欢迎光临澳大利亚新华书店网 [登录 | 免费注册]

    • 可靠的机器学习(影印版)(英文版)
      • 作者:(美国)凯茜·陈//(爱尔兰)尼尔·理查德·墨菲//(美国)克拉蒂·帕里萨//D.斯卡利//托德·安德伍德|责编:张烨
      • 出版社:东南大学
      • ISBN:9787576605525
      • 出版日期:2023/03/01
      • 页数:376
    • 售价:47.6
  • 内容大纲

        无论你是小型创业公司还是跨国公司的一员,这本实践用书都为你(数据科学家、软件和网站可靠性工程师、产品经理或企业主)展示了如何在组织内可靠、有效和负责地运行和建立机器学习。你将深入了解其中涉及的方方面面,从如何在生产中进行模型监控到如何在产品组织中运营一个完善的模型开发团队。
        通过将SRE思维应用于机器学习,作为本书作者和工程专业人士的Cathy Chen、Kranti Parisa、Niall Richard Murphy、D.Sculley、Todd Underwood以及特邀作者向你展示了如何运行高效可靠的机器学习系统。无论你是想增加收入、优化决策、解决问题,还是想理解和影响客户行为,你都将学到如何执行日常的机器学习任务,同时保持更广阔的视野。
        本书内容包括:
        ·什么是ML:运作方式以及依赖什么
        ·用于理解机器学习“环路”如何工作的概念框架
        ·有效的生产如何使机器学习系统易于监控、部署和操作
        ·为什么机器学习系统使生产故障排除更加困难,以及如何进行相应的补偿
        ·机器学习、产品和生产团队如何有效沟通
  • 作者介绍

  • 目录

    Foreword
    Preface
    1. Introduction
      The ML Lifecycle
        Data Collection and Analysis
        ML Training Pipelines
        Build and Validate Applications
        Quality and Performance Evaluation
        Defining and Measuring SLOs
        Launch
        Monitoring and Feedback Loops
       Lessons from the Loop
    2. Data Management Principles
      Data as Liability
      The Data Sensitivity of ML Pipelines
      Phases of Data
        Creation
        Ingestion
        Processing
        Storage
        Management
        Analysis and Visualization
      Data Reliability
        Durability
        Consistency
        Version Control
        Performance
        Availability
      Data Integrity
        Security
        Privacy
        Policy and Compliance
      Conclusion
    3. Basic Introduction to Models
      What Is a Model?
      A Basic Model Creation Work_flow
      Model Architecture Versus Model Definition Versus Trained Model
      Where Are the Vulnerabilities?
        Training Data
        Labels
        Training Methods
      Infrastructure and Pipelines
        Platforms
        Feature Generation
        Upgrades and Fixes
      A Set of Useful Questions to Ask About Any Model
      An Example ML System
        Yarn Product Click-Prediction Model
        Features
        Labels for Features

        Model Updating
        Model Serving
        Common Failures
      Conclusion
    4. Feature and Training Data
      Features
        Feature Selection and Engineering
        Lifecycle of a Feature
        Feature Systems
      Labels
      Human-Generated Labels
        Annotation Workforces
        Measuring Human Annotation Quality
      ……
    5. Evaluating Model Validity and Quality
    6. Fairness, Privacy, and Ethical ML Systems
    7. Training Systems
    8. Serving
    9. Monitoring and Observability for Models
    10. Continuous ML
    11. Incident Response
    12. How Product and ML Interact
    13. Integrating ML into Your Organization
    14. Practical ML Org Implementation Examples
    15. Case Studies: MLOps in Practice
    Index