book

Building Machine Learning Powered Applications

Name: Building Machine Learning Powered Applications
Author: Emmanuel Ameisen
ISBN: 9781492045113

by Emmanuel Ameisen

January 2020

Beginner to intermediate

257 pages

6h 48m

English

O'Reilly Media, Inc.

Read now

Unlock full access

Preface
The Goal of Using Machine Learning Powered ApplicationsUse ML to Build Practical ApplicationsAdditional ResourcesPractical MLWhat This Book CoversPrerequisitesOur Case Study: ML–Assisted WritingThe ML ProcessConventions Used in This BookUsing Code ExamplesO’Reilly Online LearningHow to Contact UsAcknowledgments
I. Find the Correct ML Approach
1. From Product Goal to ML Framing
Estimate What Is PossibleModelsDataFraming the ML EditorTrying to Do It All with ML: An End-to-End FrameworkThe Simplest Approach: Being the AlgorithmMiddle Ground: Learning from Our ExperienceMonica Rogati: How to Choose and Prioritize ML ProjectsConclusion
2. Create a Plan
Measuring SuccessBusiness PerformanceModel PerformanceFreshness and Distribution ShiftSpeedEstimate Scope and ChallengesLeverage Domain ExpertiseStand on the Shoulders of GiantsML Editor PlanningInitial Plan for an EditorAlways Start with a Simple ModelTo Make Regular Progress: Start SimpleStart with a Simple PipelinePipeline for the ML EditorConclusion
II. Build a Working Pipeline
3. Build Your First End-to-End Pipeline
The Simplest ScaffoldingPrototype of an ML EditorParse and Clean DataTokenizing TextGenerating FeaturesTest Your WorkflowUser ExperienceModeling ResultsML Editor Prototype EvaluationModelUser ExperienceConclusion
4. Acquire an Initial Dataset
Iterate on DatasetsDo Data ScienceExplore Your First DatasetBe Efficient, Start SmallInsights Versus ProductsA Data Quality RubricLabel to Find Data TrendsSummary StatisticsExplore and Label EfficientlyBe the AlgorithmData TrendsLet Data Inform Features and ModelsBuild Features Out of PatternsML Editor FeaturesRobert Munro: How Do You Find, Label, and Leverage Data?Conclusion
III. Iterate on Models
5. Train and Evaluate Your Model
The Simplest Appropriate ModelSimple ModelsFrom Patterns to ModelsSplit Your DatasetML Editor Data SplitJudge PerformanceEvaluate Your Model: Look Beyond AccuracyContrast Data and PredictionsConfusion MatrixROC CurveCalibration CurveDimensionality Reduction for ErrorsThe Top-k MethodOther ModelsEvaluate Feature ImportanceDirectly from a ClassifierBlack-Box ExplainersConclusion
6. Debug Your ML Problems
Software Best PracticesML-Specific Best PracticesDebug Wiring: Visualizing and TestingStart with One ExampleTest Your ML CodeDebug Training: Make Your Model LearnTask DifficultyOptimization ProblemsDebug Generalization: Make Your Model UsefulData LeakageOverfittingConsider the Task at HandConclusion

7. Using Classifiers for Writing Recommendations
Extracting Recommendations from ModelsWhat Can We Achieve Without a Model?Extracting Global Feature ImportanceUsing a Model’s ScoreExtracting Local Feature ImportanceComparing ModelsVersion 1: The Report CardVersion 2: More Powerful, More UnclearVersion 3: Understandable RecommendationsGenerating Editing RecommendationsConclusion
IV. Deploy and Monitor
8. Considerations When Deploying Models
Data ConcernsData OwnershipData BiasSystemic BiasModeling ConcernsFeedback LoopsInclusive Model PerformanceConsidering ContextAdversariesAbuse Concerns and Dual-UseChris Harland: Shipping ExperimentsConclusion
9. Choose Your Deployment Option
Server-Side DeploymentStreaming Application or APIBatch PredictionsClient-Side DeploymentOn DeviceBrowser SideFederated Learning: A Hybrid ApproachConclusion
10. Build Safeguards for Models
Engineer Around FailuresInput and Output ChecksModel Failure FallbacksEngineer for PerformanceScale to Multiple UsersModel and Data Life Cycle ManagementData Processing and DAGsAsk for FeedbackChris Moody: Empowering Data Scientists to Deploy ModelsConclusion
11. Monitor and Update Models
Monitoring Saves LivesMonitoring to Inform Refresh RateMonitor to Detect AbuseChoose What to MonitorPerformance MetricsBusiness MetricsCI/CD for MLA/B Testing and ExperimentationOther ApproachesConclusion
Index

Content preview from Building Machine Learning Powered Applications

About the Author

Emmanuel Ameisen has been building ML-powered products for years. He is currently working on ML engineering at Stripe. Previously, he was head of AI at Insight Data Science, where he led more than 150 ML projects. Before, Emmanuel was a data scientist at Zipcar, where he worked on on-demand prediction and building frameworks and services to help deploy ML models in production. He has a background at the intersection of ML and business, with an M.S. in AI from Université Paris-Sud, an M.S. in Engineering from CentraleSupélec, and an M.S. in Management from ESCP Europe.

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.

Read now

Unlock full access

More than 5,000 organizations count on O’Reilly

O’Reilly covers everything we've got, with content to help us build a world-class technology community, upgrade the capabilities and competencies of our teams, and improve overall team performance as well as their engagement.

Julian F.

Head of Cybersecurity

I wanted to learn C and C++, but it didn't click for me until I picked up an O'Reilly book. When I went on the O’Reilly platform, I was astonished to find all the books there, plus live events and sandboxes so you could play around with the technology.

Addison B.

Field Engineer

I’ve been on the O’Reilly platform for more than eight years. I use a couple of learning platforms, but I'm on O'Reilly more than anybody else. When you're there, you start learning. I'm never disappointed.

Amir M.

Data Platform Tech Lead

I'm always learning. So when I got on to O'Reilly, I was like a kid in a candy store. There are playlists. There are answers. There's on-demand training. It's worth its weight in gold, in terms of what it allows me to do.

Mark W.

Embedded Software Engineer

Machine Learning with PyTorch and Scikit-Learn

Publisher Resources

ISBN: 9781492045106Errata Page

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Building Machine Learning Powered Applications

by Emmanuel Ameisen

About the Author

Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.