Coarse-Graining Hidden Representations: Unsupervised Neuron Selection via Mapping Entropy

SummaryAI generated

On September 7, 2026, arXiv cs.LG published the paper “Coarse-Graining Hidden Representations: Unsupervised Neuron Selection via Mapping Entropy”. This study proposes a unsupervised neuron selection mechanism based on mapping entropy, aimed at coarse-graining hidden layer representations to optimize neural network structure and improve model efficiency.

Related eventsRELATED EVENTS

2026-09-09 05:29OpenAI claims huge maths breakthrough on a famed ‘Millennium Problem’
2026-09-09 06:11OpenAI claims to have solved the Navier-Stokes equation problem, but its originality is questioned by the academic community
2026-09-08 22:00Google DeepMind Maps 9 Billion Possible DNA Variants

Key entitiesKEY ENTITIES

arXiv

SignalsSIGNALS

Keyword heat

arXiv1

All reports (1)SOURCES

↗

A arXiv cs.LG en 2026-09-07 12:00

研究提出一种基于映射熵（ME）的无监督神经元选择方法，旨在从过参数化神经网络的隐藏层中筛选出关键单元。该方法通过最小化因丢弃部分网络神经元而导致的区分能力损失来评估候选选择，仅依赖隐藏激活统计信息即可运作。在教师 - 学生网络和翻译增强 MNIST 任务上验证显示，ME 优化能恢复与教师网络一致的最小表示并保留与残差变异性成比例的额外单元；所选子网络在同等大小下表现优于随机子集，尤其在强压缩条件下显著提升了预测性能。