Hebbian Learning with Global Direction

2026-01-29

AI Summary

This work proposes a model-agnostic, globally guided Hebbian learning (GHL) framework that overcomes the limitations of traditional Hebbian learning, which is constrained by purely local update rules and struggles to balance biological plausibility with performance in large-scale networks and complex tasks. By integrating task-dependent global sign signals with local Oja’s rule and competitive learning mechanisms, GHL introduces directional guidance from global objectives while preserving biological realism. The approach substantially improves Hebbian learning performance on challenging benchmarks such as ImageNet, significantly narrowing the gap with backpropagation-based methods and enabling effective scaling of biologically inspired learning to deep neural architectures.

Technology Category

• Machine Learning: Bio-inspired Learning
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Social networks and social learning
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Backpropagation algorithm has driven the remarkable success of deep neural networks, but its lack of biological plausibility and high computational costs have motivated the ongoing search for alternative training methods. Hebbian learning has attracted considerable interest as a biologically plausible alternative to backpropagation. Nevertheless, its exclusive reliance on local information, without consideration of global task objectives, fundamentally limits its scalability. Inspired by the biological synergy between neuromodulators and local plasticity, we introduce a novel model-agnostic Global-guided Hebbian Learning (GHL) framework, which seamlessly integrates local and global information to scale up across diverse networks and tasks. In specific, the local component employs Oja's rule with competitive learning to ensure stable and effective local updates. Meanwhile, the global component introduces a sign-based signal that guides the direction of local Hebbian plasticity updates. Extensive experiments demonstrate that our method consistently outperforms existing Hebbian approaches. Notably, on large-scale network and complex datasets like ImageNet, our framework achieves the competitive results and significantly narrows the gap with standard backpropagation.

Problem

• Hebbian learning
• biological plausibility
• global information
• scalability
• local plasticity

Innovation

• Hebbian learning
• global guidance
• biological plausibility
• Oja's rule
• model-agnostic