Research map / Architectures and efficiency / RNNs and normalization
Epitopological learning and Cannistraci-Hebb network shape intelligence brain-inspired theory for ultra-sparse advantage in deep learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes brain-inspired epitopological learning and the Cannistraci-Hebb training (CHT) rule for dynamic sparse training of ultra-sparse deep networks, along with an ultra-sparse advantage challenge.
Problem
Sparse training aims to replace fully connected ANNs with sparse or ultra-sparse networks, but it is unclear to what extent ultra-sparse networks can match or outperform dense ones, and which brain-inspired link-prediction-style learning paradigms can guide sparse topology evolution.
Approach
Applies epitopological learning from network science (link prediction via local-community-likelihood and Hebbian-like rules) to dynamically evolve sparse connectivity during training, so that ultra-sparse networks can be trained with topology driven by Cannistraci-Hebb principles rather than random or magnitude-only rules.
Experimental setup
Baselines: sparse evolutionary training dense networks
Builds on: sparse evolutionary training dynamic sparse training
Closest papers by meaning
- Brain network science modelling of sparse neural networks enables Transformers and LLMs to perform as fully connected (NeurIPS 2025 · similarity 0.89)
- Random Sparse Lifts: Construction, Analysis and Convergence of finite sparse networks (ICLR 2024 · similarity 0.83)
- Rethinking Hebbian Principle: Low-Dimensional Structural Projection for Unsupervised Learning (NeurIPS 2025 · similarity 0.83)
- How connectivity structure shapes rich and lazy learning in neural circuits (ICLR 2024 · similarity 0.83)
- Improving the Sparse Structure Learning of Spiking Neural Networks from the View of Compression Efficiency (ICLR 2025 · similarity 0.83)
- The Graphon Limit Hypothesis: Understanding Neural Network Pruning via Infinite Width Analysis (NeurIPS 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.