Saved in:
Bibliographic Details
Main Authors: de Freitas, João Machado, Geiger, Bernhard C.
Format: Preprint
Published: 2022
Subjects:
Online Access:https://arxiv.org/abs/2211.01446
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914137932562432
author de Freitas, João Machado
Geiger, Bernhard C.
author_facet de Freitas, João Machado
Geiger, Bernhard C.
contents Ensuring trustworthiness in machine learning -- by balancing utility, fairness, and privacy -- remains a critical challenge, particularly in representation learning. In this work, we investigate a family of closely related information-theoretic objectives, including information funnels and bottlenecks, designed to extract invariant representations from data. We introduce the Conditional Privacy Funnel with Side-information (CPFSI), a novel formulation within this family, applicable in both fully and semi-supervised settings. Given the intractability of these objectives, we derive neural-network-based approximations via amortized variational inference. We systematically analyze the trade-offs between utility, invariance, and representation fidelity, offering new insights into the Pareto frontiers of these methods. Our results demonstrate that CPFSI effectively balances these competing objectives and frequently outperforms existing approaches. Furthermore, we show that by intervening on sensitive attributes in CPFSI's predictive posterior enhances fairness while maintaining predictive performance. Finally, we focus on the real-world applicability of these approaches, particularly for learning robust and fair representations from tabular datasets in data scarce-environments -- a modality where these methods are often especially relevant.
format Preprint
id arxiv_https___arxiv_org_abs_2211_01446
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Trustworthy Representation Learning via Information Funnels and Bottlenecks
de Freitas, João Machado
Geiger, Bernhard C.
Machine Learning
Ensuring trustworthiness in machine learning -- by balancing utility, fairness, and privacy -- remains a critical challenge, particularly in representation learning. In this work, we investigate a family of closely related information-theoretic objectives, including information funnels and bottlenecks, designed to extract invariant representations from data. We introduce the Conditional Privacy Funnel with Side-information (CPFSI), a novel formulation within this family, applicable in both fully and semi-supervised settings. Given the intractability of these objectives, we derive neural-network-based approximations via amortized variational inference. We systematically analyze the trade-offs between utility, invariance, and representation fidelity, offering new insights into the Pareto frontiers of these methods. Our results demonstrate that CPFSI effectively balances these competing objectives and frequently outperforms existing approaches. Furthermore, we show that by intervening on sensitive attributes in CPFSI's predictive posterior enhances fairness while maintaining predictive performance. Finally, we focus on the real-world applicability of these approaches, particularly for learning robust and fair representations from tabular datasets in data scarce-environments -- a modality where these methods are often especially relevant.
title Trustworthy Representation Learning via Information Funnels and Bottlenecks
topic Machine Learning
url https://arxiv.org/abs/2211.01446