book

Privacy-Preserving Machine Learning

Name: Privacy-Preserving Machine Learning
Author: Srinivasa Rao Aravilli
ISBN: 9781800564671

by Srinivasa Rao Aravilli

May 2024

Intermediate to advanced

402 pages

10h 35m

English

Packt Publishing

Read now

Unlock full access

Privacy-Preserving Machine Learning
ForewordContributorsAbout the authorAbout the reviewers
Preface
Who this book is forWhat this book coversTo get the most out of this bookDownload the hands-on labs and example code filesConventions usedGet in touchShare Your ThoughtsDownload a free PDF copy of this book
Part 1: Introduction to Data Privacy and Machine Learning
Chapter 1: Introduction to Data Privacy, Privacy Breaches, and Threat Modeling
What do privacy and data privacy mean?Privacy regulationsPrivacy by Design and a case studyExample – Privacy by Design in a social media platformPrivacy breachesEquifax privacy breachClearview AI Privacy breachPrivacy threat modelingPrivacy threat modeling – definitionThe importance of privacy threat modelingPrivacy threat modeling’s alignment to Privacy by Design principlesSteps in privacy threat modelingPrivacy threat modeling frameworksThe LINDDUN frameworkStep 1 – modeling the systemStep 2 – eliciting and documenting threatsStep 3 – mitigating threatsThe need for privacy-preserving MLCase study – privacy-preserving ML in financial institutionsSummary
Chapter 2: Machine Learning Phases and Privacy Threats/Attacks in Each Phase
ML typesSupervised MLUnsupervised MLReinforced MLOverview of ML phasesThe main phases of MLPrivacy threats/attacks in ML phasesCollaborative roles in ML projectsPrivacy threats/attacks in MLMembership inference attackModel extraction attackReconstruction attacks—model inversion attacksModel inversion attacks in neural networksSummary
Part 2: Use Cases of Privacy-Preserving Machine Learning and a Deep Dive into Differential Privacy
Chapter 3: Overview of Privacy-Preserving Data Analysis and an Introduction to Differential Privacy
Privacy in data analysisThe need for privacy in data analysisPrivacy-preserving techniquesData anonymization and algorithms for data anonymizationData aggregationPrivacy-enhancing technologiesDifferential privacyFederated learningSecure multi-party computation (SMC)Homomorphic encryptionAnonymizationDe-identificationDifferential privacySummary
Chapter 4: Overview of Differential Privacy Algorithms and Applications of Differential Privacy
Differential privacy algorithmsLaplace distributionGaussian distributionComparison of noise-adding algorithms to apply differential privacyGenerating aggregates using differential privacySensitivityQueries that use differential privacyClippingOverview of real-life applications of differential privacyDifferential privacy usage at UberDifferential privacy usage at AppleDifferential privacy usage in the US CensusDifferential privacy at GoogleSummary
Chapter 5: Developing Applications with Differential Privacy Using Open Source Frameworks
Open source frameworks to implement differential privacyIntroduction to the PyDP framework and its key featuresExamples and demonstrations of PyDP in actionDeveloping a sample banking application with PyDP to showcase differential privacy techniquesProtecting against membership inference attacksApplying differential privacy to large datasetsUse case – generating differentially private aggregates on a large datasetPipelineDP high-level architectureTumult AnalyticsMachine learning using differential privacySynthetic Dataset Generation: Introducing Fraudulent TransactionsDevelop a classification model using scikit-learnHigh-level implementation of the SGD algorithmApplying differential privacy options using machine learningGenerating gradients using differential privacyClustering using differential privacyDeep learning using differential privacyFraud detection model using PyTorchFraud detection model with differential privacy using the Opacus frameworkDifferential privacy machine learning frameworksLimitations of differential privacy and strategies to overcome themSummary
Part 3: Hands-On Federated Learning

Chapter 6: Federated Learning and Implementing FL Using Open Source Frameworks
Federated learningPreserving privacyFL definitionCharacteristics of FLFL algorithmsFedSGDFedAvgFed Adaptative OptimizationThe steps involved in implementing FLOpen source frameworks to implement FLTensorFlow FederatedFlowerAn end-to-end use case of implementing fraud detection using FLDeveloping an FL model for fraud detection using the Flower frameworkFL with differential privacyApproach oneApproach twoA sample application using FL-DPSummary
Chapter 7: Federated Learning Benchmarks, Start-Ups, and the Next Opportunity
FL benchmarksThe importance of FL benchmarksFL datasetsFrameworks for FL benchmarksSelecting an FL framework for a projectA comparison of FedScale, FATE, Flower, and TensorFlow FederatedState-of-the-art research in FLCommunication-efficient FLPrivacy-preserving FLFederated Meta-LearningAdaptive FLFederated reinforcement learningKey company products related to FLSummary
Part 4: Homomorphic Encryption, SMC, Confidential Computing, and LLMs
Chapter 8: Homomorphic Encryption and Secure Multiparty Computation
Encryption, anonymization, and de-identificationData anonymizationDe-identificationExploring Homomorphic encryptionRing-basedLattice-basedElliptic curve-basedExploring the mathematics behind HEEncryptionHomomorphismTypes of HEFully Homomorphic Encryption (FHE)Somewhat Homomorphic Encryption (SHE)Partially Homomorphic Encryption (PHE)Paillier schemePyfhelSEAL PythonTenSEALpheImplementing HEImplementing PHEImplementing HE using the TenSEAL libraryComparison of HE frameworksPyfhelTenSEALPALISADEPySEALTFHEMachine learning with HEEncrypted evaluation of ML models and inferenceLimitations of HESecure Multiparty ComputationBasic principles of SMCApplications of SMCTechniques used for SMCImplementing SMC – high-level stepsPython frameworks that can be used to implement SMCImplementing Private Set Interaction (PSI) SMC – case studyZero-knowledge proofsBasic conceptsTypes of ZKPsApplications of ZKPsSummary
Chapter 9: Confidential Computing – What, Why, and the Current State
Privacy/security attacks on data in memoryData at restData in motionData in memoryConfidential computationWhat is confidential computing?Benefits of confidential computingTrusted execution environments – attestation of source code and how it helps protect against insider threat attacksIndustry standards for ML in TEEsConfidential Computing ConsortiumHigh-level comparison of Intel SGX, AWS Nitro Enclaves, Google Asylo, Azure enclaves, and AnjunaPros and cons of TEEsSummary
Chapter 10: Preserving Privacy in Large Language Models
Key concepts/terms used in LLMsPrompt example using ChatGPT (closed source LLM)Prompt example using open source LLMsComparison of open source LLMs and closed source LLMsAI standards and terminology of attacksNISTOWASP Top 10 for LLM applicationsPrivacy attacks on LLMsMembership inference attacks against generative modelsExtracting training data attack from generative modelsPrompt injection attacksPrivacy-preserving technologies for LLMsText attacks on ML models and LLMsPrivate transformers – training LLMs using differential privacySTOA – Privacy-preserving technologies for LLMsSummary
Index
Why subscribe?
Other Books You May EnjoyPackt is searching for authors like youShare Your ThoughtsDownload a free PDF copy of this book

Content preview from Privacy-Preserving Machine Learning

Part 4: Homomorphic Encryption, SMC, Confidential Computing, and LLMs

This part covers homomorphic encryption and SMC. It covers, in detail, privacy preservation in large language models (LLMs).

We cover the concepts of homomorphic encryption (HE) and SMC as privacy-enhancing techniques. We highlight the significance of these cryptographic approaches in enabling secure computations on encrypted data without compromising privacy.

SMC enables multiple parties to collaboratively compute results while keeping their individual input private. Our summary emphasizes the importance of HE and SMC in scenarios where sensitive data needs to be analyzed or processed in a privacy-preserving manner. By utilizing these techniques, organizations and individuals ...

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: 9781800564671

Cloud Computing

Data Engineering

Data Science

AI & ML

Programming Languages

Software Architecture

IT/Ops

Security

Design

Business

Soft Skills

Privacy-Preserving Machine Learning

by Srinivasa Rao Aravilli

Part 4: Homomorphic Encryption, SMC, Confidential Computing, and LLMs

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.