欢迎光临澳大利亚新华书店网 [登录 | 免费注册]

    • 机器学习数据训练(影印版)(英文版)
      • 作者:(美)安东尼·萨尔基斯|责编:张烨
      • 出版社:东南大学
      • ISBN:9787576612028
      • 出版日期:2024/03/01
      • 页数:306
    • 售价:47.2
  • 内容大纲

        训练数据与算法本身一样关系到数据项目的成败,因为大多数AI系统的失败都与训练数据有关。尽管训练数据是AI和机器学习成功的基础,但却很少有全面的资源能帮助你掌握这一过程。
        在这本实践指南中,作者Anthony Sarkis(Diffgram AI数据训练软件的首席工程师)向技术专业人员、管理人员、主题专家展示了如何使用和扩展训练数据,同时阐明了监督机器的人性化一面。工程领导者、数据工程师、数据科学专业人士都将深入了解使用训练数据取得成功所需的概念、工具和流程。
        通过本书,你将学习如何:
        有效地使用包括模式、原始数据、注释在内的训练数据;
        改造你的工作、团队或组织,使其更加以AI,ML数据为中心;
        向其他员工、团队成员、利益相关者清晰地解释训练数据概念;
        为生产级AI应用设计、部署、交付训练数据;
        识别并纠正新的基于训练数据的故障模式,如数据偏差;
        自信地使用自动化技术来更有效地创建训练数据;
        成功维护、操作、改进训练数据记录系统。
  • 作者介绍

        安东尼·萨尔基斯(Anthony Sarkis)是Diffgram AI数据训练软件的首席工程师,也是Diffgram公司的首席技术官和创始人。在此之前,他是Skidmore.Owings&Merrill公司的研发软件工程师,并与他人共同创办了DriveCarma.ca。
  • 目录

    Preface
    1. Training Data Introduction
        Training Data Intents
          What Can You Do With Training Data?
          What Is Training Data Most Concerned With?
        Training Data Opportunities
          Business Transformation
          Training Data Efficiency
          Tooling Proficiency
          Process Improvement Opportunities
        Why Training Data Matters
          ML Applications Are Becoming Mainstream
          The Foundation of Successful AI
          Training Data Is Here to Stay
          Training Data Controls the ML Program
          New Types of Users
        Training Data in the Wild
          What Makes Training Data Difficult?
          The Art of Supervising Machines
          A New Thing for Data Science
          ML Program Ecosystem
           Data-Centric Machine Learning
           Failures
           History of Development Affects Training Data Too
           What Training Data Is Not
        Generative AI
         Human Alignment Is Human Supervision
       Summary
    2. Getting Up and Running
       Introduction
       Getting Up and Running
         Installation
         Tasks Setup
         Annotator Setup
         Data Setup
         Workflow Setup
         Data Catalog Setup
         Initial Usage
         Optimization
       Tools Overview
         Training Data for Machine Learning
         Growing Selection of Tools
         People, Process, and Data
         Embedded Supervision
         Human Computer Supervision
         Separation of End Concerns
         Standards
         Many Personas
         A Paradigm to Deliver Machine Learning Software
       Trade-Offs

         Costs
         Installed Versus Software as a Service
         Development System
         Scale
         Installation Options
         Annotation Interfaces
         Modeling Integration
         Multi-User versus Single-User Systems
         Integrations
         Scope
         Hidden Assumptions
         Security
         Open Source and Closed Source
       History
         Open Source Standards
    ……
    3.Schema
    4.Data Engineering
    5.Workflow
    6.Theories, Concepts, and Maintenance
    7.AI Transformation and Use Cases
    8.Automation
    9.Case Studies and Stories