Small Perturbation

Small perturbation analysis investigates the impact of minor changes to inputs or model parameters on the behavior of machine learning models, particularly deep neural networks. Current research focuses on leveraging perturbation for tasks such as improving model robustness (e.g., against adversarial attacks or noisy data), enhancing interpretability, and achieving efficient model compression through techniques like quantization and weight pruning. This work is significant because understanding and controlling the effects of small perturbations is crucial for building reliable, trustworthy, and efficient AI systems across diverse applications, from image recognition and natural language processing to robotics and medical image analysis.

Papers

November 6, 2024

Explaining Human Activity Recognition with SHAP: Validating Insights with Perturbation and Quantitative Measures
Felix Tempel, Espen Alexander F. Ihlen, Lars Adde, Inga Strümke
Action Recognition Activity Recognition DCU Insight AQ Fusion SHAP Small Perturbation Skeleton Data

November 1, 2024

Latent Paraphrasing: Perturbation on Layers Improves Knowledge Injection in Language Models
Minki Kang, Sung Ju Hwang, Gibbeum Lee, Jaewoong Cho
Language Model Multi Layer Paraphrase Generation Knowledge Injection Small Perturbation

August 28, 2024

StyleRemix: Interpretable Authorship Obfuscation via Distillation and Perturbation of Style Elements
Jillian Fisher, Skyler Hallinan, Ximing Lu, Mitchell Gordon, Zaid Harchaoui, Yejin Choi
Large Language Model Mutual Distillation Obfuscation Technique Stylistic Change Small Perturbation Authorship Analysis Authorship Obfuscation

August 23, 2024

Perturbation on Feature Coalition: Towards Interpretable Deep Neural Networks
Xuran Hu, Mingzhe Zhu, Zhenpeng Feng, Miloš Daković, Ljubiša Stanković
Explainable AI Interpretable Deep Learning Small Perturbation Perturbation Analysis Interpretable Network Feature Communication

August 14, 2024

Attention-Guided Perturbation for Unsupervised Image Anomaly Detection
Tingfeng Huang, Yuxuan Cheng, Jingbo Xia, Rui Yu, Yuxuan Cai, Jinhai Xiang, Xinwei He, Xiang Bai
Anomaly Detection Unsupervised Anomaly Detection Small Perturbation Plain Neural

August 7, 2024

Mixstyle-Entropy: Domain Generalization with Causal Intervention and Perturbation
Luyao Tang, Yuxuan Yuan, Chaoqi Chen, Xinghao Ding, Yue Huang
Causal Intervention Causal Variable Small Perturbation Causal Entropy

June 25, 2024

VarBench: Robust Language Model Benchmarking Through Dynamic Variable Perturbation
Kun Qian, Shunji Wan, Claudia Tang, Youzhi Wang, Xuanming Zhang, Maximillian Chen, Zhou Yu
Large Language Model Language Model Data Contamination Small Perturbation Benchmark Data

June 8, 2024

April 22, 2024

Explicit Lipschitz Value Estimation Enhances Policy Robustness Against Perturbation
Xulin Chen, Ruipeng Liu, Garrett E. Katz
Reinforcement Learning Value Function Robust Policy Robust Reinforcement Learning Better Robustness Small Perturbation Lipschitz Regularization

April 21, 2024

Robust EEG-based Emotion Recognition Using an Inception and Two-sided Perturbation Model
Shadi Sartipi, Mujdat Cetin
Emotion Recognition Robust Version Terminator Economy Feature Generation Small Perturbation Inception Module

March 11, 2024

What Makes Quantization for Large Language Models Hard? An Empirical Study from the Lens of Perturbation
Zhuocheng Gong, Jiahao Liu, Jingang Wang, Xunliang Cai, Dongyan Zhao, Rui Yan
Large Language Model Empirical Study Camera Lens Quantization Operator Weight Quantization Small Perturbation LLM Quantization

December 23, 2023

ZO-AdaMU Optimizer: Adapting Perturbation by the Momentum and Uncertainty in Zeroth-order Optimization
Shuoran Jiang, Qingcai Chen, Youchen Pan, Yang Xiang, Yukang Lin, Xiangping Wu, Chuanyi Liu, Xiaobao Song
Large Language Model High Uncertainty Anticipation Zeroth Order Residual Momentum Small Perturbation Gradient Momentum Momentum Based Method

December 20, 2023

Stability of Graph Convolutional Neural Networks through the lens of small perturbation analysis
Lucia Testa, Claudio Battiloro, Stefania Sardellitti, Sergio Barbarossa
Core Stability Camera Lens Graph Convolutional Neural Network Small Perturbation Eigen Portfolio Random Perturbation Laplacian Matrix

November 26, 2023

Hessian Aware Low-Rank Perturbation for Order-Robust Continual Learning
Jiaqi Li, Yuanhao Lai, Rui Wang, Changjian Shui, Sabyasachi Sahoo, Charles X. Ling, Shichun Yang, Boyu Wang, Christian Gagné, Fan Zhou
Continual LEArning Hessian Matrix Low Rank Approximation Small Perturbation Robust Continual Learning Robust Task

June 29, 2023

PCDAL: A Perturbation Consistency-Driven Active Learning Approach for Medical Image Segmentation and Classification
Tao Wang, Xinlin Zhang, Yuanbo Zhou, Junlin Lan, Tao Tan, Min Du, Qinquan Gao, Tong Tong
Classification Code Active Learning Medical Image Segmentation Medical Image Classification Medical Image Datasets PCa Detection Small Perturbation Based Active Learning

April 28, 2023

Using Perturbation to Improve Goodness-of-Fit Tests based on Kernelized Stein Discrepancy
Xing Liu, Andrew B. Duncan, Axel Gandy
Bayesian Perspective Score Matching Small Perturbation Transition Kernel Goodness of Fit Stein Discrepancy

March 7, 2023

Amplitude-Varying Perturbation for Balancing Privacy and Utility in Federated Learning
Xin Yuan, Wei Ni, Ming Ding, Kang Wei, Jun Li, H. Vincent Poor
Differential Privacy Task Utility Small Perturbation Gaussian Mechanism Differential Privacy Mechanism

February 22, 2023

Mitigating Adversarial Attacks in Deepfake Detection: An Exploration of Perturbation and AI Techniques
Saminder Dhesi, Laura Fontes, Pedro Machado, Isibor Kennedy Ihianle, Farhad Fassihi Tash, David Ada Adama
Deep Learning Adversarial Attack Deep Learning Model Adversarial Example Environment Exploration Deepfake Detection Deep Fake Artificial Intelligence Technique Small Perturbation Adversarial Threat

November 12, 2022

Generating Textual Adversaries with Minimal Perturbation
Xingyi Zhao, Lu Zhang, Depeng Xu, Shuhan Yuan
Adversarial Text Small Perturbation Attack Algorithm Word Level Adversarial

Small Perturbation

Papers

Explaining Human Activity Recognition with SHAP: Validating Insights with Perturbation and Quantitative Measures

Latent Paraphrasing: Perturbation on Layers Improves Knowledge Injection in Language Models

StyleRemix: Interpretable Authorship Obfuscation via Distillation and Perturbation of Style Elements

Perturbation on Feature Coalition: Towards Interpretable Deep Neural Networks

Attention-Guided Perturbation for Unsupervised Image Anomaly Detection

Mixstyle-Entropy: Domain Generalization with Causal Intervention and Perturbation

VarBench: Robust Language Model Benchmarking Through Dynamic Variable Perturbation

Perturbation Towards Easy Samples Improves Targeted Adversarial Transferability

One Perturbation is Enough: On Generating Universal Adversarial Perturbations against Vision-Language Pre-training Models

Explicit Lipschitz Value Estimation Enhances Policy Robustness Against Perturbation

Robust EEG-based Emotion Recognition Using an Inception and Two-sided Perturbation Model

What Makes Quantization for Large Language Models Hard? An Empirical Study from the Lens of Perturbation

ZO-AdaMU Optimizer: Adapting Perturbation by the Momentum and Uncertainty in Zeroth-order Optimization

Stability of Graph Convolutional Neural Networks through the lens of small perturbation analysis

Hessian Aware Low-Rank Perturbation for Order-Robust Continual Learning

PCDAL: A Perturbation Consistency-Driven Active Learning Approach for Medical Image Segmentation and Classification

Using Perturbation to Improve Goodness-of-Fit Tests based on Kernelized Stein Discrepancy

Amplitude-Varying Perturbation for Balancing Privacy and Utility in Federated Learning

Mitigating Adversarial Attacks in Deepfake Detection: An Exploration of Perturbation and AI Techniques

Generating Textual Adversaries with Minimal Perturbation