book

Platform and Model Design for Responsible AI

Name: Platform and Model Design for Responsible AI
ISBN: 9781803237077

by Dr. Amita Kapoor, Sharmistha Chatterjee

April 2023

Intermediate to advanced

516 pages

14h 59m

English

Packt Publishing

Read now

Unlock full access

Platform and Model Design for Responsible AI
ContributorsAbout the authorsAbout the reviewers
Preface
Who this book is forWhat this book coversTo get the most out of this bookDownload the example code filesConventions usedGet in touchShare Your ThoughtsDownload a free PDF copy of this book
Part 1: Risk Assessment Machine Learning Frameworks in a Global Landscape
Chapter 1: Risks and Attacks on ML Models
Technical requirementsDiscovering risk elementsStrategy riskFinancial riskTechnical riskPeople and processes riskTrust and explainability riskCompliance and regulatory riskExploring risk mitigation strategies with vision, strategy, planning, and metricsDefining a structured risk identification processEnterprise-wide controlsMicro-risk management and the reinforcement of controlsAssessing potential impact and loss due to attacksDiscovering different types of attacksData phishing privacy attacksPoisoning attacksEvasion attacksModel stealing/extractionPerturbation attacksScaffolding attackModel inversionTransfer learning attacksSummaryFurther reading
Chapter 2: The Emergence of Risk-Averse Methodologies and Frameworks
Technical requirementsAnalyzing the threat matrix and defense techniquesResearching and planning during the system and model design/architecture phaseModel training and developmentML model live in productionAnonymization and data encryptionData maskingData swappingData perturbationData generalizationK-anonymityL-diversityT-closenessPseudonymizationHomomorphic encryptionSecure Multi-Party Computation (MPC/SMPC)Differential Privacy (DP)SensitivityProperties of DPHybrid privacy methods and modelsAdversarial risk mitigation frameworksModel robustnessSummaryFurther reading
Chapter 3: Regulations and Policies Surrounding Trustworthy AI
Regulations and enforcements under different authoritiesRegulations in the European UnionPropositions/acts passed by other countriesSpecial regulations for children and minority groupsPromoting equality for minority groupsEducational initiativesInternational AI initiatives and cooperative actionsNext steps for trustworthy AIProposed solutions and improvement areasSummaryFurther reading
Part 2: Building Blocks and Patterns for a Next-Generation AI Ecosystem
Chapter 4: Privacy Management in Big Data and Model Design Pipelines
Technical requirementsDesigning privacy-proven pipelinesBig data pipelinesArchitecting model design pipelinesIncremental/continual ML training and retrainingScaling defense pipelinesEnabling differential privacy in scalable architecturesDesigning secure microservicesVaultCloud security architectureDeveloping in a sandbox environmentManaging secrets in cloud orchestration servicesMonitoring and threat detectionSummaryFurther reading
Chapter 5: ML Pipeline, Model Evaluation, and Handling Uncertainty
Technical requirementsUnderstanding different components of ML pipelinesML tasks and algorithmsUncertainty in MLTypes of uncertaintyQuantifying uncertaintyUncertainty in regression tasksUncertainty in classification tasksTools for benchmarking and quantifying uncertaintyThe Uncertainty Baselines libraryKeras-UncertaintyRobustness metricsSummaryReferences
Chapter 6: Hyperparameter Tuning, MLOps, and AutoML
Technical requirementsIntroduction to AutoMLIntroducing H2O AutoMLUnderstanding Amazon SageMaker AutopilotThe need for MLOpsTFX – a scalable end-to-end platform for AI/ML workflowsUnderstanding KubeflowKatib for hyperparameter tuningVertex AIDatasetsTraining and experiments in Vertex AIVertex AI WorkbenchSummaryFurther reading

Part 3: Design Patterns for Model Optimization and Life Cycle Management
Chapter 7: Fairness Notions and Fair Data Generation
Technical requirementsUnderstanding the impact of data on fairnessReal-world bias examplesCauses of biasDefining fairnessTypes of fairness based on statistical metricsTypes of fairness based on the metrics of predicted outcomesTypes of fairness based on similarity-based measuresTypes of fairness based on causal reasoningThe role of data audits and quality checks in fairnessAssessing fairnessLinear regressionThe variance inflation factorMutual informationSignificance testsEvaluating group fairnessEvaluating counterfactual fairnessBest practicesFair synthetic datasetsMOSTLY AI’s self-supervised fair synthetic data generatorA GAN-based fair synthetic data generatorSummaryFurther reading
Chapter 8: Fairness in Model Optimization
Technical requirementsThe notion of fairness in MLUnfairness mitigation methodsIn-processing methodsExplicit unfairness mitigationFairness constraints for a classification taskFairness constraints for a regression taskFairness constraints for a clustering taskFairness constraints for a reinforcement learning taskFairness constraints for recommendation systemsChallenges of fairnessMissing sensitive attributesMultiple sensitive attributesChoice of fairness measurementsIndividual versus group fairness trade-offInterpretation and fairnessFairness versus model performanceLimited datasetsSummaryFurther reading
Chapter 9: Model Explainability
Technical requirementsIntroduction to Explainable AIScope of XAIChallenges in XAIExplain Like I’m Five (ELI5)LIMESHAPUnderstanding churn modeling using XAI techniquesBuilding a modelUsing ELI5 to understand classifier modelsHands-on with LIMESHAP in actionCausalNexDoWhy for causal inferenceDoWhy in actionAI Explainability 360 for interpreting modelsSummaryReferences
Chapter 10: Ethics and Model Governance
Technical requirementsModel Risk Management (MRM)Types of model inventory managementCost savings with MRMA transformative journey with MRMModel risk tieringModel risk calibrationModel version controlModelDBWeights & BiasesFurther reading
Part 4: Implementing an Organization Strategy, Best Practices, and Use Cases
Chapter 11: The Ethics of Model Adaptability
Technical requirementsAdaptability framework for data and model driftStatistical methodsStatistical process controlUnderstanding model explainability during concept drift/calibrationExplainability and calibrationChallenges with calibration and fairnessSummaryFurther reading
Chapter 12: Building Sustainable Enterprise-Grade AI Platforms
Technical requirementsThe key to sustainable enterprise-grade AI platformsSustainable solutions with AI as an organizational roadmapOrganizational standards for sustainable frameworksSustainability practices and metrics across different cloud platformsEmission metrics on Google CloudBest practices and strategies for carbon-free energyThe energy efficiency of data centersCarbon emission trackersThe FL carbon calculatorCentralized learning carbon emissions calculatorAdopting sustainable model training and deployment with FLCO2e emission metricsComparing emission factors – centralized learning versus FLIllustrating how FL works better than centralized learningThe CO2 footprint of FLHow to compensate for equivalent CO2e emissionsDesign patterns of FL-based model trainingSustainability in model deploymentsDesign patterns of FL-based model deploymentsSummaryFurther reading
Chapter 13: Sustainable Model Life Cycle Management, Feature Stores, and Model Calibration
Sustainable model development practicesOrganizational standards for sustainable, trustworthy frameworksExplainability, privacy, and sustainability in feature storesFeature store components and functionalitiesFeature stores for FLExploring model calibrationDetermining whether a model is well calibratedCalibration techniquesModel calibration using scikit-learnBuilding sustainable, adaptable systemsConcept drift-aware federated averaging (CDA-FedAvg)SummaryFurther reading
Chapter 14: Industry-Wide Use Cases
Technical requirementsBuilding ethical AI solutions across industriesBiased chatbotsEthics in XR/AR/VRUse cases in retailPrivacy in the retail industryFairness in the retail industryInterpretability – the role of counterfactuals (CFs)Supply chain use casesUse cases in BFSIDeepfakesUse cases in healthcareHealthcare system architecture using Google CloudSurvival analysis for Responsible AI healthcare applicationsSummaryFurther reading
Index
Why subscribe?
Other Books You May EnjoyPackt is searching for authors like youShare Your ThoughtsDownload a free PDF copy of this book

Overview

Platform and Model Design for Responsible AI is a comprehensive guide to creating trustworthy machine learning systems. This book will teach you how to design algorithms that are ethical, transparent, and resilient, addressing the crucial challenges of AI ethics and governance.

What this Book will help me do

Identify and quantify risks associated with machine learning models and frameworks.
Design and implement ethical AI systems that comply with privacy and fairness standards.
Develop explainable ML models and integrate them into enterprise-grade ecosystems.
Apply ML orchestration tools effectively across various cloud platforms.
Optimize model performance while minimizing data bias and ensuring compliance with regulations.

Author(s)

Amita Kapoor and Sharmistha Chatterjee are seasoned experts in AI and machine learning, with a focus on ethical and sustainable AI practices. Amita Kapoor has authored several influential books in the field and brings a wealth of practical experience. Sharmistha Chatterjee is a recognized consultant in responsible AI platform design. Together, they provide a thoughtful and detailed approach to mastering responsible AI.

Who is it for?

This book is ideal for intermediate to advanced machine learning professionals, data scientists, and AI system architects. It specifically addresses the needs of those interested in understanding and mitigating AI risks. Readers should be motivated to integrate ethical and regulatory considerations into their work. If you're keen on building resilient and accountable AI systems, this book is for you.

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

Publisher Resources

ISBN: 9781803237077

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills