paper-with-me

Papers

Entropy Regularizing Activation: Boosting Continuous Control, Large Language Models, and Image Classification with Activation as Entropy Constraints

2025-10-09 · Zilin Kang, Chonghua Liao, Tingqiang Xu, Huazhe Xu arxiv

We propose ERA, a new paradigm that constrains the sampling entropy above given thresholds by applying specially designed activations to the outputs of models. Our approach demonstrates broad effectiveness across different domains: 1) for large language models(LLMs), boosting the AIME 2025 score for Qwen2.5-Math-7B by 37.4%; 2) for continuous control reinforcement learning agents, improving performance by more than 30% over strong baselines such as SAC on the challenging HumanoidBench; 3) for image classification, enhancing ImageNet top-1 accuracy by 0.69% for ResNet-50. These gains are achieved with a computational overhead of less than 7%. Our work validates output activation as a powerful tool for entropy control, opening a new direction for designing simpler and more robust algorithms.

📄 PDF Abstract BibTeX arXiv:2510.08549

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement LearningImage ClassificationContinuous Control

Similar Papers 제목 키워드 기반

Neural Networks with Smooth Adaptive Activation Functions for Regression

2016-08-23 · Le Hou, Dimitris Samaras, Tahsin M. Kurc, Yi Gao 외

In Neural Networks (NN), Adaptive Activation Functions (AAF) have parameters that control the shapes of activation functions. These parameters are trained along with other parameters in the NN. AAFs have improved perform…

regression

Regularization Matters in Policy Optimization - An Empirical Study on Continuous Control

2021-01-01 · ICLR 2021 1 · Zhuang Liu, Xuanlin Li, Bingyi Kang, Trevor Darrell

Deep Reinforcement Learning (Deep RL) has been receiving increasingly more attention thanks to its encouraging performance on a variety of control tasks. Yet, conventional regularization techniques in training neural ne…

continuous-controlContinuous ControlDeep Reinforcement Learning

Boosting Reasoning in Large Multimodal Models via Activation Replay

2025-11-25 · Yun Xing, Xiaobin Hu, Qingdong He, Jiangning Zhang 외 arxiv

Recently, Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as an effective approach to incentivizing reasoning capability in Large Multimodal Models (LMMs), while the underlying mechanisms behind this po…

Reinforcement LearningMultimodal Reasoning

Regularization Matters in Policy Optimization

2019-10-21 · Zhuang Liu, Xuanlin Li, Bingyi Kang, Trevor Darrell

Deep Reinforcement Learning (Deep RL) has been receiving increasingly more attention thanks to its encouraging performance on a variety of control tasks. Yet, conventional regularization techniques in training neural net…

continuous-controlContinuous ControlDeep Reinforcement LearningReinforcement Learning+1

ConvNets with Smooth Adaptive Activation Functions for Regression

2017-01-01 · Le Hou ; Dimitris Samaras ; Tahsin M. Kurc ; Yi Gao ; Joel H. Saltz

Within Neural Networks (NN), the parameters of Adaptive Activation Functions (AAF) control the shapes of activation functions. These parameters are trained along with other parameters in the NN. AAFs have improved perfo…

Age And Gender ClassificationPose Estimationregression