Are the High-weight Neurons the Important Ones in Image Classification Neural Networks?

2026-07-28

AI Summary

This study challenges the common assumption that high-weight neurons are necessarily critical to classification performance, a relationship that remains poorly understood. Through three complementary experimental approaches—overlap analysis, selective perturbation, and targeted pruning followed by retraining—the authors systematically evaluate neuron importance on CIFAR-10 and Mini-ImageNet. Their findings reveal that only about 25% of the top 10% highest-weight neurons are genuinely indispensable; perturbing certain low-weight neurons can degrade accuracy by 10–17%; and post-pruning retraining demonstrates that neuron importance cannot be linearly inferred from weight magnitude alone. These results underscore that neuron importance is distributed nonlinearly and emphasize the necessity of assessing functional impact rather than relying solely on weight-based heuristics to gauge neuronal contribution.

Abstract

As neural network models for image classification advance, neurons play critical roles in pruning, backdoor defense, and interpretability. Yet existing work lacks clarity on the weight-importance relationship. We address this with a neuron importance assessment method using three experiments: quantifying overlap between high-weight and accuracy-impacting neurons, analyzing high-weight neuron perturbation effects, and testing post-retraining accuracy after high-weight neuron ablation. Experiments on CIFAR-10 and Mini-ImageNet reveal key patterns. Overlap analysis shows top 10\% high-weight neurons overlap with important ones by only about 25\% at maximum, dropping further in subsequent intervals. Perturbation tests find top 10\% high-weight neurons cause 45-80\% accuracy degradation under certain operations compared to 3-7\% for random perturbations, but a third of them show minimal impact. Ablation-retraining results show removing top 10\% high-weight neurons leaves accuracy 10-20\% below baseline with no recovery, while ablating top 0.1\% allows near-full recovery. Notably, some low-weight intervals show 10-17\% degradation when perturbed, comparable to mid-range high-weight neurons. These results confirm not all high-weight neurons are important: their importance is nonlinear. Low-weight neurons also contribute significantly. This challenges weight-importance equivalence, offering refined neuron role insights. It supports applications like encryption prioritizing critical high-weight neurons and pruning removing non-critical ones, advancing neural network analysis.

Problem

• neuron importance
• weight-importance relationship
• image classification
• neural networks
• model interpretability

Innovation

• neuron importance
• weight-importance relationship
• neural network pruning
• perturbation analysis
• ablation study