Investigating the interpretability of biometric face templates using gated sparse autoencoders and differentiable image parametrizations

Rot, Peter (author), Grm, Klemen (author)

State-of-the-art face recognition models rely on deep, complex neural net architectures that produce relatively compact template vectors, making their mechanisms of operation difficult to interpret and understand. Recently, mechanistic interpretability has emerged as a promising approach to explain large language models. In this paper, we aim to apply such approaches to explain face recognition models. Our method involves transforming face image templates into sparse representations and analyzing their components by identifying images that maximize activation. Our results demonstrate that existing mechanistic interpretability techniques generalize well to previously unconsidered tasks and architectures, and that differentiable image parametrizations can serve as a useful additional means of confirming the interpretation of sparse representations.

• Language: English
• Keywords: biometry, face recognition, sparse auto-encoders
• Typology: 1.08 - Published Scientific Conference Contribution
• Organization: FE - Faculty of Electrical Engineering
• Publication status: Published
• Publication version: Version of Record
• Year: 2024
• Number of pages: 5 str.
• PID: 20.500.12556/RUL-160088
• UDC: 004.93'1
• COBISS.SI-ID: 204387843
• Publication date in RUL: 19.08.2024

Record is a part of a monograph

• Title: ICML 2024 Workshop on Mechanistic Interpretability : ICML 2024 MI Workshop
• Place of publishing: [Massachusetts
• Publisher: OpenReview
• Year: 2024
• COBISS.SI-ID: 204384771

Licences

• License: CC BY 4.0, Creative Commons Attribution 4.0 International
• Link: http://creativecommons.org/licenses/by/4.0/
• Description: This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.

Secondary language

• Language: Slovenian
• Keywords: biometrija, razpoznavanje obrazov, redki samo-kodirniki

Projects

• Funder: ARIS - Slovenian Research and Innovation Agency
• Project number: P2-0250
• Name: Metrologija in biometrični sistemi

• Funder: ARIS - Slovenian Research and Innovation Agency
• Project number: J2-50069
• Name: Interpretacija mehanizmov za razložljivo biometrično umetno inteligenco (MIXBAI)