kth.sePublications KTH
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Upper Bounds on the Generalization Error of Private Algorithms for Discrete Data
KTH, School of Electrical Engineering and Computer Science (EECS), Intelligent systems, Information Science and Engineering.ORCID iD: 0000-0002-0862-1333
KTH, School of Electrical Engineering and Computer Science (EECS), Intelligent systems, Information Science and Engineering. Ericsson Res, S-16483 Stockholm, Sweden..ORCID iD: 0000-0002-8974-6591
KTH, School of Electrical Engineering and Computer Science (EECS), Intelligent systems, Information Science and Engineering.ORCID iD: 0000-0002-7926-5081
2021 (English)In: IEEE Transactions on Information Theory, ISSN 0018-9448, E-ISSN 1557-9654, Vol. 67, no 11, p. 7362-7379Article in journal (Refereed) Published
Abstract [en]

In this work, we study the generalization capability of algorithms from an information-theoretic perspective. It has been shown that the expected generalization error of an algorithm is bounded from above by a function of the relative entropy between the conditional probability distribution of the algorithm's output hypothesis, given the dataset with which it was trained, and its marginal probability distribution. We build upon this fact and introduce a mathematical formulation to obtain upper bounds on this relative entropy. Assuming that the data is discrete, we then develop a strategy using this formulation, based on the method of types and typicality, to find explicit upper bounds on the generalization error of stable algorithms, i.e., algorithms that produce similar output hypotheses given similar input datasets. In particular, we show the bounds obtained with this strategy for the case of epsilon-DP and mu-GDP algorithms.

Place, publisher, year, edition, pages
Institute of Electrical and Electronics Engineers (IEEE) , 2021. Vol. 67, no 11, p. 7362-7379
Keywords [en]
Generalization, differential privacy, mutual information, mixture approximation
National Category
Telecommunications Control Engineering Signal Processing
Identifiers
URN: urn:nbn:se:kth:diva-304555DOI: 10.1109/TIT.2021.3111480ISI: 000709078200028Scopus ID: 2-s2.0-85114716254OAI: oai:DiVA.org:kth-304555DiVA, id: diva2:1609475
Note

QC 20211108

Available from: 2021-11-08 Created: 2021-11-08 Last updated: 2024-04-02Bibliographically approved
In thesis
1. An Information-Theoretic Approach to Generalization Theory
Open this publication in new window or tab >>An Information-Theoretic Approach to Generalization Theory
2024 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

In this thesis, we investigate the in-distribution generalization of machine learning algorithms, focusing on establishing rigorous upper bounds on the generalization error. We depart from traditional complexity-based approaches by introducing and analyzing information-theoretic bounds that quantify the dependence between a learning algorithm and the training data.

We consider two categories of generalization guarantees:

  • Guarantees in expectation. These bounds measure performance in the average case. Here, the dependence between the algorithm and the data is often captured by the mutual information or other information measures based on ƒ-divergences. While these measures offer an intuitive interpretation, they might overlook the geometry of the algorithm's hypothesis class. To address this limitation, we introduce bounds using the Wasserstein distance, which incorporates geometric considerations at the cost of being mathematically more involved. Furthermore, we propose a structured, systematic method to derive bounds capturing the dependence between the algorithm and an individual datum, and between the algorithm and subsets of the training data, conditioned on knowing the rest of the data. These types of bounds provide deeper insights, as we demonstrate by applying them to derive generalization error bounds for the stochastic gradient Langevin dynamics algorithm.      
  • PAC-Bayesian guarantees. These bounds measure the performance level with high probability. Here, the dependence between the algorithm and the data is often measured by the relative entropy. We establish connections between the Seeger--Langford and Catoni's bounds, revealing that that the former is optimized by the Gibbs posterior. Additionally, we introduce novel, tighter bounds for various types of loss functions, including those with a bounded range, cumulant generating function, moment, or variance. To achieve this, we introduce a new technique to optimize parameters in probabilistic statements.

We also study the limitations of these approaches. We present a counter-example where most of the existing (relative entropy-based) information-theoretic bounds fail, and where traditional approaches do not. Finally, we explore the relationship between privacy and generalization. We show that algorithms with a bounded maximal leakage generalize. Moreover, for discrete data, we derive new bounds for differentially private algorithms that vanish as the number of samples increases, thus guaranteeing their generalization even with a constant privacy parameter. This is in contrast with previous bounds in the literature, that require the privacy parameter to decrease with the number of samples to ensure generalization.

Abstract [sv]

I denna avhandling undersöker vi maskininlärningsalgoritmers generaliseringsförmåga för likafördelad data, med fokus på att upprätta stränga övre gränser för generaliseringsfelet. Vi avviker från traditionella komplexitetsbaserade metoder genom att introducera och analysera informationsteoretiska gränser som kvantifierar beroendet mellan en inlärningsalgoritm och träningsdata.

Vi studerar två kategorier av generaliseringsgarantier:

  • Väntevärdegarantier. Dessa gränser mäter prestanda i genomsnitt. I detta fall fångas beroendet mellan algoritmen och data ofta av ömsesidig information eller andra informationsmått baserade på ƒ-divergenser. Trots att dessa mått erbjuder en intuitiv tolkning, kan de förbise geometrin hos algoritmens hypotesklass. För att hantera denna begränsning introducerar vi gränser som använder Wassersteinavstånd, vilket innehåller geometriska överväganden på bekostnad av att vara mer matematiskt invecklat. Dessutom föreslår vi en strukturerad, systematisk metod för att härleda gränser som fångar beroendet mellan algoritmen och en enskild datapunkt, samt mellan algoritmen och delmängder av träningsdata, under förutsättning att resten av datan är känd. Dessa typer av gränser ger djupare insikter, vilket vi demonstrerar genom att tillämpa dem för att härleda generaliseringsgarantier för algoritmen stokastisk gradient Langevindynamik.   
  • PAC-Bayesiska garantier. Dessa gränser mäter prestandanivån med hög sannolikhet. I detta fall fångas beroendet mellan algoritmen och data ofta av relativ entropi. Vi härleder samband mellan Seeger--Langfords- och Catonis gränser, vilket avslöjar att den förra optimeras av Gibbsfördelningen. Dessutom introducerar vi nya, starkare gränser för olika typer av kostnadsfunktioner, inklusive de med begränsad värdemängd, momentgenererande funktion, moment, eller varians. För att uppnå detta introducerar vi en ny teknik för att optimera parametrar i sannolikhetsuttryck.

Vi studerar också begränsningarna för dessa metoder. Vi presenterar ett motexempel där de flesta av de existerande (relativ entropibaserade) informationsteoretiska gränserna misslyckas, och där traditionella metoder fungerar. Slutligen utforskar vi sambandet mellan dataintegritet och generaliseringsförmåga. Vi visar att algoritmer med ett begränsad maximalt läckage generaliserar. Dessutom härleder vi för diskreta data nya gränser för algoritmer med differentiellt dataintegritet, som försvinner när antal sampel ökar, vilket garanterar deras generaliseringsförmåga även med en konstant dataintegritetsparameter. Detta står i kontrast till tidigare gränser i litteraturen, som kräver att dataintegritetsparametern minskar med antalet sampel för att säkerställa generaliseringsförmåga.

Place, publisher, year, edition, pages
KTH Royal Institute of Technology, 2024. p. 259
Series
TRITA-EECS-AVL ; 2024:31
Keywords
Generalization, Information-Theoretic Bounds
National Category
Computer Sciences
Research subject
Electrical Engineering
Identifiers
urn:nbn:se:kth:diva-344833 (URN)978-91-8040-878-3 (ISBN)
Public defence
2024-04-23, F3, Lindstedtsvägen 26, Stockholm, 13:00 (English)
Opponent
Supervisors
Funder
Swedish Research Council, 2019-03606
Note

QC 20240402

Available from: 2024-04-02 Created: 2024-04-02 Last updated: 2024-04-15Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textScopus

Authority records

Rodríguez Gálvez, BorjaBassi, GermánSkoglund, Mikael

Search in DiVA

By author/editor
Rodríguez Gálvez, BorjaBassi, GermánSkoglund, Mikael
By organisation
Information Science and Engineering
In the same journal
IEEE Transactions on Information Theory
TelecommunicationsControl EngineeringSignal Processing

Search outside of DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 193 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf