon S3 无缝集成，自动从 S3 加载数据并写回 S3，并通过数百 GBps 的并行处理加速工作负载。
Amazon DynamoDB	Fully managed NoSQL database. Enables fast read and write operations. Automatic scaling feature. 完全托管的 NoSQL 数据库。支持快速读写作。自动缩放功能。
Amazon Redshift   亚马逊 Redshift	Petabyte-scale data warehouse. Enables fast query processing. Columnar storage. PB 级数据仓库。支持快速查询处理。柱状存储。
Amazon OpenSearch Service Amazon OpenSearch 服务	An Elasticsearch-compatible search and analytics engine service. In addition to full-text search and real-time analytics, it provides vector database functionality supporting neural search and k-Nearest Neighbors (k-NN) vector search. Compatible with log analysis, application search, and security analytics, as well as AI applications such as recommendations and semantic search. Advanced search capabilities are enabled through integration with large language models using OpenSearch Neural Search functionality. 与 Elasticsearch 兼容的搜索和分析引擎服务。除了全文搜索和实时分析之外，它还提供支持神经搜索和 k 最近邻 （k-NN） 向量搜索的向量数据库功能。与日志分析、应用程序搜索和安全分析以及推荐和语义搜索等 AI 应用程序兼容。通过使用 OpenSearch 神经搜索功能与大型语言模型集成来启用高级搜索功能。
Amazon DocumentDB	MongoDB-compatible document database. Equipped with vector search functionality. Scalable document management. 与 MongoDB 兼容的文档数据库。配备矢量搜索功能。可扩展的文档管理。

Basic Concepts of Machine Learning
机器学习的基本概念

AI/ML Fundamental Concepts
AI/ML 基本概念

Term   术语	Description   描述
Artificial Intelligence (AI) 人工智能 （AI）	Computer systems that exhibit human-like intelligent behavior. Possess capabilities such as learning, reasoning, and problem-solving. Includes specialized AI for specific tasks and general AI for broad intelligence. 表现出类似人类智能行为的计算机系统。具备学习、推理和解决问题等能力。包括用于特定任务的专用 AI 和用于广泛智能的通用 AI。
Machine Learning (ML)   机器学习 （ML）	Algorithms or systems that learn patterns from data and perform tasks without explicit programming. Realized through a combination of statistical methods and algorithms. 从数据中学习模式并在没有显式编程的情况下执行任务的算法或系统。通过统计方法和算法的组合实现。
Deep Learning   深度学习	A machine learning technique using multi-layer neural networks. Demonstrates high performance in image recognition, natural language processing, etc. Requires large amounts of data and computational resources. 一种使用多层神经网络的机器学习技术。在图像识别、自然语言处理等方面表现出高性能。需要大量的数据和计算资源。
Feature   特征	Individual variables or attributes used as input to a model. Meaningful information extracted from data. 用作模型输入的单个变量或属性。从数据中提取的有意义信息。
Label   标签	Correct answer data in supervised learning. Target values or classification categories that the model should predict. 监督学习中的正确答案数据。模型应预测的目标值或分类类别。
Instance   实例	Individual data points. Composed of a combination of features and labels. 单个数据点。由特征和标签的组合组成。
Batch   批	A set of data processed simultaneously during model training. Affects memory efficiency and training speed. 在模型训练期间同时处理的一组数据。影响内存效率和训练速度。
Epoch   时代	A unit representing one complete processing of all training data. Model is gradually improved through multiple epochs of learning. 一个单元，表示对所有训练数据的一次完整处理。模型通过多个学习时期逐渐改进。
Iteration   迭 代	One update of model parameters. Often refers to processing per batch. 模型参数的一次更新。通常指每批的处理。
Parameters   参数	Values optimized by the model during the learning process. Includes weights and biases. 模型在学习过程中优化的值。包括权重和偏差。
Hyperparameters   超参数	Control parameters set before model learning. Includes learning rate and batch size. 在模型学习之前设置的控制参数。包括学习率和批量大小。
Inductive Bias   电感偏置	Assumptions or hypotheses inherent in the model. Characterizes the nature of the learning algorithm. 模型中固有的假设或假设。描述学习算法的性质。
Generalization Performance 泛化性能	The model's predictive ability on unseen data. Balancing overfitting and underfitting is important. 模型对看不见的数据的预测能力。平衡过拟合和欠拟合很重要。

Generative AI Related Concepts
生成式 AI 相关概念

Term   术语	Description   描述
Foundation Model (FM)   基础模型 （FM）	A general-purpose AI model pre-trained on large-scale data. Adaptable to various tasks. Forms the basis for transfer learning and fine-tuning. 在大规模数据上预先训练的通用 AI 模型。适应各种任务。构成迁移学习和微调的基础。
Large Language Model (LLM) 大型语言模型 （LLM）	A large-scale foundation model specialized in natural language processing. Examples include GPT and BERT. Demonstrates high performance in text generation and comprehension tasks. 专门用于自然语言处理的大规模基础模型。示例包括 GPT 和 BERT。在文本生成和理解任务中表现出高性能。
RAG (Retrieval-Augmented Generation) RAG（检索增强一代）	A method to improve the output quality of generative AI by searching and referencing external knowledge. Effective in preventing hallucination and improving accuracy. 一种通过搜索和引用外部知识来提高生成式 AI 输出质量的方法。有效防止幻觉和提高准确性。
Prompt   提示	Input text to generative AI models. Instructions or context to control the model's output. 将文本输入到生成式 AI 模型。用于控制模型输出的指令或上下文。
Token   令 牌	The smallest unit for dividing text in prompts. Composed of words or substrings. The basis for input/output limitations of models. 在提示中划分文本的最小单位。由单词或子字符串组成。模型输入/输出限制的基础。
Temperature   温度	A parameter in prompts that controls the randomness of generation. Higher values lead to more diverse outputs, lower values to more deterministic outputs. 提示中的一个参数，用于控制生成的随机性。值越高，输出越多样化，值越低，输出的确定性越高。
Top-p sampling   Top-p 采样	A method in prompts for selecting the next token based on cumulative probability. Controls the balance between output diversity and quality. 中的一种方法提示根据累积概率选择下一个标记。控制输出多样性和质量之间的平衡。
Top-k sampling   Top-k 采样	A method in prompts for selecting the next token from the top k tokens by probability. Used to control output. 提示按概率从前 k 个标记中选择下一个标记的方法。用于控制输出。
Context window   上下文窗口	The maximum length of input that the model can process at once in prompts. Affects the understanding of long contexts. 模型可以在提示中一次处理的最大输入长度。影响对长上下文的理解。
In-context learning   情境学习	The ability to learn tasks through examples within the prompt. Adaptation without additional training. 通过提示中的示例学习任务的能力。无需额外培训即可适应。
Fine-tuning   微调	The process of adapting a foundation model to specific tasks or domains. Specialization through additional learning. 使基础模型适应特定任务或领域的过程。通过额外学习实现专业化。
Prompt engineering   快速工程	The technique of designing effective prompts. Improves the quality and consistency of outputs. 设计有效提示的技术。提高输出的质量和一致性。
Hallucination   幻觉	The phenomenon where a model generates information not based on facts. A challenge for reliability. 模型生成信息不基于事实的现象。可靠性的挑战。
Style transfer   样式迁移	A generative technique to change the style of existing content. Used for images and text. 一种用于更改现有内容样式的生成技术。用于图像和文本。
Latent space   潜在空间	A compressed representation space of data learned by generative models. Controls the diversity of generation. 生成模型学习的数据的压缩表示空间。控制生成的多样性。
Attention mechanism   注意力机制	A mechanism to focus on important parts of the input. Core technology of Transformer models. 一种专注于输入重要部分的机制。Transformer 模型的核心技术。
Self-attention mechanism 自我注意机制	A mechanism to learn relationships between elements within a sequence. Effective for capturing long-range dependencies. 一种用于学习序列中元素之间关系的机制。对于捕获长期依赖关系有效。
Decoder   译码器	The part that generates the desired output from latent representations. An important component of generative models. 从 latent representations 生成所需输出的部分。生成模型的重要组成部分。
Encoder   编码器	The part that converts input into latent representations. Responsible for information compression and feature extraction. 将 input 转换为 latent representation 的部分。负责信息压缩和特征提取。
Transformer   变压器	An architecture based on self-attention mechanism. The foundation of modern generative AI. 一种基于自我注意机制的架构。现代生成式 AI 的基础。
Multimodal   模 态	The ability to handle multiple data formats such as text, images, and audio. 能够处理多种数据格式，例如文本、图像和音频。
Zero-shot capability   零点能力	The ability to perform new tasks using only pre-training in prompts. Adaptation without examples. 在提示中仅使用预训练来执行新任务的能力。没有例子的改编。
Few-shot capability   少发能力	The ability to perform new tasks with a few examples in prompts. Efficient adaptive learning. 能够执行新任务，并在提示中提供一些示例。高效的自适应学习。

Machine Learning Approaches
机器学习方法

Term   术语	Description   描述
Parametric learning   参数学习	An approach where the model shape is fixed and the number of parameters is constant. Examples include linear regression and logistic regression. 一种模型形状固定且参数数量恒定的方法。示例包括线性回归和 Logistic 回归。
Non-parametric learning   非参数学习	An approach where the complexity of the model changes according to the data. Examples include k-NN and kernel methods. 一种模型的复杂性根据数据而变化的方法。示例包括 k-NN 和内核方法。
Ensemble learning   集成学习	An approach that combines multiple learners to improve performance. Examples include Random Forest and Boosting. 一种将多个学习者组合在一起以提高绩效的方法。示例包括 Random Forest 和 Boosting。

Foundational Learning Theories
基础学习理论

Term   术语	Description   描述
Maximum Likelihood Estimation 最大似然估计	A method to estimate parameters by maximizing the probability of obtaining the data. 一种通过最大化获取数据的概率来估计参数的方法。
Bayesian Estimation   贝叶斯估计	A method to estimate parameters by calculating posterior probability from prior probability and data likelihood. 一种通过根据先验概率和数据似然计算后验概率来估计参数的方法。
Empirical Risk Minimization 实证风险最小化	A principle to minimize prediction errors on training data. 最小化训练数据预测误差的原则。
Structural Risk Minimization 结构性风险最小化	A principle to minimize prediction errors while considering model complexity. 在考虑模型复杂性的同时最小化预测误差的原则。

Loss Functions  损失函数

Term   术语	Description   描述
Squared Loss   平方损失	The square of the difference between predicted and actual values. Commonly used in regression problems. 预测值和实际值之差的平方。常用于回归问题。
Cross-Entropy Loss   交叉熵损失	A loss function used in classification problems. Measures the distance between probability distributions. 分类问题中使用的损失函数。测量概率分布之间的距离。
Hinge Loss   铰链丢失	A loss function used in SVMs. Achieves margin maximization. SVM 中使用的损失函数。实现边距最大化。
Huber Loss   Huber 损失	A loss function robust to outliers. A combination of squared loss and absolute loss. 对异常值具有鲁棒性的损失函数。平方损失和绝对损失的组合。

Optimization Theory  优化理论

Term   术语	Description   描述
Convex Optimization   凸优化	A special optimization problem where local optima are global optima. 一个特殊的优化问题，其中局部最优值是全局最优值。
Stochastic Optimization   随机优化	A method that uses randomness to search for optimal solutions. 一种使用随机性搜索最佳解决方案的方法。
Constrained Optimization 约束优化	An optimization problem under constraints. Solved using methods like Lagrange multipliers. 约束下的优化问题。使用拉格朗日乘子等方法求解。

Terms Related to Learning Process
与学习过程相关的术语

Term   术语	Description   描述
Vanishing Gradient Problem 梯度消失问题	A phenomenon in deep neural networks where gradients vanish during backpropagation. Makes learning difficult in deep layers. 深度神经网络中的一种现象，其中梯度在反向传播期间消失。使深层学习变得困难。
Exploding Gradient   分解渐变	A phenomenon in deep neural networks where gradients grow exponentially. Causes instability in learning. 深度神经网络中的一种现象，其中梯度呈指数增长。导致学习不稳定。
Sparsity   稀疏	A property where many of the data or model parameters are zero. Affects computational efficiency and generalization performance. 许多数据或模型参数为零的属性。影响计算效率和泛化性能。
Curse of Dimensionality   维度诅咒	A problem where the required amount of data increases exponentially as the number of feature dimensions increases. A challenge in high-dimensional data analysis. 所需数据量随着特征维度数的增加而呈指数级增长的问题。高维数据分析中的挑战。

Data Quality Related Terms
数据质量相关术语

Term   术语	Description   描述
Data Imbalance   数据不平衡	A state where there is a large difference in the number of samples between classes. Makes learning difficult for minority classes. 类之间的样本数存在较大差异的状态。使少数族裔阶层的学习变得困难。
Noise   噪声	Unwanted variations or errors in data. Can hinder model learning. 数据中不需要的变体或错误。可能会阻碍模型学习。
Outliers   异常	Values that deviate significantly from the general distribution of data. Can negatively affect model learning. 与数据的一般分布明显不同的值。可能会对模型学习产生负面影响。
Missing Values   缺失值	Unrecorded or unmeasured values in a dataset. Requires appropriate handling. 数据集中未记录或未测量的值。需要适当的处理。

Model Evaluation Related Terms
模型评估相关术语

Term   术语	Description   描述
Baseline   基线	A simple model or performance metric used as a comparison standard. Used to evaluate improvement. 用作比较标准的简单模型或性能指标。用于评估改善情况。
Significance Testing   显著性检验	A method to evaluate whether performance differences between models are statistically meaningful. 一种评估模型之间的性能差异是否具有统计意义的方法。
Cross-Entropy   交叉熵	A metric that measures the difference between predicted probabilities and true distribution in classification problems. 度量分类问题中预测概率与真实分布之间的差异的指标。
Confusion Matrix   混淆矩阵	A table that aggregates prediction results by classification. Used for performance evaluation. 按分类聚合预测结果的表。用于性能评估。

Activation Functions  激活函数

Term   术语	Description   描述
ReLU   ReLU 系列	The most commonly used activation function. A simple non-linear function that sets negative inputs to zero. 最常用的激活函数。一个简单的非线性函数，用于将负输入设置为零。
Sigmoid   乙状结肠	Converts output to a range of 0-1. Often used in the output layer for binary classification. 将输出转换为 0-1 的范围。通常用于二进制分类的输出层。
tanh   丹	Converts output to a range of -1 to 1. Mitigates the vanishing gradient problem better than sigmoid. 将输出转换为 -1 到 1 的范围。比 sigmoid 更好地缓解梯度消失问题。
Softmax	Outputs probability distribution for multiple classes. Used in the output layer for multi-class classification. 输出多个类的概率分布。在输出层中用于多类分类。

Types of Learning Algorithms
学习算法的类型

Term   术语	Description   描述
Perceptron   感知器	The most basic neural network. Suitable for linearly separable problems. 最基本的神经网络。适用于线性可分问题。
SVM (Support Vector Machine) SVM （支持向量机）	A method that determines the classification boundary by maximizing margin. Non-linear classification is possible with kernel trick. 一种通过最大化边距来确定分类边界的方法。非线性分类可以通过 kernel trick 实现。
Decision Tree   决策树	A method that makes predictions by hierarchically dividing data. High interpretability and easy evaluation of feature importance. 一种通过分层划分数据进行预测的方法。高可解释性，易于评估特征重要性。
k-Nearest Neighbors   k 最近邻	A method that makes predictions based on the majority of the k nearest training data. Simple but computationally expensive. 一种根据 k 个最接近的训练数据中的大多数进行预测的方法。简单但计算成本高。

Data Quality Indicators  数据质量指标

Term   术语	Description   描述
Data Completeness   数据完整性	An indicator of the degree of missing values, duplicates, and inconsistencies in a dataset. 数据集中缺失值、重复项和不一致程度的指示器。
Data Consistency   数据一致性	An indicator of whether data formats and value ranges are as expected. 指示数据格式和值范围是否符合预期的指标。
Data Freshness   数据新鲜度	An indicator of the update time and expiration date of data. 数据的更新时间和到期日期的指示符。
Data Representativeness   数据代表性	An indicator of whether the sample appropriately represents the population. 指示样本是否适当地代表总体的指标。

Model Quality Indicators
模型质量指标

Term   术语	Description   描述
Prediction Stability   预测稳定性	An indicator of prediction consistency for similar inputs. 类似输入的预测一致性指标。
Model Confidence   模型置信度	An indicator of the model's confidence in each prediction. 模型对每个预测的置信度的指示器。
Explainability   可解释性	An indicator of the ease of interpreting the reasons for model predictions. 表示易于解释模型预测原因的指标。
Robustness   鲁棒性	An indicator of the model's resistance to noise and outliers. 模型对噪声和异常值的抵抗力的指示器。

Statistical Concepts  统计概念

Term   术语	Description   描述
Analysis of Variance   方差分析	A statistical method for analyzing sources of variation in data. 一种用于分析数据变异来源的统计方法。
Hypothesis Testing   假设检验	A method for verifying statistical hypotheses. 一种验证统计假设的方法。
Confidence Interval   置信区间	A range that quantifies the uncertainty of an estimate. 量化估计的不确定性的范围。
Effect Size	An indicator of the practical magnitude of statistical differences. 统计差异的实际量级的指标。

Model Development Process
模型开发过程

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Studio provides an integrated development environment (IDE) that enables one-stop execution from notebook creation to model development, training, and deployment, realizing centralized management of ML workflows.
* Amazon SageMaker Studio 提供集成开发环境 （IDE），支持从笔记本创建到模型开发、训练和部署的一站式执行，实现 ML 工作流的集中管理。
* Amazon SageMaker Canvas provides a no-code ML development environment that allows data preparation to model deployment through drag & drop without writing code, enabling development for business analysts.
* Amazon SageMaker Canvas 提供了一个无代码的 ML 开发环境，允许数据准备通过拖放进行模型部署，而无需编写代码，从而为业务分析师提供开发支持。

Model Development Process
模型开发过程

Phase   阶段	Description   描述
Data Collection   数据采集	Collection and integration of data necessary for learning. Includes identification of data sources and quality checks. Consider data representativeness and balance. 收集和整合学习所需的数据。包括数据源的标识和质量检查。考虑数据代表性和平衡性。
Data Preprocessing   数据预处理	Implement data splitting, cleaning, data labeling, feature engineering, scaling (normalization, standardization). Improve data quality and convert to a format suitable for learning. Include handling of missing values and outliers. 实现数据拆分、清理、数据标记、特征工程、扩展（规范化、标准化）。提高数据质量并转换为适合学习的格式。包括对缺失值和异常值的处理。
Model Selection   型号选择	Select appropriate algorithms and architectures based on problem type (classification/regression, etc.), data characteristics, requirements (accuracy/speed/explainability). Consider computational resource constraints and deployment environment. Also consider the possibility of using pre-trained models. 根据问题类型（分类/回归等）、数据特征、要求（准确性/速度/可解释性）选择合适的算法和架构。考虑计算资源约束和部署环境。还要考虑使用预训练模型的可能性。
Model Training   模型训练	Train the model using the selected algorithm. Include hyperparameter optimization. Conduct performance evaluation through cross-validation. 使用所选算法训练模型。包括超参数优化。通过交叉验证进行性能评估。
Model Evaluation   模型评估	Verify model performance. Analyze from multiple angles using various evaluation metrics. Confirm generalization performance on test data. 验证模型性能。使用各种评估指标从多个角度进行分析。确认测试数据的泛化性能。
Deployment   部署	Deploy the model to the production environment. Include scaling and monitoring settings. Also conduct A/B testing to verify effectiveness. 将模型部署到生产环境。包括扩展和监控设置。此外，还要进行 A/B 测试以验证有效性。
Inference   推理	Execute predictions on new data using the deployed model. Perform predictions in real-time or batch processing. 使用部署的模型对新数据执行预测。实时或批处理执行预测。
Monitoring   监测	Continuously monitor model performance. Detect drift and make decisions on retraining. Track quality metrics and set alerts. 持续监控模型性能。检测偏差并做出重新训练的决策。跟踪质量指标并设置警报。

Data Collection  数据采集

Types of Data  数据类型

Type   类型	Description   描述
Structured Data   结构化数据	Data organized in tabular form. Such as data managed in RDBMS. Has a clear schema. 以表格形式组织的数据。例如在 RDBMS 中管理的数据。具有清晰的架构。
Unstructured Data   非结构化数据	Data without a fixed structure. Such as text, images, audio, video. Requires special techniques for processing. 没有固定结构的数据。例如文本、图像、音频、视频。需要特殊的加工技术。
Semi-structured Data   半结构化数据	Partially structured data. Such as JSON, XML, HTML. Has a flexible schema. 部分结构化数据。如 JSON、XML、HTML。具有灵活的架构。
Vector Data   矢量数据	Data represented as numerical vectors. Such as word embeddings, feature vectors. Suitable for similarity calculations. 以数值向量表示的数据。例如单词嵌入、特征向量。适用于相似度计算。

ETL (Extract, Transform, Load)
ETL （提取、转换、加载）

Phase   阶段	Description   描述
Extract   提取	Extract data from various sources. Check data format and quality. Perform consistency checks. 从各种来源提取数据。检查数据格式和质量。执行一致性检查。
Transform   变换	Transform and process data. Execute cleansing, normalization, aggregation, etc. Transform according to business rules. 转换和处理数据。执行清理、规范化、聚合等。根据业务规则进行转换。
Load   负荷	Save and load processed data. Store in data warehouses or data lakes. Ensure consistency. 保存和加载已处理的数据。存储在数据仓库或数据湖中。确保一致性。

Data Preprocessing  数据预处理

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Data Wrangler provides tools to streamline data preparation and preprocessing, enabling data cleansing to feature engineering via GUI.
* Amazon SageMaker Data Wrangler 提供简化数据准备和预处理的工具，使数据清理能够通过 GUI 进行特征工程。
* Amazon SageMaker Canvas provides functionality to perform data preprocessing, feature engineering, data transformation, etc. via GUI without writing code, enabling data preparation by business analysts.
* Amazon SageMaker Canvas 提供通过 GUI 执行数据预处理、特征工程、数据转换等的功能，而无需编写代码，从而使业务分析师能够准备数据。

Data Splitting  数据拆分

Term   术语	Description   描述
Training Data   训练数据	Dataset used for model learning. Typically accounts for about 60-80% of all data. 用于模型学习的数据集。通常占所有数据的 60-80% 左右。
Validation Data   验证数据	Dataset used for model hyperparameter tuning and performance evaluation. Typically accounts for about 10-20% of all data. 用于模型超参数优化和性能评估的数据集。通常占所有数据的 10-20% 左右。
Test Data   测试数据	Independent dataset used for final model evaluation. Typically accounts for about 10-20% of all data. 用于最终模型评估的独立数据集。通常占所有数据的 10-20% 左右。
Holdout Method   保持方法	Basic method of splitting data into training and evaluation sets. Used when data quantity is sufficient. 将数据拆分为训练集和评估集的基本方法。当数据量足够时使用。
Stratified Sampling   分层抽样	Method of splitting data while maintaining class ratios. Important for imbalanced datasets. 在保持类比率的同时拆分数据的方法。对于不平衡的数据集很重要。

Cleansing (Cleaning)  清洁 （Cleaning）

Task   任务	Description   描述
Noise Removal   噪	Detection and removal of outliers and noise. Improves data quality. Utilizes statistical methods and domain knowledge. 检测并删除异常值和噪声。提高数据质量。利用统计方法和领域知识。
Missing Value Handling   缺失值处理	Completion or removal of missing data. Impute with mean, median, predicted values, etc. Consider MAR and MCAR assumptions. 完成或删除缺失数据。使用平均值、中位数、预测值等进行插补。考虑 MAR 和 MCAR 假设。
Outlier Detection   异常值检测	Identification of outliers using statistical methods or ML techniques. Important to check consistency with domain knowledge. 使用统计方法或 ML 技术识别异常值。检查与领域知识的一致性很重要。
Duplicate Data Elimination 重复数据消除	Detection and removal of duplicate records. Ensures data consistency. Normalization of key items. 检测和删除重复记录。确保数据一致性。关键项目的规范化。

Data Labeling  数据标注

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Ground Truth provides a data labeling service for creating high-quality training datasets.
* Amazon SageMaker Ground Truth 提供用于创建高质量训练数据集的数据标记服务。

Method   方法	Description   描述
Manual Labeling   手动贴标	Direct labeling by humans. High quality but time and cost intensive. Effective when specialized knowledge is required. 由人类直接标记。高质量，但时间和成本密集。在需要专业知识时有效。
Semi-Automatic Labeling   半自动贴标	Combination of AI prediction and human verification. Enables efficient labeling. Achieves balance between quality and efficiency. AI 预测和人工验证相结合。实现高效贴标。实现质量和效率之间的平衡。
Active Learning   主动学习	Selection of target data for efficient labeling. Prioritizes data with high uncertainty. Optimizes labeling costs. 选择目标数据以实现高效标记。优先考虑具有高不确定性的数据。优化标签成本。
Label Quality Management 标签质量管理	Checks for consistency and errors. Includes consensus building among multiple annotators. Setting and monitoring of quality metrics. 检查一致性和错误。包括在多个注释者之间建立共识。设置和监控质量指标。

Feature Engineering  特征工程

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
・Amazon SageMaker Feature Store provides a centralized repository for managing and sharing features, ensuring consistency of features both online and offline.
・Amazon SageMaker Feature Store 提供了一个用于管理和共享功能的集中式存储库，确保在线和离线功能的一致性。

Technique   技术	Description   描述
Feature Selection   特征选择	Selection of useful input variables for the model. Selection based on correlation analysis and importance evaluation. Contributes to dimensionality reduction and model performance improvement. Used when there are many features or when you want to remove unnecessary features to prevent model overfitting. Particularly effective when analyzing datasets with many variables, such as medical or financial data. 为模型选择有用的输入变量。基于相关性分析和重要性评估的选择。有助于降维和提高模型性能。当有许多特征或您想要删除不必要的特征以防止模型过拟合时使用。在分析具有许多变量的数据集（例如医疗或财务数据）时特别有效。
Feature Extraction   特征提取	The process of extracting meaningful features from raw data. Examples include Fourier transform in signal processing and edge detection from images. Used when there is a need to extract useful information from complex raw data, especially important in image processing, speech processing, and sensor data analysis. In time series data analysis, it is utilized for extracting statistical measures and frequency characteristics. 从原始数据中提取有意义特征的过程。示例包括信号处理中的傅里叶变换和图像中的边缘检测。当需要从复杂的原始数据中提取有用信息时使用，这在图像处理、语音处理和传感器数据分析中尤其重要。在时间序列数据分析中，它用于提取统计度量和频率特征。
Feature Scaling   特征缩放	The process of adjusting the value range of features. Includes standardization and normalization. Essential when dealing with datasets that have features on different scales, particularly important for machine learning algorithms using gradient descent methods and clustering algorithms that perform distance calculations. 调整特征值范围的过程。包括标准化和规范化。在处理具有不同比例特征的数据集时，这一点至关重要，对于使用梯度下降方法的机器学习算法和执行距离计算的聚类算法尤其重要。
Feature Interaction   特征交互	Creation of new features by combining multiple features. Used when wanting to capture non-linear relationships or model phenomena that cannot be explained by individual features alone. Particularly utilized in regression analysis and predictive models to improve prediction accuracy. 通过组合多个特征来创建新特征。当想要捕获非线性关系或无法单独用单个特征解释的模型现象时使用。特别用于回归分析和预测模型，以提高预测准确性。
Dimensionality Reduction 降维	Techniques for reducing the feature dimensions of data. PCA and t-SNE are representative methods. Contributes to improved computational efficiency and performance. Used when visualization of high-dimensional data is necessary or when wanting to reduce computational costs. Particularly useful when dealing with high-dimensional data in image recognition and document classification. 减小数据特征维度的技术。PCA 和 t-SNE 是代表性方法。有助于提高计算效率和性能。当需要高维数据可视化或希望降低计算成本时使用。在图像识别和文档分类中处理高维数据时特别有用。
Encoding   编码	Numerical conversion of categorical values. Uses methods such as One-Hot, Label, and Target encoding. Select appropriate methods based on data characteristics. Essential when building machine learning models that handle categorical data, especially important when there are many categories or when relationships between categories need to be considered. 分类值的数值转换。使用 One-Hot、Label 和 Target 编码等方法。根据数据特征选择合适的方法。在构建处理分类数据的机器学习模型时，这一点至关重要，当存在许多类别或需要考虑类别之间的关系时尤其重要。
Embedding   嵌入	Conversion of high-dimensional data into low-dimensional vector representations. Examples include Word2Vec and BERT. Preserves semantic similarity. Used when dealing with text data or large-scale categorical data, playing a particularly important role in natural language processing and recommender system construction. 将高维数据转换为低维向量表示。示例包括 Word2Vec 和 BERT。保留语义相似性。用于处理文本数据或大规模分类数据，在自然语言处理和推荐系统构建中起着特别重要的作用。
Data Augmentation   数据增强	Enhancement of training data by transforming existing data. Includes rotation, scaling, etc. Improves model generalization performance. Used when training data is limited or when wanting to prevent model overfitting. Particularly effective in tasks using deep learning, such as image recognition and speech recognition. 通过转换现有数据来增强训练数据。包括旋转、缩放等。改进了模型泛化性能。当训练数据受到限制或想要防止模型过拟合时使用。在使用深度学习的任务中特别有效，例如图像识别和语音识别。

Encoding Techniques  编码技术

Encoding Technique   编码技术	Description and Use Cases 描述和使用案例
Label Encoding   标签编码	A technique that converts categorical values into continuous integer values. Suitable when there is an ordinal relationship between categories (e.g., education level, age group). Memory-efficient and often used in decision tree-based algorithms. However, caution is needed for non-ordinal categorical data as it introduces a numerical relationship between categories. 一种将分类值转换为连续整数值的技术。当类别之间存在序数关系（例如，教育水平、年龄组）时适用。内存效率高，通常用于基于决策树的算法。但是，对于非有序分类数据，需要谨慎，因为它会在类别之间引入数值关系。
One-Hot Encoding   One-Hot 编码	A technique that converts categorical values into binary vectors. Optimal for cases where there is no ordinal relationship between categories (e.g., color, gender, occupation). Treats each category equally, but can lead to high memory consumption and increased computational cost due to dimensionality increase when there are many categories. Particularly important in linear models and neural networks. 一种将分类值转换为二进制向量的技术。最适合类别之间没有序数关系的情况（例如，颜色、性别、职业）。平等地对待每个类别，但当有许多类别时，由于维度增加，可能会导致高内存消耗和计算成本增加。在线性模型和神经网络中尤为重要。
Target Encoding   目标编码	A technique that replaces categorical values with the mean of the target variable. Effective when there are a very large number of categories or when there is a strong association between categories and the target variable. However, there is a risk of overfitting, so appropriate regularization and cross-validation are necessary. Particularly useful for improving performance in predictive models. 一种将分类值替换为目标变量的平均值的技术。当类别数量非常多或类别与目标变量之间存在很强关联时有效。但是，存在过拟合的风险，因此需要适当的正则化和交叉验证。对于提高预测模型的性能特别有用。
Frequency Encoding   频率编码	A technique that converts categories to numerical values based on their frequency of occurrence. Suitable when the frequency of categories holds significant meaning (e.g., product popularity, usage frequency). Easy to implement and interpret, but has the limitation of not being able to distinguish between categories with the same frequency. 一种根据类别出现频率将类别转换为数值的技术。当类别的频率具有重要意义（例如，产品受欢迎程度、使用频率）时，它适用。易于实现和解释，但存在无法区分相同频率的类别的局限性。
Binary Encoding   二进制编码	A technique that converts categories into binary representations. More memory-efficient than One-Hot Encoding as it can represent with fewer dimensions, useful when there are many categories. However, the generated features can be difficult to interpret, and relationships between categories may be lost. 一种将类别转换为二进制表示的技术。比 One-Hot Encoding 内存效率更高，因为它可以用更少的维度表示，这在有许多类别时很有用。但是，生成的要素可能难以解释，并且类别之间的关系可能会丢失。
Hash Encoding   哈希编码	A technique that uses a hash function to convert categories into fixed-dimensional features. Suitable for cases with an extremely large number of categories or when new categories are continuously added. Memory-efficient and can handle online learning, but there is a possibility of information loss due to hash collisions. 一种使用哈希函数将类别转换为固定维度特征的技术。适用于类别数量极多或不断添加新类别的情况。内存效率高，可以处理在线学习，但可能会因哈希冲突而丢失信息。

Feature Extraction Techniques for Text Data
文本数据的特征提取技术

Technique   技术	Description   描述
TF-IDF	A technique that calculates word importance based on term frequency and inverse document frequency. A fundamental feature in text analysis. Widely used in document classification, information retrieval, keyword extraction, and other tasks where word importance needs to be considered. 一种根据词频和逆向文档频率计算单词重要性的技术。文本分析中的一个基本功能。广泛用于文档分类、信息检索、关键字提取和其他需要考虑单词重要性的任务。
Word2Vec	A technique that converts words into fixed-length dense vectors. Captures semantic similarity between words. Used in natural language processing tasks that require consideration of word meaning relationships and context, such as sentiment analysis, document classification, question-answering systems, and machine translation. 一种将单词转换为固定长度的密集向量的技术。捕获单词之间的语义相似性。用于需要考虑词义关系和上下文的自然语言处理任务，例如情感分析、文档分类、问答系统和机器翻译。
Doc2Vec   Doc2Vec 的	An extension of Word2Vec that learns vector representations for entire documents. Used for document similarity calculations. Suitable for tasks requiring semantic comparison at the document level, such as document classification, clustering, recommender systems, and similar document search. Word2Vec 的扩展，用于学习整个文档的矢量表示。用于文档相似性计算。适用于需要在文档级别进行语义比较的任务，例如文档分类、聚类、推荐系统和类似文档搜索。
FastText   FastText 快文本	A word embedding technique that considers substrings. Capable of handling unknown words. Particularly effective in processing languages with rich morphology, handling text with spelling errors, and analyzing social media posts where new or modified words frequently appear. 一种考虑子字符串的单词嵌入技术。能够处理未知单词。在处理具有丰富形态的语言、处理具有拼写错误的文本以及分析经常出现新单词或修改单词的社交媒体帖子时特别有效。
BERT Tokenization   BERT 分词	A tokenization technique for BERT models. Uses the WordPiece algorithm. Used as essential preprocessing when using BERT models for advanced natural language processing tasks that consider context, such as sentiment analysis, named entity recognition, and question answering. 一种用于 BERT 模型的标记化技术。使用 WordPiece 算法。在使用 BERT 模型执行考虑上下文的高级自然语言处理任务（例如情感分析、命名实体识别和问答）时，用作必要的预处理。
BPE (Byte Pair Encoding) BPE （字节对编码）	A technique that learns subword units by merging frequent character strings. Allows control of vocabulary size. Particularly effective in machine translation, multilingual processing, and processing languages with complex morphology where efficient handling of large vocabularies is necessary. 一种通过合并频繁的字符串来学习子词单位的技术。允许控制词汇表大小。在机器翻译、多语言处理和需要有效处理大量词汇的复杂形态语言中特别有效。
Bag of Words (BoW) 词袋 （BoW）	The most basic technique that vectorizes word frequency in documents. Does not consider word order. Used in basic text analysis tasks where word frequency alone is sufficient for performance, such as spam email detection, document classification, and topic classification. 对文档中的词频进行矢量化的最基本技术。不考虑词序。用于基本文本分析任务，其中仅字频就足以提高性能，例如垃圾邮件检测、文档分类和主题分类。
n-gram   n-gram 元语法	A technique that uses combinations of n consecutive words or characters as features. Captures local context. Used in language modeling, spell checking, author identification, predictive input for programming languages, and other cases where local context or word order is important. 一种使用 n 个连续单词或字符的组合作为特征的技术。捕获本地上下文。用于语言建模、拼写检查、作者识别、编程语言的预测输入以及本地上下文或词序很重要的其他情况。

Text Data Preprocessing  文本数据预处理

Method   方法	Description   描述
Tokenization   分词化	Splitting text into words or substrings. Selection of appropriate splitting method based on language characteristics. 将文本拆分为单词或子字符串。根据语言特性选择合适的拆分方法。
Normalization   正常化	Implement unification of uppercase and lowercase, accent removal, character type standardization, etc. 实现大小写统一、去除重音、字符类型标准化等。
Stop Word Removal   停用词删除	Removal of common words with little information (articles, prepositions, etc.). 删除信息较少的常用词（冠词、介词等）。
Lemmatization/Stemming   词形还原/词干提取	Convert words to their base form. Use morphological analysis or stemming. 将单词转换为其基本形式。使用形态分析或词干提取。
Noise Removal   噪	Removal of special characters, HTML/XML tags, unnecessary spaces, etc. 删除特殊字符、HTML/XML 标签、不必要的空格等。

Image Data Preprocessing
图像数据预处理

Method   方法	Description   描述
Resize   调整	Standardization of image size. Convert to a size suitable for model input. 图像大小的标准化。转换为适合模型输入的大小。
Normalization   正常化	Standardization of pixel values. Generally converted to a range of 0-1 or -1-1. 像素值的标准化。通常转换为 0-1 或 -1-1 的范围。
Color Space Conversion   色彩空间转换	Conversion to color spaces such as RGB, HSV, grayscale according to purpose. 根据用途转换为 RGB、HSV、灰度等色彩空间。
Noise Removal   噪	Noise reduction using median filters or Gaussian filters. 使用中值滤波器或高斯滤波器进行降噪。
Data Augmentation   数据增强	Enhancement of training data through rotation, flipping, scaling, etc. 通过旋转、翻转、缩放等来增强训练数据。

Scaling Methods  缩放方法

Method   方法	Description   描述
Normalization   正常化	A transformation that fits data into a specific range (usually 0-1). Unifies the scale between features, making them comparable. 将数据拟合到特定范围 （通常为 0-1） 的转换。统一要素之间的比例，使其具有可比性。
Standardization   标准化	A transformation that converts data to a distribution with mean 0 and standard deviation 1. Susceptible to outliers but suitable for normally distributed data. 将数据转换为均值为 0 且标准差为 1 的分布的转换。易受异常值的影响，但适用于正态分布的数据。
Standard Scaler   标准定标器	A scaler that implements standardization. Converts to mean 0 and standard deviation 1. Suitable for data following normal distribution. 实现标准化的缩放器。转换为平均值 0 和标准差 1。适用于遵循正态分布的数据。
Robust Scaler   强大的缩放器	Robust scaling using median and interquartile range. Less affected by outliers. 使用中位数和四分位数范围的稳健缩放。受异常值的影响较小。
Min Max Scaler   最小最大缩放器	A scaler that implements normalization. Converts data to a specified range such as 0-1. Suitable for neural network inputs. 实现规范化的缩放器。将数据转换为指定的范围，例如 0-1。适用于神经网络输入。
Max Absolute Scaler	Normalizes by maximum absolute value. Suitable for sparse data. Maintains zero-centered scale. 按最大绝对值进行规格化。适用于稀疏数据。保持以零为中心的刻度。

Model Selection  型号选择

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker JumpStart functions as an ML hub providing pre-trained models and solutions, deployable with one click.
* Amazon SageMaker JumpStart 充当 ML 中心，提供预先训练的模型和解决方案，只需单击一下即可部署。

Model Types  模型类型

Classification   分类	Description   描述
Supervised Learning Models 监督式学习模型	Predictive models for classification or regression. Uses labeled data. Examples include RandomForest, SVM, Neural Networks. 用于分类或回归的预测模型。使用标记的数据。示例包括 RandomForest、SVM、神经网络。
Unsupervised Learning Models 无监督学习模型	Models for clustering and pattern discovery. Uses unlabeled data. Examples include K-means, PCA, Auto-encoders. 用于聚类和模式发现的模型。使用未标记的数据。示例包括 K-means、PCA、自动编码器。
Semi-Supervised Learning Models 半监督学习模型	Models that learn using a small amount of labeled data and a large amount of unlabeled data. Examples include pseudo-labeling, co-training. 使用少量标记数据和大量未标记数据进行学习的模型。示例包括伪标记、共同训练。
Generative Models   生成模型	Models that generate data or learn distributions. Examples include GANs, VAE, Diffusion Models. 生成数据或学习分布的模型。示例包括 GAN、VAE、扩散模型。
Transfer Learning Models 迁移学习模型	Models that utilize pre-trained knowledge. Based on foundation models such as BERT, GPT, ResNet. 利用预先训练知识的模型。基于 BERT、GPT、ResNet 等基础模型。

Selection Criteria  纳入排除标准

Criteria   标准	Description   描述
Data Characteristics   数据特征	Model selection based on data quantity, dimensionality, presence of noise, class balance, sparsity, etc. 根据数据量、维度、噪声的存在、类平衡、稀疏性等进行模型选择。
Computational Resources   计算资源	Selection based on available memory, CPU/GPU, and training time constraints. 根据可用内存、CPU/GPU 和训练时间限制进行选择。
Prediction Performance   预测性能	Selection based on target performance metrics such as accuracy, recall, F1 score. 根据目标性能指标（如准确性、召回率、F1 分数）进行选择。
Inference Speed   推理速度	Selection based on real-time requirements, batch processing requirements. 根据实时需求、批处理要求进行选择。
Explainability   可解释性	Selection based on requirements for model transparency and interpretability. 根据模型透明度和可解释性的要求进行选择。
Scalability   可扩展性	Selection based on ability to handle increases in data volume and system expansion. 根据处理数据量增加和系统扩展的能力进行选择。
Cost   成本	Selection based on total cost of ownership including development, training, and operation. 根据总拥有成本（包括开发、培训和运营）进行选择。

Common Algorithms  常用算法

Algorithm   算法	Application Scenarios   应用场景
Linear Regression   线性回归	Simple regression problems, when interpretation of relationships is important. 当关系的解释很重要时，会出现简单的回归问题。
Logistic Regression   Logistic 回归	Binary classification problems, when probability prediction is needed. 二元分类问题，当需要概率预测时。
Random Forest	Classification and regression of structured data, analysis of feature importance. 结构化数据的分类和回归，特征重要性分析。
XGBoost/LightGBM	High-performance prediction problems with structured data. 结构化数据的高性能预测问题。
Neural Networks   神经网络	Complex pattern recognition, image and speech processing. 复杂的模式识别、图像和语音处理。
BERT/Transformer   BERT/变压器	Text processing, natural language understanding tasks. 文本处理、自然语言理解任务。
CNN	Image recognition, pattern detection. 图像识别、模式检测。
RNN/LSTM	Time series data analysis, sequence data processing. 时间序列数据分析、序列数据处理。
Reinforcement Learning Models 强化学习模型	Decision making, game strategies, robot control, etc. 决策、游戏策略、机器人控制等。

Architecture Considerations
架构注意事项

Element   元素	Description   描述
Model Size   模型大小	Consideration of number of parameters, memory requirements, storage requirements. 考虑参数数量、内存要求、存储要求。
Layer Configuration   图层配置	Selection of number of layers, number of units, activation functions for neural networks. 选择神经网络的层数、单元数、激活函数。
Ensemble Methods   集成方法	Methods of combining multiple models, voting or averaging strategies. 组合多个模型、投票或平均策略的方法。
Quantization and Compression 量化和压缩	Model lightweighting, adaptation to edge deployment. 模型轻量化，适应边缘部署。
Batch Size   批量大小	Balance between memory usage and speed during training and inference. 在训练和推理期间平衡内存使用和速度。
Distributed Learning Support 分布式学习支持	Possibility of learning on multiple GPUs or multi-nodes. 可以在多个 GPU 或多节点上学习。

Model Selection Strategies
模型选择策略

Strategy   策略	Description   描述
Baseline Construction   基线构造	Strategy to start with simple models and gradually increase complexity. 从简单模型开始，然后逐渐增加复杂性的策略。
AutoML   自动机器学习	Utilization of tools for automatic model selection and optimization. 利用工具进行自动模型选择和优化。
Algorithm Comparison   算法比较	Evaluate multiple models in parallel and select the optimal one. 并行评估多个模型并选择最佳模型。
Experiment Management   实验管理	Tracking and documentation of the model selection process. 跟踪和记录模型选择过程。
A/B Testing   A/B 测试	Comparison of different models' performance in real environments. 不同模型在真实环境中的性能比较。
Gradual Optimization   逐步优化	Strategy to optimize while gradually increasing model complexity. 在逐渐增加模型复杂性的同时进行优化的策略。

Model Training  模型训练

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Debugger performs debugging and monitoring of the training process, enabling visualization of metrics and setting of alerts.
* Amazon SageMaker Debugger 执行训练过程的调试和监控，从而实现指标可视化和警报设置。

Basic Concepts of Model Learning
模型学习的基本概念

Concept   概念	Description   描述
Inductive Bias   电感偏置	Assumptions or hypotheses inherent in the model. Determines the nature of the learning algorithm. Appropriate bias improves generalization performance. 模型中固有的假设或假设。确定学习算法的性质。适当的偏差可以提高泛化性能。
Bias-Variance Tradeoff   偏差-方差权衡	The tradeoff relationship between model complexity and generalization performance. Balance between overfitting and underfitting. 模型复杂性和泛化性能之间的权衡关系。过拟合和欠拟合之间的平衡。
Cross-Entropy Loss   交叉熵损失	A loss function commonly used in classification problems. Measures the difference between predicted probabilities and true distribution. 分类问题中常用的损失函数。测量预测概率与真实分布之间的差异。

Training Methods  训练方法

Method   方法	Description   描述
Supervised Learning   监督式学习	A method of learning using data with correct labels. Used for classification and regression tasks. The quality and quantity of data determine performance. 一种使用具有正确标签的数据进行学习的方法。用于分类和回归任务。数据的质量和数量决定了性能。
Unsupervised Learning   无监督学习	A method to discover patterns from unlabeled data. Used for clustering and anomaly detection. Reveals latent structures in data. 一种从未标记的数据中发现模式的方法。用于聚类和异常检测。揭示数据中的潜在结构。
Semi-Supervised Learning 半监督学习	A method of learning using a small amount of labeled data and a large amount of unlabeled data. Achieves high performance while suppressing labeling costs. 一种使用少量标记数据和大量未标记数据的学习方法。在降低标签成本的同时实现高性能。
Reinforcement Learning   强化学习	A method to learn actions that maximize rewards through interaction with the environment. Acquires optimal action policies through trial and error. 一种通过与环境交互来学习实现奖励最大化的行动的方法。通过反复试验获得最佳作策略。
Transfer Learning   迁移学习	A method to apply knowledge learned from one task to another task. Improves learning efficiency by utilizing pre-trained models. 一种将从一项任务中学到的知识应用于另一项任务的方法。利用预先训练的模型提高学习效率。
Batch Learning   批量学习	A learning method that processes all training data at once. Enables stable learning with good computational efficiency. Requires retraining when data is updated. 一种一次处理所有训练数据的学习方法。实现稳定的学习和良好的计算效率。更新数据时需要重新训练。
Online Learning   在线学习	A learning method that processes data sequentially and continuously updates the model. Quick adaptation to new patterns. Risk of instability. 一种按顺序处理数据并持续更新模型的学习方法。快速适应新模式。不稳定的风险。
Incremental Learning   增量学习	A method to perform additional learning on existing models with new data. Enables model updates without complete retraining. 一种使用新数据对现有模型执行额外学习的方法。无需完全重新训练即可启用模型更新。
Pre-training   训练前	The basic process of training a model from scratch with large-scale data. Acquires general knowledge and patterns. Forms the foundation for subsequent tasks. 使用大规模数据从头开始训练模型的基本过程。获得一般知识和模式。为后续任务奠定基础。
Fine-tuning   微调	A method to adapt pre-trained models to specific tasks. Achieves high performance even with small amounts of data. A type of transfer learning. 一种使预训练模型适应特定任务的方法。即使数据量很小，也能实现高性能。一种迁移学习。
Continuous Pre-training   持续的预培训	Periodic retraining with new data. Effective for maintaining performance of domain-specific models. Important as a drift countermeasure. 使用新数据定期重新训练。有效维护域特定模型的性能。作为漂移对策很重要。
RLHF	Reinforcement learning through human feedback. Improves quality and safety of generative AI models. Addresses alignment problems. 通过人工反馈进行强化学习。提高生成式 AI 模型的质量和安全性。解决对齐问题。
Custom Vocabulary Learning 自定义词汇学习	A method to train models on specialized terminology in specific fields. Important for building domain-specific models. Contributes to improving expertise. 一种根据特定领域的专业术语训练模型的方法。对于构建特定于域的模型非常重要。有助于提高专业知识。
Meta-learning   元学习	A method to learn the learning algorithm itself. Realizes high flexibility and efficiency for new tasks. The foundation of automatic ML systems. Enables quick adaptation to new tasks. 一种学习学习算法本身的方法。为新任务实现高度灵活性和效率。自动 ML 系统的基础。能够快速适应新任务。
Few-shot Learning   小样本学习	A method that enables learning from a small number of samples. A form of meta-learning. 一种支持从少量样本中学习的方法。元学习的一种形式。
Zero-shot Learning   零样本学习	A method that can infer even classes not seen during learning. An advanced form of transfer learning. 一种甚至可以推断学习过程中看不到的类的方法。一种高级的迁移学习形式。
Self-Supervised Learning 自我监督学习	A method to learn by automatically generating teaching signals from unlabeled data. Effective for pre-training. 一种通过从未标记的数据自动生成教学信号来学习的方法。对训练前有效。
Multi-task Learning   多任务学习	A method to learn multiple tasks simultaneously. Enables efficient learning through knowledge sharing between tasks. 一种同时学习多个任务的方法。通过任务之间的知识共享实现高效学习。
Federated Learning   联邦学习	A method to perform cooperative learning on multiple clients while keeping data distributed. Effective for privacy protection. 一种在多个客户端上执行协作学习同时保持数据分布式的方法。有效保护隐私。
Knowledge Distillation   知识蒸馏	A method to transfer knowledge from a large model to a small model. Used for model lightweighting. 一种将知识从大型模型转移到小型模型的方法。用于模型轻量化。

Model Optimization Methods
模型优化方法

Method   方法	Description   描述
Hyperparameter Tuning   超参数优化	Optimization of model configuration parameters. Uses grid search, random search, Bayesian optimization, etc. Consider the balance between computational cost and performance. 模型配置参数的优化。使用网格搜索、随机搜索、贝叶斯优化等。考虑计算成本和性能之间的平衡。
Regularization   正规化	Addition of penalty terms to prevent overfitting. L1 (Lasso), L2 (Ridge) regularization, etc. Controls model complexity. 添加了罚项以防止过度拟合。L1 （Lasso）、L2 （Ridge） 正则化等控制模型复杂性。
Early Stopping   提前停止	Ends learning when improvement in validation performance is no longer observed, preventing overfitting. Also contributes to efficient use of computational resources. 当不再观察到验证性能的改进时结束学习，防止过拟合。还有助于有效利用计算资源。
Cross-validation   交叉验证	Conducts evaluation by dividing data into multiple parts. Accurately estimates model's generalization performance. Particularly effective when data quantity is limited. 将数据划分为多个部分来进行评估。准确估计模型的泛化性能。当数据量有限时特别有效。
Ensemble Learning   集成学习	Improves performance by combining multiple models. Random Forest, Gradient Boosting, etc. Compensates for weaknesses of individual models. 通过组合多个模型来提高性能。Random Forest、Gradient Boosting 等弥补单个模型的弱点。
Gradient Descent   梯度下降	A method to search for optimal solutions by updating parameters based on the gradient of the loss function. There are variations such as Stochastic Gradient Descent (SGD) and Mini-batch Gradient Descent. 一种通过根据损失函数的梯度更新参数来搜索最优解的方法。有随机梯度下降 （SGD） 和小批量梯度下降等变体。
Learning Rate Adjustment 学习率调整	Adjustment of parameters that control the update amount in gradient descent. There are adaptive methods such as AdaGrad, Adam, RMSprop. 调整了控制梯度下降更新量的参数。有自适应方法，例如 AdaGrad、Adam 和 RMSprop。
Momentum   动量	A method to accelerate optimization by using past gradient information. Effective for avoiding local optima and accelerating convergence. 一种使用过去的梯度信息加速优化的方法。有效避免局部最优值并加速收敛。

Optimizers (Optimization Algorithms)
优化器（优化算法）

Term   术语	Description   描述
SGD (Stochastic Gradient Descent) SGD （随机梯度下降）	A basic optimization algorithm that calculates gradients and updates parameters on a mini-batch basis. 一种基本的优化算法，用于计算梯度并按小批量更新参数。
Adam   亚当	A popular optimization algorithm that combines momentum and adaptive learning rates. Shows good convergence in many cases. 一种结合了动量和自适应学习率的常用优化算法。在许多情况下显示出良好的收敛性。
RMSprop   RMS 属性	An adaptive optimization algorithm that considers past gradients using exponential moving average. 一种自适应优化算法，它使用指数移动平均线考虑过去的梯度。
AdaGrad	An adaptive optimization algorithm that applies different learning rates for each parameter. 一种自适应优化算法，可为每个参数应用不同的学习率。

Regularization Methods  正则化方法

Method   方法	Description   描述
Dropout   辍学	A method to prevent overfitting by randomly disabling neurons. 一种通过随机禁用神经元来防止过拟合的方法。
Batch Normalization   批量规范化	A method to normalize inputs on a mini-batch basis, stabilizing and accelerating learning. 一种在小批量基础上规范化输入的方法，以稳定和加速学习。
Layer Normalization   图层归一化	A method to perform normalization at the layer level. Effective for RNNs and Transformers. 一种在层级别执行归一化的方法。对 RNN 和 Transformer 有效。
Weight Decay   权重衰减	A regularization method that penalizes the magnitude of weights. Also called L2 regularization. 一种对权重的大小进行惩罚的正则化方法。也称为 L2 正则化。
Label Smoothing   标签平滑	A method to prevent model overconfidence by softening teacher labels. 一种通过软化教师标签来防止模型过度自信的方法。

Learning Rate Scheduling
学习率调度

Method   方法	Description   描述
Step Decay   步进衰减	A method to decrease the learning rate in stages every certain number of epochs. 一种每固定数量的 epoch 分阶段降低学习率的方法。
Exponential Decay   指数衰减	A method to decay the learning rate exponentially. 一种以指数方式衰减学习率的方法。
Cosine Annealing   余弦退火	A method to periodically change the learning rate based on a cosine function. 一种根据余弦函数定期更改学习率的方法。
Warm-up   热身	A method to gradually increase the learning rate at the beginning of learning, then transition to normal learning rate scheduling. 一种在学习开始时逐渐提高学习率，然后过渡到正常学习率调度的方法。

Model Evaluation  模型评估

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Clarify can evaluate bias detection and explainability (interpretation of predictions) in model evaluation.
* Amazon SageMaker Clarify 可以在模型评估中评估偏差检测和可解释性（预测的解释）。
* Amazon SageMaker Experiments provides tools for tracking and managing machine learning experiments, automatically recording experiment results such as training runs, parameters, metrics, and enabling comparative analysis.
* Amazon SageMaker Experiments 提供用于跟踪和管理机器学习实验的工具，自动记录实验结果（如训练运行、参数、指标）并启用比较分析。

Baseline Evaluation  基线评估

Metric   度量	Description   描述
Rule-based Baseline   基于规则的基线	Performance of a prediction model based on simple rules. Used as a baseline for improvement. 基于简单规则的预测模型的性能。用作改进的基线。
Random Baseline   随机基线	Performance when making random predictions. Used as a minimum performance standard. 进行随机预测时的性能。用作最低性能标准。
Industry Standard Baseline 行业标准基准	Performance standards generally accepted in the industry. Used as a benchmark for competitive comparison. 业界普遍接受的性能标准。用作竞争比较的基准。

Classification Tasks  分类任务

Metric   度量	Description   描述
Accuracy   准确性	The proportion of correct predictions among all predictions. Effective when classes are balanced. Caution is needed with imbalanced data. Often used in cases where the number of data points is similar across classes, such as image classification and document classification. Examples: handwritten character recognition, general object recognition tasks, etc. 正确预测在所有预测中的比例。当课程平衡时有效。需要谨慎处理不平衡的数据。通常用于不同类的数据点数量相似的情况，例如图像分类和文档分类。示例：手写字符识别、一般物体识别任务等。
Precision   精度	The proportion of true positives among positive predictions. Important when minimizing false positives is crucial. Emphasized in spam filters, etc. Used when the cost of false positives is high. Examples: spam email detection, fraudulent transaction detection, quality control inspection, etc., where incorrectly classifying normal items as abnormal can cause significant problems. 正预测中真阳性的比例。当最大限度地减少误报至关重要时，这一点很重要。在垃圾邮件过滤器等中强调。当误报成本较高时使用。示例：垃圾邮件检测、欺诈交易检测、质量控制检查等，其中错误地将正常商品分类为异常商品可能会导致重大问题。
Recall   召回	The proportion of correct predictions among actual positives. Important when minimizing false negatives is crucial. Emphasized in disease diagnosis, etc. Used when the cost of false negatives is high. Examples: cancer screening, security systems, earthquake prediction, etc., where missed detections can lead to serious consequences. 正确预测与实际正数的比例。当最大限度地减少假阴性至关重要时，这一点很重要。强调疾病诊断等。当假阴性的成本较高时使用。例如：癌症筛查、安全系统、地震预测等，其中漏检会导致严重后果。
F1 Score   F1 分数	The harmonic mean of precision and recall. A balanced evaluation metric. Used as a single comprehensive evaluation. Used when both precision and recall are important. Examples: information retrieval systems, product recommendation, document classification, etc., where both accuracy and comprehensiveness are required. 精确率和召回率的调和平均值。平衡的评估指标。用作单个综合评估。当精度和召回率都很重要时使用。例如：信息检索系统、产品推荐、文档分类等，其中需要准确性和全面性。
ROC Curve   ROC 曲线	A graph plotting true positive rate vs false positive rate for each classification threshold. Performance is evaluated by AUC (Area Under the Curve). Used when a comprehensive evaluation of model performance is desired or when determining the optimal classification threshold is necessary. Examples: credit scoring, medical diagnostic systems, risk assessment models, etc., where threshold adjustment is important. 绘制每个分类阈值的真阳性率与假阳性率的图表。性能通过 AUC （Area Under the Curve） 进行评估。当需要对模型性能进行全面评估或需要确定最佳分类阈值时使用。示例：信用评分、医疗诊断系统、风险评估模型等，其中阈值调整很重要。

Regression Tasks  回归任务

Metric   度量	Description   描述
MSE (Mean Squared Error) MSE （均方误差）	The average of the squared differences between predicted and actual values. As it squares the errors, it is strongly affected by outliers. Commonly used in general regression problems, especially when emphasizing outliers or when larger errors need to be penalized more severely. 预测值和实际值之间的平方差的平均值。当它对误差进行平方时，它会受到离群值的强烈影响。常用于一般回归问题，尤其是在强调异常值或需要更严厉地惩罚较大的错误时。
RMSE (Root Mean Squared Error) RMSE（均方根误差）	The square root of the mean squared error. Evaluates the magnitude of prediction errors in the original unit. Strongly affected by outliers. Suitable for tasks like housing price prediction or sales forecasting where interpreting the predicted values in the original scale is desired. Taking the square root of MSE allows for more intuitive interpretation. 均方误差的平方根。以原始单位计算预测误差的大小。受离群值影响很大。适用于需要解释原始比例中的预测值的任务，例如房价预测或销售预测。取 MSE 的平方根可以进行更直观的解释。
MAE (Mean Absolute Error) MAE （平均绝对误差）	An evaluation metric less sensitive to outliers. It's the average of the absolute differences between predicted and actual values. Suitable for demand forecasting or inventory management where you want to assess the average magnitude of errors while suppressing the impact of outliers. It doesn't overestimate prediction errors and is easy to understand intuitively. 对离群值不太敏感的评估量度。它是预测值和实际值之间绝对差值的平均值。适用于需求预测或库存管理，您希望评估误差的平均幅度，同时抑制异常值的影响。它不会高估预测误差，并且很容易直观地理解。
MAPE (Mean Absolute Percentage Error) MAPE （平均绝对百分比误差）	Evaluates relative errors. Allows comparison between data of different scales. Suitable for sales forecasting or stock price prediction where actual values are greater than 0 and you want to assess the relative magnitude of errors. Useful for comparing prediction accuracy across companies or products of different sizes. 计算相对误差。允许比较不同尺度的数据。适用于实际值大于 0 且您希望评估误差的相对大小的销售预测或股票价格预测。用于比较不同规模的公司或产品的预测准确性。
R² (Coefficient of Determination) R² （决定系数）	Expresses the goodness of fit of a model on a scale from 0 to 1, with values closer to 1 indicating higher prediction accuracy. Can also take negative values. Used to evaluate the overall explanatory power of a model. Particularly used as an indicator for variable selection in multiple regression analysis and is useful for model comparison and selection. 以 0 到 1 的等级表示模型的拟合优度，值越接近 1 表示预测准确性越高。也可以取负值。用于评估模型的整体解释能力。特别用作多元回归分析中变量选择的指标，可用于模型比较和选择。
Adjusted R²   调整后的 R²	A modified version of R². It corrects for the influence of the number of explanatory variables, allowing for more accurate model evaluation. Suitable for variable selection and model comparison, especially when comparing models with different numbers of explanatory variables. Used to prevent overfitting. R² 的修改版本。它校正了解释变量数量的影响，从而可以进行更准确的模型评估。适用于变量选择和模型比较，尤其是在比较具有不同解释变量数量的模型时。用于防止过拟合。
RMSLE (Root Mean Squared Logarithmic Error) RMSLE（均方根对数误差）	The RMSE after taking the logarithm of predicted and actual values. Evaluates relative errors and mitigates the impact of large values. Suitable for sales forecasting or population prediction where the range of data values is wide and relative errors are emphasized. Particularly useful when predicted values have vastly different scales. 取预测值和实际值的对数之后的 RMSE。评估相对误差并减轻较大值的影响。适用于数据值范围广且强调相对误差的销售预测或人口预测。当预测值具有截然不同的尺度时，尤其有用。
MSLE (Mean Squared Logarithmic Error) MSLE（均方对数误差）	The MSE after taking the logarithm of predicted and actual values. Evaluates relative errors and mitigates the impact of large values. Used for similar purposes as RMSLE, particularly suitable for cases where predicted values increase exponentially or for predicting ratios. 取预测值和实际值的对数之后的 MSE。评估相对误差并减轻较大值的影响。用于与 RMSLE 类似的目的，特别适用于预测值呈指数增长的情况或用于预测比率。
MedAE (Median Absolute Error) MedAE （中位绝对误差）	The median of absolute errors. The evaluation metric least affected by outliers. Particularly useful for prediction tasks with noisy datasets or when outliers are present. Suitable for analyzing sensor data or evaluating actual measurement data that may contain anomalies. 绝对误差的中位数。受离群值影响最小的评估量度。对于具有嘈杂数据集或存在异常值的预测任务特别有用。适用于分析传感器数据或评估可能包含异常的实际测量数据。

Text Generation  文本生成

Metric   度量	Description   描述
ROUGE	Evaluates the similarity between generated text and reference text. Calculates n-gram matching. Commonly used for summarization tasks. Particularly effective for evaluating the performance of news article summarization and document summarization systems. 评估生成的文本和引用文本之间的相似性。计算 n 元语法匹配。通常用于摘要任务。对于评估新闻文章摘要和文档摘要系统的性能特别有效。
Human Evaluation   人工评价	Subjectively assesses quality, coherence, relevance, etc. Sets qualitative evaluation criteria and judges with multiple evaluators. Used when subtle nuances and contextual understanding that cannot be fully captured by automatic evaluation metrics are required, or when evaluating creative text generation. 主观评估质量、连贯性、相关性等。设定定性评价标准，并与多个评价者进行评判。当需要自动评估指标无法完全捕获的细微差别和上下文理解时，或在评估创意文本生成时使用。

Evaluation Metrics for Generative AI Models
生成式 AI 模型的评估指标

Metric   度量	Description   描述
BLEU	A metric used for evaluating machine translation. Calculates n-gram matching between generated and reference sentences. Particularly effective for quality assessment of multilingual translation systems and comparison of different translation models. 用于评估机器翻译的指标。计算生成句子和引用句子之间的 n 元语法匹配。对于多语言翻译系统的质量评估和不同翻译模型的比较特别有效。
METEOR	An evaluation metric for translation and generated text. Allows for flexible evaluation considering synonyms and morphological variations. Used in translation tasks where there are significant differences in grammatical structures between languages or when considering diversity of expressions. 翻译和生成文本的评估量度。允许考虑同义词和形态变化进行灵活评估。用于语言之间语法结构存在显著差异的翻译任务，或考虑表达方式的多样性。
BERTScore   BERTS 核心	A metric that evaluates semantic similarity of sentences using BERT's contextual word embeddings. Used when semantic similarity assessment is important beyond surface-level matching, or when evaluating paraphrasing. 使用 BERT 的上下文词嵌入来评估句子语义相似性的指标。当语义相似性评估的重要性超出表面匹配或评估释义时使用。
Perplexity   困惑	A metric for evaluating the predictive performance of language models. Lower values indicate better models. Used for evaluating the learning process of language models and comparing language models with different architectures. 用于评估语言模型预测性能的指标。值越低表示模型越好。用于评估语言模型的学习过程和比较不同架构的语言模型。

Explainability Evaluation
可解释性评估

Method   方法	Description   描述
LIME	A method providing local explainability. Generates interpretable explanations for individual predictions. Used in cases where explanation of individual decision bases is important, such as medical diagnostics or financial credit assessments. 提供本地可解释性的方法。为单个预测生成可解释的解释。用于解释个人决策基础很重要的情况，例如医疗诊断或财务信用评估。
SHAP	A feature importance calculation method based on game theory. Evaluates the contribution of each feature to predictions. Used when there is a need to understand the decision-making process of complex models or when ranking feature importance. 一种基于博弈论的特征重要性计算方法。评估每个特征对预测的贡献。当需要了解复杂模型的决策过程或对特征重要性进行排名时使用。
Attention Visualization   注意力可视化	Visualization of the attention mechanism in Transformer models. Visually represents the basis of model decisions. Used to confirm the areas of focus in natural language processing tasks or when analyzing model behavior. Transformer 模型中注意力机制的可视化。直观地表示模型决策的基础。用于确认自然语言处理任务中的关注区域或分析模型行为时的关注区域。
Feature Attribution   特征归因	A method to quantify the contribution of each feature to prediction results. Analyzes the decision process of models. Used for evaluating model fairness or detecting bias when necessary. 一种量化每个特征对预测结果的贡献的方法。分析模型的决策过程。用于评估模型公平性或在必要时检测偏差。

Correlation Analysis  相关分析

Method   方法	Description   描述
Pearson Correlation   皮尔逊相关性	Measures the strength of linear relationships on a scale from -1 to 1. Used for evaluating relationships between continuous variables. Applied when analyzing variables expected to have a linear relationship, such as height and weight. 在 -1 到 1 的范围内测量线性关系的强度。用于评估连续变量之间的关系。在分析预期具有线性关系的变量（如身高和体重）时应用。
Spearman Correlation   Spearman 相关性	Evaluates the strength of ordinal relationships. Applicable to non-linear relationships. Suitable for ordinal variables. Used when analyzing monotonic but not necessarily linear relationships, such as between customer satisfaction and purchase amount. 评估序数关系的强度。适用于非线性关系。适用于有序变量。在分析单调但不一定是线性关系时使用，例如客户满意度和购买金额之间的关系。
Chi-square Test   卡方检验	Statistically tests the association between categorical variables. Used for verifying independence. Applied when analyzing relationships between categorical data, such as the association between gender and product selection. 对分类变量之间的关联进行统计检验。用于验证独立性。在分析分类数据之间的关系（如性别与产品选择之间的关联）时应用。
Phi Coefficient   Phi 系数	Measures correlation between binary variables. Applied to 2x2 contingency table data. Used when measuring the strength of relationships between binary data, such as the association between pass/fail and male/female. 测量二进制变量之间的相关性。应用于 2x2 列联表数据。在测量二进制数据之间关系的强度时使用，例如通过/失败和男性/女性之间的关联。

Model Challenges and Phenomena
模拟挑战和现象

Term   术语	Description   描述
Overfitting   过拟合	A state where the model excessively fits to the training data, reducing generalization performance on new data. Also called overlearning, balance between model complexity and training data amount is important. 模型过度拟合训练数据，从而降低新数据的泛化性能的状态。也称为过度学习，模型复杂性和训练数据量之间的平衡很重要。
Underfitting   欠拟合	A state where the model fails to capture patterns in the training data sufficiently. Caused by lack of model expressiveness or insufficient learning. Requires more complex models or additional learning. 模型无法充分捕获训练数据中的模式的状态。由于缺乏模型表现力或学习不足而引起。需要更复杂的模型或额外的学习。
Bias   偏见	Systematic error between model predictions and true values. Increases when model expressiveness is insufficient, causing underfitting. 模型预测与真实值之间的系统误差。当模型表现力不足时增加，导致欠拟合。
Variance   方差	Variability in model predictions. The magnitude of prediction value fluctuations to small changes in training data. If too high, it can cause overfitting. 模型预测的可变性。预测值波动到训练数据中微小变化的幅度。如果太高，可能会导致过度拟合。
Hallucination   幻觉	When AI models generate incorrect information not based on facts. Particularly problematic in generative AI, can be mitigated by techniques like RAG. 当 AI 模型生成不正确的信息而不是基于事实时。在生成式 AI 中尤其存在问题，可以通过 RAG 等技术来缓解。
Drift   漂移	Changes in data distribution or model performance over time. Includes concept drift (changes in target variable relationships) and data drift (changes in input distribution). 数据分布或模型性能随时间的变化。包括概念漂移 （目标变量关系的变化） 和数据漂移 （输入分布的变化）。

Dealing with Data Imbalance
处理数据不平衡

Method   方法	Description   描述
Oversampling   过采样	A method to increase data of minority classes. Includes synthetic data generation like SMOTE. Improves class balance. Effective when overall data quantity is small and you want to maximize use of information, or when there are ample computational resources. 一种增加少数类数据的方法。包括 SMOTE 等合成数据生成。改善兵种平衡。当总体数据量较小且您希望最大限度地利用信息时，或者当有充足的计算资源时有效。
Undersampling   欠采样	A method to balance by reducing data from majority classes. Risk of information loss but computationally efficient. Effective when data quantity is sufficient and there are strict constraints on computation time or memory. 一种通过减少来自多数类的数据来平衡的方法。存在信息丢失风险，但计算效率高。当数据量足够且对计算时间或内存有严格限制时有效。
SMOTE	A method to generate synthetic data for minority classes. Uses k-nearest neighbors to generate new samples. Ensures data diversity. Effective when simple duplication risks overfitting or when you want to learn more diverse features of minority classes. 一种为少数类生成合成数据的方法。使用 k 最近邻生成新样本。确保数据多样性。当简单的重复存在过度拟合的风险或您想要了解少数类的更多不同特征时有效。
Class Weighting   类权重	Adjusts balance by weighting minority classes during learning. Modifies the model's loss function to balance. Effective when you want to maintain the original data distribution or avoid data modification. 通过在学习过程中对少数类进行加权来调整平衡。将模型的损失函数修改为 balance。当您希望保持原始数据分布或避免数据修改时有效。

Bias and Fairness Evaluation
偏差和公平性评估

Metric   度量	Description   描述
Demographic Parity   人口平价	Evaluates the uniformity of prediction result distribution across different demographic groups. 评估不同人口统计群体之间预测结果分布的一致性。
Equal Opportunity   机会均等	An indicator that confirms true positive rates are equal across protected attributes. 确认受保护属性的真阳性率相等的指示器。
Predictive Parity   预测奇偶校验	Evaluates the consistency of prediction accuracy between different groups. 评估不同组之间预测准确性的一致性。
Individual Fairness   个人公平性	Evaluates the consistency of predictions for individuals with similar characteristics. 评估具有相似特征的个体的预测一致性。
Bias Amplification	Measures the degree to which the model amplifies existing biases in the data. 度量模型放大数据中现有偏差的程度。

Performance Stability Evaluation
性能稳定性评估

Metric   度量	Description   描述
Prediction Variance   预测方差	Evaluates the variability of model predictions. Used as an indicator of stability. 评估模型预测的可变性。用作稳定性指标。
Threshold Stability   阈值稳定性	Evaluates the robustness of performance to changes in classification thresholds. 评估性能对分类阈值变化的稳健性。
Cross-validation Standard Deviation 交叉验证标准差	Evaluates the variability of performance across different data splits. 评估不同数据分片的性能变化。
Noise Resistance   抗噪性	Evaluates the stability of predictions against input noise. 根据输入噪声评估预测的稳定性。
Temporal Stability   时间稳定性	Evaluates the consistency of prediction performance in time series data. 评估时间序列数据中预测性能的一致性。

Reliability and Robustness Evaluation
可靠性和稳健性评估

Metric   度量	Description   描述
Adversarial Attack Resistance 对抗性攻击抵抗	Evaluates the model's robustness against adversarial samples. Identifies security vulnerabilities. 评估模型对对抗性样本的稳健性。识别安全漏洞。
Model Uncertainty   模型不确定性	Quantification of prediction confidence and uncertainty. Evaluation using Bayesian methods or ensemble methods. 预测置信度和不确定性的量化。使用贝叶斯方法或集成方法进行评估。
Stress Test   压力测试	Evaluation of model behavior in extreme cases or boundary conditions. Understanding system limitations. 评估极端情况或边界条件下的模型行为。了解系统限制。
Data Quality Sensitivity 数据质量敏感度	Evaluation of model sensitivity to deterioration in input data quality. Used as an indicator of robustness. 评估模型对输入数据质量恶化的敏感性。用作稳健性的指标。
Fail-safe Property   故障安全属性	Evaluation of safety in case of model abnormal operation. Confirmation of fallback mechanism effectiveness. 模型异常运行情况下的安全性评估。确认回退机制的有效性。

Cost Efficiency Evaluation
成本效益评估

Metric   度量	Description   描述
Computational Cost   计算成本	Evaluation of computational resources required for model training and inference. GPU time, memory usage, etc. 评估模型训练和推理所需的计算资源。GPU 时间、内存使用情况等
Infrastructure Cost   基础设施成本	Evaluation of infrastructure costs required for model operation. Storage, network, etc. 评估模型运行所需的基础设施成本。存储、网络等
Maintenance Cost   维护成本	Evaluation of human resources and time required for model maintenance and updates. 评估模型维护和更新所需的人力资源和时间。
ROI Analysis   ROI 分析	Evaluation of return on investment from model deployment. Quantification of cost reduction or revenue increase. 评估模型部署的投资回报。成本降低或收入增加的量化。

Model Deployment  模型部署

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Model Registry catalogs and version-manages ML models, managing model metadata.
* Amazon SageMaker Model Registry 对 ML 模型进行编目和版本管理，管理模型元数据。

Deployment Strategies  部署策略

Term   术语	Description   描述
Canary Deployment   金丝雀部署	A technique that applies a new version of the model to only a portion of the traffic and gradually expands. Allows validation of new models while minimizing risk. 一种将新版本的模型仅应用于部分流量并逐渐扩展的技术。允许验证新模型，同时最大限度地降低风险。
Blue/Green Deployment   蓝/绿部署	A deployment method that prepares production (blue) and new (green) environments in parallel and switches between them. Enables immediate rollback. 一种部署方法，可并行准备生产 （蓝色） 和新 （绿色） 环境，并在它们之间切换。启用立即回滚。
Shadow Deployment   影子部署	A method that mirrors production traffic to a new model for parallel evaluation. Enables performance verification under actual workloads. 一种将生产流量镜像到新模型以进行并行评估的方法。支持在实际工作负载下进行性能验证。
A/B Testing   A/B 测试	A method to operate multiple versions of models simultaneously and compare their performance. Enables data-driven decision making. 一种同时作多个版本的模型并比较其性能的方法。支持数据驱动的决策。
Rolling Update   滚动更新	A gradual deployment method that updates instances sequentially. Minimizes service interruption. 一种按顺序更新实例的逐步部署方法。最大限度地减少服务中断。
Rollback Plan   回滚计划	Setting of recovery procedures and trigger conditions in case of problems. Ensures consistency of data and models. 设置恢复程序和出现问题时的触发条件。确保数据和模型的一致性。

Inference Options  推理选项

Term   术语	Description   描述
Real-time Inference   实时推理	A method that executes inference in real-time for requests. Suitable for use cases requiring low latency. In Amazon SageMaker, it's provided as persistent, fully managed endpoints that can handle payloads up to 6MB and processing times up to 60 seconds. A scalable solution capable of handling continuous traffic. 一种对请求实时执行推理的方法。适用于需要低延迟的使用案例。在 Amazon SageMaker 中，它作为持久性、完全托管的终端节点提供，可以处理高达 6MB 的负载和长达 60 秒的处理时间。能够处理连续流量的可扩展解决方案。
Batch Inference   批量推理	A method that executes inference in bulk for large amounts of data. Suitable for periodic prediction processing. In Amazon SageMaker, it's provided as batch transform, capable of processing large-scale datasets of several GB. Optimal for offline processing or preprocessing that doesn't require persistent endpoints. 一种对大量数据批量执行推理的方法。适用于周期性预测处理。在 Amazon SageMaker 中，它以批量转换的形式提供，能够处理数 GB 的大规模数据集。最适合离线处理或不需要持久性终端节点的预处理。
Asynchronous Inference   异步推理	A method that queues and processes inference requests requiring large payloads or long processing times. In Amazon SageMaker, it supports payloads up to 1GB and processing times up to 1 hour. Can scale down to 0 when there's no traffic. 一种对需要大量负载或较长处理时间的推理请求进行排队和处理的方法。在 Amazon SageMaker 中，它支持高达 1GB 的负载和长达 1 小时的处理时间。当没有流量时，可以缩减到 0。
Serverless Inference   无服务器推理	An event-driven inference execution method. Automatically scales according to demand. In Amazon SageMaker, it provides a model that requires no infrastructure management and charges only for usage, for intermittent or unpredictable traffic. Supports payloads up to 4MB and processing times up to 60 seconds. 一种事件驱动的推理执行方法。根据需求自动扩展。在 Amazon SageMaker 中，它提供了一种不需要基础设施管理的模型，并且只对使用量、间歇性或不可预测的流量收费。支持高达 4MB 的有效负载和长达 60 秒的处理时间。

Endpoint Options  端点选项

Term   术语	Description   描述
Single Model Endpoint   单一模型终端节点	A basic endpoint configuration for deploying a single model. Simple and easy to manage. 用于部署单个模型的基本终端节点配置。简单易懂。
Multi-Model Endpoint   多模型终端节点	An endpoint configuration that serves multiple models of the same framework in a single container. In Amazon SageMaker, it improves endpoint utilization and reduces deployment overhead, realizing cost optimization. 在单个容器中为同一框架的多个模型提供服务的终端节点配置。在 Amazon SageMaker 中，它提高了终端节点利用率并降低了部署开销，从而实现了成本优化。
Multi-Container Endpoint 多容器终端节点	An endpoint configuration that serves multiple models of different frameworks in separate containers. In Amazon SageMaker, it allows flexible deployment of various frameworks and models. 一种终端节点配置，可在单独的容器中为不同框架的多个模型提供服务。在 Amazon SageMaker 中，它允许灵活部署各种框架和模型。
Serial Inference Pipeline 串行推理管道	An endpoint configuration that executes preprocessing, inference, and post-processing as a series of pipelines. In Amazon SageMaker, all containers are hosted on the same EC2 instance and fully managed, achieving low latency. 一种终端节点配置，用于将预处理、推理和后处理作为一系列管道执行。在 Amazon SageMaker 中，所有容器都托管在同一个 EC2 实例上，并且完全托管，从而实现低延迟。
Scalable Endpoint   可扩展终端节点	An endpoint configuration that automatically scales according to load. Flexibly responds to traffic fluctuations. 根据负载自动扩展的终端节点配置。灵活应对流量波动。
High Availability Endpoint 高可用性终端节点	An endpoint configuration deployed across multiple availability zones, ensuring redundancy. 跨多个可用区部署的终端节点配置，确保冗余。

Infrastructure  基础设施

Term   术语	Description   描述
Model Containerization   模型容器化	Packaging of models using container technologies like Docker. Ensures consistency and portability of environments. 使用 Docker 等容器技术打包模型。确保环境的一致性和可移植性。
Scaling Strategy   扩展策略	Setting of automatic scaling according to load. Selection and policy setting of horizontal/vertical scaling. 根据负载设置自动缩放。水平/垂直缩放的选择和策略设置。
Service Mesh   服务网格	Management of traffic control and inter-service communication in microservice architectures. 在微服务架构中管理流量控制和服务间通信。
Deployment Pipeline   部署管道	Automation and standardization of deployments. Construction of CI/CD pipelines and setting of quality gates. 部署的自动化和标准化。构建 CI/CD 管道并设置质量关卡。

Optimization and Security
优化和安全性

Term   术语	Description   描述
Model Optimization   模型优化	Model optimization before deployment. Lightweighting and acceleration through quantization, pruning, distillation, etc. 部署前的模型优化。通过量化、修剪、蒸馏等实现轻量化和加速。
Security Settings   安全设置	Setting of access control, encryption, authentication and authorization. Configuration of secure endpoints. 访问控制、加密、认证和授权的设置。安全端点的配置。
API Versioning   API 版本控制	Version management of model APIs and ensuring compatibility between versions. 模型 API 的版本管理，确保版本之间的兼容性。
Monitoring Settings   监控设置	Configuration of metric collection, logging, alert settings. Establishment of performance and quality monitoring system. 配置指标收集、日志记录、警报设置。建立绩效和质量监控系统。

Inference  推理

Prompting Techniques  提示技术

Technique   技术	Description   描述
Prompt Engineering   快速工程	A technique to obtain desired outputs by crafting inputs (prompts) to AI models. Optimizes methods of setting context and presenting constraints. 一种通过为 AI 模型制作输入（提示）来获得所需输出的技术。优化设置上下文和呈现约束的方法。
Zero-shot Prompting   零样本提示	One of the prompt engineering techniques. Executes tasks directly without examples. Utilizes the model's generalization ability to handle new tasks. 提示工程技术之一。直接执行任务，无需示例。利用模型的泛化功能来处理新任务。
Few-shot Prompting   Few-shot 提示	One of the prompt engineering techniques. Teaches how to execute tasks by showing a few examples. Controls model behavior through concrete examples. 提示工程技术之一。通过展示几个示例来教授如何执行任务。通过具体示例控制模型行为。
Chain-of-Thought Prompting 思路提示	One of the prompt engineering techniques. Guides the model to solve complex problems step by step (Chain of Thought). Encourages explicit expansion of the reasoning process. 提示工程技术之一。指导模型逐步解决复杂问题 （Chain of Thought）。鼓励显式扩展推理过程。

Prompt Optimization Techniques
提示优化技术

Technique   技术	Description   描述
Prompt Templates   提示模板	Design of reusable fixed prompts. Ensures consistent outputs. 设计可复用的固定提示。确保一致的输出。
Hallucination Countermeasures 幻觉对策	Techniques to prevent generation not based on facts. Incorporation of knowledge base references and fact-checking. 防止不基于事实的生成的技术。纳入知识库参考和事实核查。
Context Management   上下文管理	Effective setting and control of context information in prompts. Leads to more accurate responses. 有效设置和控制 Prompt 中的上下文信息。导致更准确的响应。
Prompt Variation   提示变化	Experimentation and optimization of different expression methods for the same intent. Improves robustness. 针对同一目的对不同的表达方法进行实验和优化。提高稳健性。

Monitoring  监测

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Model Monitor continuously monitors model quality in production environments and detects data drift and bias.
* Amazon SageMaker Model Monitor 持续监控生产环境中的模型质量，并检测数据漂移和偏差。
* Amazon SageMaker Clarify can also be used in model quality monitoring, continuously evaluating model bias and explainability.
* Amazon SageMaker Clarify 还可用于模型质量监控，持续评估模型偏差和可解释性。

Model Quality Monitoring
模型质量监控

Term   术语	Description   描述
Data Drift Detection   数据偏差检测	Monitors changes in input data distribution. Used as an indicator to determine timing for model retraining. 监视输入数据分布的变化。用作确定模型重新训练时间的指标。
Concept Drift Detection   概念漂移检测	Monitors changes in the relationship between inputs and outputs. Used as an indicator to determine the need for model updates. 监视输入和输出之间关系的变化。用作确定模型更新需求的指示器。
Prediction Quality Monitoring 预测质量监控	Continuously evaluates the quality of model prediction results. Includes monitoring of bias and fairness. 持续评估模型预测结果的质量。包括对偏见和公平性的监控。
Explainability Monitoring 可解释性监控	Monitors metrics related to model explainability. Ensures transparency and reliability of predictions. 监控与模型可解释性相关的指标。确保预测的透明度和可靠性。
Input Data Validation   输入数据验证	Continuously validates the validity of input data schema, type, range, etc. 持续验证输入数据架构、类型、范围等的有效性。
Data Completeness Monitoring 数据完整性监控	Monitors data quality indicators such as missing values, outliers, duplicates. 监控数据质量指标，例如缺失值、异常值、重复项。
Feature Stability Monitoring 特征稳定性监控	Tracks changes in statistical properties of features. Detects distribution shifts. 追踪特征统计属性的变化。检测分布偏移。
Data Source Monitoring   数据源监控	Monitors availability, freshness, and consistency of data sources. 监控数据源的可用性、新鲜度和一致性。

Performance Monitoring  性能监控

Term   术语	Description   描述
Latency Monitoring   延迟监控	Monitors inference processing time. Checks SLA compliance status and used for performance optimization. 监控推理处理时间。检查 SLA 合规性状态并用于性能优化。
Throughput Monitoring   吞吐量监控	Monitors the number of processes per unit time. Used for capacity planning. 监控单位时间内的进程数。用于容量规划。
Resource Utilization Monitoring 资源利用率监控	Monitors infrastructure metrics such as CPU, memory, disk usage. Used for scaling decisions. 监控基础设施指标，例如 CPU、内存、磁盘使用情况。用于扩展决策。
Error Rate Monitoring   错误率监控	Monitors the occurrence rate of inference errors and system errors. Used for maintaining service quality. 监控推理错误和系统错误的发生率。用于维护服务质量。

Security Monitoring  安全监控

Term   术语	Description   描述
Access Monitoring   访问监控	Monitors access patterns and authentication status to API endpoints. Detection of unauthorized access. 监控 API 端点的访问模式和身份验证状态。检测未经授权的访问。
Data Security Monitoring 数据安全监控	Monitors data encryption status, access control, and privacy protection status. 监控数据加密状态、访问控制和隐私保护状态。
Compliance Monitoring   合规性监控	Continuously monitors compliance with regulatory requirements. Maintenance of audit trails. 持续监控对法规要求的遵守情况。维护审计跟踪。

Operational Monitoring  运营监控

Term   术语	Description   描述
Alert Settings   警报设置	A mechanism to notify when important metrics exceed thresholds. Enables early response. 一种在重要指标超过阈值时发出通知的机制。启用早期响应。
Log Analysis   日志分析	Analysis of system logs and application logs. Used for identifying causes of failures and trend analysis. 分析系统日志和应用程序日志。用于确定失败的原因和趋势分析。
Incident Tracking   事件跟踪	Tracking of occurrence history and response status of failures and abnormalities. Used for formulating recurrence prevention measures. 跟踪故障和异常的发生历史记录和响应状态。用于制定防止复发的措施。
Capacity Management   容量管理	Prediction and planning of resource usage. Formulation of appropriate scaling strategies. 预测和规划资源使用情况。制定适当的扩展策略。

Business Impact Monitoring
业务影响监控

Term   术语	Description   描述
ROI Analysis   ROI 分析	Continuous evaluation of costs and effects of model operation. Measurement of return on investment. 持续评估模型作的成本和效果。投资回报率的衡量。
Business Metrics   业务指标	Monitoring of indicators showing the business contribution of the model. Measurement of effects such as sales and cost reduction. 监控显示模型业务贡献的指标。衡量销售额和成本降低等效果。
User Satisfaction   用户满意度	Tracking of feedback and service evaluations from end users. 跟踪来自最终用户的反馈和服务评估。

System Health Monitoring
系统运行状况监控

Term   术语	Description   描述
Infrastructure Availability Monitoring 基础设施可用性监控	Operational status and health checks of system components. 系统组件的运行状态和运行状况检查。
Network Monitoring   网络监控	Monitoring of network connectivity, latency, bandwidth. 监控网络连接、延迟、带宽。
Cache Efficiency Monitoring 缓存效率监控	Tracking of cache hit rates, memory usage efficiency. 跟踪缓存命中率、内存使用效率。
Batch Processing Monitoring 批处理监控	Monitoring of batch job execution status, success rates, processing times. 监控批处理作业执行状态、成功率、处理时间。

Fairness Monitoring  公平性监控

Term   术语	Description   描述
Bias Metrics   偏差指标	Monitoring of demographic biases in model predictions. Tracking of fairness indicators. 监测模型预测中的人口偏差。公平性指标跟踪。
Attribute-based Monitoring 基于属性的监控	Monitoring of prediction biases based on protected attributes. Detection of discriminatory results. 监控基于 protected attributes 的预测偏差。检测歧视性结果。
Fairness Score   公平性分数	Evaluation of prediction accuracy uniformity across different groups. Quantitative measurement of fairness. 评估不同组之间的预测准确性均匀性。公平性的定量衡量。
Impact Analysis   影响分析	Analysis of the impact of model predictions on different populations. Evaluation of social impact. 分析模型预测对不同人群的影响。社会影响评估。

Governance Monitoring  治理监控

Term   术语	Description   描述
Policy Compliance   策略合规性	Monitoring of compliance with the organization's AI governance policies. Confirmation of guideline adherence. 监控对组织 AI 治理策略的遵守情况。确认指南遵守情况。
Accountability Tracking   问责制跟踪	Monitoring of transparency and explainability in the model's decision-making process. Ensuring accountability. 监控模型决策过程中的透明度和可解释性。确保问责制。
Ethical Risk Monitoring   道德风险监控	Continuous assessment of AI's ethical impacts and potential risks. Fulfillment of social responsibility. 持续评估 AI 的道德影响和潜在风险。履行社会责任。
Regulatory Compliance Tracking 监管合规跟踪	Monitoring of compliance status with new regulatory requirements. Maintenance of compliance. 监控新法规要求的合规性状态。维护合规性。

MLOps Management Process
MLOps 管理流程

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Studio provides an integrated development environment (IDE) that enables one-stop execution from notebook creation to model development, training, and deployment, realizing centralized management of ML workflows.
* Amazon SageMaker Studio 提供集成开发环境 （IDE），支持从笔记本创建到模型开发、训练和部署的一站式执行，实现 ML 工作流的集中管理。
* Amazon SageMaker Canvas provides a no-code ML development environment that allows data preparation to model deployment through drag & drop without writing code, enabling development for business analysts.
* Amazon SageMaker Canvas 提供了一个无代码的 ML 开发环境，允许数据准备通过拖放进行模型部署，而无需编写代码，从而为业务分析师提供开发支持。
* Amazon SageMaker Pipelines orchestrates ML workflows, building reproducible ML pipelines.
* Amazon SageMaker Pipelines 编排 ML 工作流，构建可重现的 ML 管道。

Experiment Management  实验管理

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Experiments provides tools for tracking and managing machine learning experiments, automatically recording experiment results such as training runs, parameters, metrics, and enabling comparative analysis.
* Amazon SageMaker Experiments 提供用于跟踪和管理机器学习实验的工具，自动记录实验结果（如训练运行、参数、指标）并启用比较分析。

Term   术语	Description   描述
Experiment Tracking   实验跟踪	Activity of recording and managing settings, parameters, and results of each experiment in model development. Uses tools like MLflow, SageMaker Experiments. 在模型开发中记录和管理每个实验的设置、参数和结果的活动。使用 MLflow、SageMaker Experiments 等工具。
Metadata Management   元数据管理	Management of associated information such as settings, environment, datasets, results related to experiments. Ensures reproducibility and traceability of experiments. 管理相关信息，例如设置、环境、数据集、与实验相关的结果。确保实验的可重复性和可追溯性。
Hyperparameter Logging   超参数日志记录	History management of model hyperparameter settings. Used for tracking and comparative analysis of optimization processes. 模型超参数设置的历史记录管理。用于优化过程的跟踪和比较分析。
Evaluation Metrics Tracking 评估指标跟踪	Time-series recording and analysis of model performance indicators. Used for understanding improvement trends and comparative evaluation. 模型性能指标的时间序列记录和分析。用于了解改进趋势和比较评估。
Artifact Management   工件管理	Storage and management of artifacts such as models, checkpoints, plots. Streamlines storage and sharing of experiment results. 存储和管理工件，例如模型、检查点、绘图。简化实验结果的存储和共享。
A/B Test Management   A/B 测试管理	Design and result management of comparative experiments of multiple models. Supports statistical significance evaluation and decision making. 多个模型比较实验的设计和结果管理。支持统计显著性评估和决策。
Experiment Environment Management 实验环境管理	Management of development environment configurations, dependencies, resource settings, etc. Maintains reproducibility and consistency of environments. 管理开发环境配置、依赖关系、资源设置等保持环境的可重复性和一致性。

Version Control  版本控制

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Model Registry catalogs and version-manages ML models, managing model metadata.
* Amazon SageMaker Model Registry 对 ML 模型进行编目和版本管理，管理模型元数据。

Term   术语	Description   描述
Data Versioning   数据版本控制	Change history management of training datasets. Uses tools like DVC (Data Version Control). Enables tracking of data lineage. 训练数据集的更改历史记录管理。使用 DVC（数据版本控制）等工具。启用数据沿袭跟踪。
Model Versioning   模型版本控制	Management of different versions of trained models. Implemented with tools like SageMaker Model Registry. Controls switching and rollback in production environments. 管理不同版本的训练模型。使用 SageMaker Model Registry 等工具实施。控制生产环境中的切换和回滚。
Code Version Control   代码版本控制	Version management of model development code. Uses Git, etc. Setting of branch strategies and merge policies. 模型开发代码的版本管理。使用 Git 等设置分支策略和合并策略。
Configuration File Management 配置文件管理	Version management of configuration files such as environment settings, parameter settings. Ensures consistency across environments. 环境设置、参数设置等配置文件的版本管理。确保跨环境的一致性。
Dependency Management   依赖关系管理	Version management of libraries and frameworks. Clarified in requirements.txt or Dockerfile. 库和框架的版本管理。在 requirements.txt 或 Dockerfile 中阐明。
Tagging   标记	Assigning meaningful tags to versions of models, data, code. Facilitates release management and tracking. 为模型、数据、代码的版本分配有意义的标签。促进发布管理和跟踪。
Baseline Management   基线管理	Management of model versions that serve as benchmarks for performance comparison. Used as indicators for quality assurance. 管理用作性能比较基准的模型版本。用作质量保证的指标。

Documentation  文档

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Model Cards creates and manages model documentation, centralizing management of detailed model information.
* Amazon SageMaker Model Cards 创建和管理模型文档，集中管理详细的模型信息。

Term   术语	Description   描述
Model Card   型号卡	A standardized document recording detailed information about the model. Includes usage, performance, limitations, ethical considerations, etc. Can be managed with SageMaker Model Cards. 记录有关模型的详细信息的标准化文档。包括用法、性能、限制、道德考虑等。可以使用 SageMaker Model Cards 进行管理。
Data Sheet   数据表	A document recording characteristics of datasets, collection methods, preprocessing procedures, license information, etc. Ensures transparency and reusability of data. 记录数据集特征、收集方法、预处理过程、许可证信息等的文档。确保数据的透明度和可重用性。
API Specification   API 规范	Describes model interface specifications, input/output formats, endpoint information, etc. Managed in standard formats like OpenAPI/Swagger. 描述模型接口规范、输入/输出格式、终端节点信息等。以 OpenAPI/Swagger 等标准格式进行管理。
Experiment Report   实验报告	A document summarizing the purpose, method, results, and discussion of experiments. Records important findings and decisions. 总结实验目的、方法、结果和讨论的文档。记录重要的发现和决策。
Operation Manual   作手册	A manual describing procedures for model deployment, monitoring, and maintenance. Includes incident response procedures. 描述模型部署、监控和维护过程的手册。包括事件响应程序。
Training Record   培训记录	Detailed record of model learning. Includes data preparation, parameter settings, learning process, summary of results. 模型学习的详细记录。包括数据准备、参数设置、学习过程、结果摘要。
Change History   更改历史记录	Records important changes to models, data, code. Documents reasons for updates and scope of impact. 记录对模型、数据、代码的重要更改。记录更新的原因和影响范围。
Risk Assessment Document 风险评估文件	A document evaluating potential risks, biases, ethical considerations of the model. Also used as evidence for regulatory compliance. 评估模型的潜在风险、偏见、道德考虑的文件。也用作法规遵从性的证据。
Quality Assurance Document 质量保证文件	Records test results, performance evaluations, validation procedures. Demonstrates compliance with quality standards. 记录测试结果、性能评估、验证程序。证明符合质量标准。
Compliance Document   合规文件	A document demonstrating compliance with regulatory requirements. Records responses to GDPR, AI governance, etc. 证明符合法规要求的文件。记录对 GDPR、AI 治理等的响应。
Architecture Diagram   架构图	Visual representation of system configuration, data flow, relationships between components. 系统配置、数据流、组件之间关系的可视化表示。
Troubleshooting Guide   故障排除指南	A guide listing common problems and their solution procedures. Contributes to improving operational efficiency. 列出常见问题及其解决过程的指南。有助于提高运营效率。
Model Lineage Diagram   模型沿袭图	Visual representation of model development process, derivative relationships, important changes. Clarifies relationships between versions. 模型开发过程、衍生关系、重要变化的可视化表示。阐明版本之间的关系。
Performance Benchmark   性能基准	Performance comparison results between different model versions. Used as evidence of improvement. 不同型号版本之间的性能比较结果。用作改善的证据。
Deployment Plan   部署计划	A plan describing model deployment strategy, schedule, risk countermeasures. 描述模型部署策略、计划、风险对策的计划。

Orchestration Management
编排管理

Term   术语	Description   描述
Pipeline Management   管道管理	Automation and control of ML workflows. Manages the series of flows from data processing to inference. ML 工作流的自动化和控制。管理从数据处理到推理的一系列流程。
Workflow Definition   工作流定义	Definition of each step in the ML process and its dependencies. DAG-based control flow design. ML 流程中每个步骤及其依赖关系的定义。基于 DAG 的控制流设计。
Automation Triggers   自动化触发器	Setting of conditions and schedules for pipeline execution. Control of event-driven processing. 设置管道执行的条件和计划。控制事件驱动的处理。
Error Handling   错误处理	Implementation of anomaly detection and recovery mechanisms. Definition of fallback strategies. 实施异常检测和恢复机制。定义回退策略。

Quality Management  质量管理

Term   术语	Description   描述
Quality Gates   质量门	Quality checkpoints before deployment. Verification of performance, security, compliance. 部署前的质量检查点。验证性能、安全性、合规性。
Test Automation   测试自动化	Automated execution system for unit tests, integration tests, performance tests. Continuous quality assurance. 用于单元测试、集成测试、性能测试的自动执行系统。持续的质量保证。
Quality Metrics   质量指标	Definition and monitoring of quality indicators for models and systems. Tracking of SLO/SLA compliance status. 定义和监控模型和系统的质量指标。跟踪 SLO/SLA 合规性状态。

Infrastructure Management
基础设施管理

Term   术语	Description   描述
Resource Optimization   资源优化	Efficient allocation and management of computational resources and storage. Cost optimization. 高效分配和管理计算资源和存储。成本优化。
Scaling Management   扩展管理	Setting and monitoring of auto-scaling policies. Flexible resource adjustment according to demand. 设置和监控 Auto Scaling 策略。根据需求灵活调整资源。
Availability Management   可用性管理	Ensuring system redundancy and fault tolerance. Management of backup and disaster recovery plans. 确保系统冗余和容错。备份和灾难恢复计划的管理。

Security Management  安全管理

[Amazon SageMaker components useful in this category]
[在此类别中有用的 Amazon SageMaker 组件]
* Amazon SageMaker Role Manager manages access permissions for ML activities, implementing security based on the principle of least privilege and providing appropriate access control.
* Amazon SageMaker Role Manager 管理 ML 活动的访问权限，根据最小权限原则实施安全性并提供适当的访问控制。

Term   术语	Description   描述
Access Control   存取控制	Role-based access management. Permission settings based on the principle of least privilege. 基于角色的访问管理。基于最小权限原则的权限设置。
Data Protection   数据保护	Encryption and anonymization of sensitive data. Implementation of privacy protection mechanisms. 敏感数据的加密和匿名化。实施隐私保护机制。
Vulnerability Management 漏洞管理	Detection and countermeasures for security vulnerabilities. Conducting regular security assessments. 安全漏洞的检测和对策。定期进行安全评估。

Governance Management  治理管理

Term   术语	Description   描述
Policy Management   策略管理	Formulation and compliance management of AI governance policies. Setting of ethical guidelines. AI 治理策略的制定和合规性管理。制定道德准则。
Audit Response   审计响应	Maintenance of audit trails and management of audit response processes. Preparation of compliance evidence. 维护审计跟踪和管理审计响应流程。准备合规证据。
Risk Management   风险管理	Identification, assessment, and implementation of mitigation measures for risks related to AI use. Continuous risk monitoring. 识别、评估和实施与 AI 使用相关的风险缓解措施。持续的风险监控。

References:   引用：
Tech Blog with curated related content
包含精选相关内容的技术博客

Summary  总结

In this article, I have compiled an "AI and Machine Learning Terminology for AWS" based on the knowledge I gained during my study process to pass the newly added AWS certifications: AWS Certified AI Practitioner and AWS Certified Machine Learning Engineer - Associate. Additionally, I included insights from co-authoring the "Quiz to Learn AWS Functions and History: Selected 'Machine Learning' Edition" in the "Compilation of Thin Books on AWS Vol.01", which was self-published for Japan's "Technical Book Festival 17".
在本文中，我根据我在学习过程中获得的知识编写了“AWS 的 AI 和机器学习术语”，以通过新添加的 AWS 认证：AWS Certified AI Practitioner 和 AWS Certified Machine Learning Engineer - Associate。此外，我还在“Compilation of Thin Books on AWS Vol.01”中加入了合著“Quiz to Learn AWS Functions and History： Selected 'Machine Learning' Edition”的见解，该书是为日本的“Technical Book Festival 17”自行出版的。

I will continue to shape ideas that can be useful for learning and utilizing AWS.
我将继续塑造对学习和使用 AWS 有用的想法。
Additionally, I plan to update this article periodically to reflect changes in AI and machine learning in AWS.
此外，我计划定期更新本文，以反映 AWS 中 AI 和机器学习的变化。