Negative Learning

Negative learning, a subfield of machine learning, focuses on leveraging negative information—examples of what a model *should not* predict—to improve performance and robustness. Current research explores its application in various contexts, including semi-supervised learning, noisy label handling, and open-set recognition, often employing techniques like contrastive learning, adversarial training, and novel loss functions designed to push predictions away from undesired classes. This approach holds significant promise for enhancing model generalization, mitigating catastrophic forgetting in continual learning, and improving the safety and reliability of AI systems, particularly in scenarios with limited labeled data or noisy labels.

Papers

July 14, 2024

Order parameters and phase transitions of continual learning in deep neural networks
Haozhe Shan, Qianyi Li, Haim Sompolinsky
Neural Network Deep Neural Network Continual LEArning Catastrophic Forgetting Anti Forgetting Order Matter Phase Transition Negative Learning

March 17, 2024

Boosting Semi-Supervised Temporal Action Localization by Learning from Non-Target Classes
Kun Xia, Le Wang, Sanping Zhou, Gang Hua, Wei Tang
Unlabeled Video Supervised Temporal Action Localization Negative Learning

March 6, 2024

Negating Negatives: Alignment without Human Positive Samples via Distributional Dispreference Optimization
Shitong Duan, Xiaoyuan Yi, Peng Zhang, Tun Lu, Xing Xie, Ning Gu
Large Language Model Alignment Problem Negative Sampling Divergence Regularization Negative Learning

February 23, 2024

Bidirectional Uncertainty-Based Active Learning for Open Set Annotation
Chen-Chen Zong, Ye-Wen Wang, Kun-Peng Ning, Haibo Ye, Sheng-Jun Huang
Active Learning Open Set Bidirectional Correspondence Uncertainty Based Active Learning Negative Learning Unknown Class

December 15, 2023

Adaptive Integration of Partial Label Learning and Negative Learning for Enhanced Noisy Label Learning
Mengmeng Sheng, Zeren Sun, Zhenhuang Cai, Tao Chen, Yichao Zhou, Yazhou Yao
Partial Label Learning Noisy Label Learning Seamless Integration Complementary Label Negative Learning Label Disambiguation

September 7, 2023

A Robust Negative Learning Approach to Partial Domain Adaptation Using Source Prototypes
Sandipan Choudhuri, Suli Adeniye, Arunabha Sen
Complementary Label Pseudo Label Refinement Negative Learning Partial Domain Adaptation Prototype Highlighting Current Challenge Target Supervision

June 13, 2023

Large Language Models Sometimes Generate Purely Negatively-Reinforced Text
Fabien Roger
Large Language Model Language Model Adversarial Training Generate Quick Biased Feedback Negative Learning

April 14, 2023

NEV-NCD: Negative Learning, Entropy, and Variance regularization based novel action categories discovery
Zahid Hasan, Masud Ahmed, Abu Zaher Md Faridee, Sanjay Purushotham, Heesung Kwon, Hyungtae Lee, Nirmalya Roy
Stable Entropy Label Space Category Discovery Unseen Action Negative Learning Variance Regularization View Action Recognition

March 20, 2023

Boosting Semi-Supervised Learning by Exploiting All Unlabeled Data
Yuhao Chen, Xin Tan, Borui Zhao, Zhaowei Chen, Renjie Song, Jiajun Liang, Xuequan Lu
Semi Supervised Learning Pseudo Label Unlabeled Data Negative Learning Pseudo Labelling Negative Pseudo Label

September 3, 2022

August 6, 2022

Learning Human Cognitive Appraisal Through Reinforcement Memory Unit
Yaosi Hu, Zhenzhong Chen
Memory Augmentation Memory Mechanism Negative Learning Two Memory Reinforcement

May 5, 2022

PI-NLF: A Proportional-Integral Approach for Non-negative Latent Factor Analysis
Ye Yuan, Xin Luo
Negative Learning LIF Model Latent Factor Analysis Non Negative Latent Factor Pi Dual

May 2, 2022

Positive-Unlabeled Learning with Adversarial Data Augmentation for Knowledge Graph Completion
Zhenwei Tang, Shichao Pei, Zhao Zhang, Yongchun Zhu, Fuzhen Zhuang, Robert Hoehndorf, Xiangliang Zhang
LeArning Abstract Knowledge Graph Completion Positive Unlabeled Learning Adversarial Data Augmentation Negative Learning Minimax Game

March 18, 2022

Negative Inner-Loop Learning Rates Learn Universal Features
Tom Starshak
Learning Rate Task Adaptation Meta Gradient Agnostic Meta learnIng Negative Learning Learning Loop

Negative Learning

Papers

Order parameters and phase transitions of continual learning in deep neural networks

Boosting Semi-Supervised Temporal Action Localization by Learning from Non-Target Classes

Negating Negatives: Alignment without Human Positive Samples via Distributional Dispreference Optimization

Bidirectional Uncertainty-Based Active Learning for Open Set Annotation

Adaptive Integration of Partial Label Learning and Negative Learning for Enhanced Noisy Label Learning

A Robust Negative Learning Approach to Partial Domain Adaptation Using Source Prototypes

Large Language Models Sometimes Generate Purely Negatively-Reinforced Text

NEV-NCD: Negative Learning, Entropy, and Variance regularization based novel action categories discovery

Boosting Semi-Supervised Learning by Exploiting All Unlabeled Data

STAD: Self-Training with Ambiguous Data for Low-Resource Relation Extraction

Noise-Robust Bidirectional Learning with Dynamic Sample Reweighting

Learning Human Cognitive Appraisal Through Reinforcement Memory Unit

PI-NLF: A Proportional-Integral Approach for Non-negative Latent Factor Analysis

Positive-Unlabeled Learning with Adversarial Data Augmentation for Knowledge Graph Completion

Negative Inner-Loop Learning Rates Learn Universal Features