paper-with-me

Papers

Shielded Representations: Protecting Sensitive Attributes Through Iterative Gradient-Based Projection

2023-05-17 · Shadi Iskander, Kira Radinsky, Yonatan Belinkov

Natural language processing models tend to learn and encode social biases present in the data. One popular approach for addressing such biases is to eliminate encoded information from the model's representations. However, current methods are restricted to removing only linearly encoded information. In this work, we propose Iterative Gradient-Based Projection (IGBP), a novel method for removing non-linear encoded concepts from neural representations. Our method consists of iteratively training neural classifiers to predict a particular attribute we seek to eliminate, followed by a projection of the representation on a hypersurface, such that the classifiers become oblivious to the target attribute. We evaluate the effectiveness of our method on the task of removing gender and race information as sensitive attributes. Our results demonstrate that IGBP is effective in mitigating bias through intrinsic and extrinsic evaluations, with minimal impact on downstream task accuracy.

📄 PDF Abstract BibTeX arXiv:2305.10204

Code (1)

technion-cs-nlp/igbp_nonlinear-removal 공식 구현 pytorch

Tasks

Attribute

Similar Papers 제목 키워드 기반

Protecting gender and identity with disentangled speech representations

2021-04-22 · Dimitrios Stoidis, Andrea Cavallaro

Besides its linguistic content, our speech is rich in biometric information that can be inferred by classifiers. Learning privacy-preserving representations for speech signals enables downstream tasks without sharing unn…

Privacy PreservingRepresentation LearningSpeaker VerificationSpeech Recognition

Declarative Privacy-Preserving Inference Queries

2024-01-22 · Hong Guan, Ansh Tiwari, Summer Gautier, Rajan Hari Ambrish 외

Detecting inference queries running over personal attributes and protecting such queries from leaking individual information requires tremendous effort from practitioners. To tackle this problem, we propose an end-to-end…

Federated LearningManagementPrivacy Preserving

PaCo-VLA: Passivity-Shielded Compliance Prior for Contact-Rich Vision-Language-Action Manipulation

2026-05-30 · Haofan Cao, Zhaoyang Li, Zhichao You, Liang Guo 외 arxiv

Contact-rich manipulation demands both high-level semantic reasoning and the safe regulation of high-frequency contact dynamics. While Vision-Language-Action (VLA) models provide unprecedented semantic generalization, th…

Information Obfuscation of Graph Neural Networks

2020-09-28 · Peiyuan Liao, Han Zhao, Keyulu Xu, Tommi Jaakkola 외

While the advent of Graph Neural Networks (GNNs) has greatly improved node and graph representation learning in many applications, the neighborhood aggregation scheme exposes additional vulnerabilities to adversaries see…

Adversarial DefenseGraph Representation LearningKnowledge GraphsRecommendation Systems+1

Debiasing Diffusion Model: Enhancing Fairness through Latent Representation Learning in Stable Diffusion Model

2025-03-16 · Lin-Chun Huang, Ching Chieh Tsao, Fang-Yi Su, Jung-Hsien Chiang

Image generative models, particularly diffusion-based models, have surged in popularity due to their remarkable ability to synthesize highly realistic images. However, since these models are data-driven, they inherit bia…

FairnessmodelRepresentation Learning