psvit-structured-pruning-spiking-vision

name	psvit-structured-pruning-spiking-vision
description	Structured pruning methodology for Spiking Vision Transformers (SViT) using uniform channel-wise filter pruning and sensitivity analysis for 22.4% memory saving
version	1.0.0
category	ai_collection
tags	["deep-learning","neuromorphic","SNN","pruning","efficiency","vision"]
arxiv	2606.03257v1
paper_title	PSViT: A Methodology for Structurally Pruning Spiking Vision Transformers
authors	["Rachmad Vidya Wicaksana Putra","Achyuta Muthuvelan","Alberto Marchisio","Muhammad Shafique"]
published	2026-06-02T00:00:00.000Z
activation_keywords	["spiking neural network","vision transformer pruning","structured pruning","neuromorphic efficiency","SViT","channel-wise pruning"]

PSViT: Structured Pruning for Spiking Vision Transformers

Core Innovation

Structured pruning for Spiking Vision Transformers (SViT) enabling efficient acceleration on existing hardware (vs. unstructured pruning requiring specialized architectures).

Problem Addressed

Unstructured pruning limitations:

• Requires specialized hardware for sparsity patterns
• Not scalable for widespread deployment
• Hardware-dependent efficiency gains

Methodology

Three-Stage Pruning Pipeline

1. Uniform channel-wise filter pruning: Structurally eliminate non-significant weights
2. Sensitivity analysis: Evaluate pruning impact per layer on accuracy/size
3. Fine-grained channel-wise pruning: Layer-specific pruning based on sensitivity

Key Advantages

• Hardware-agnostic: Works on standard computing architectures
• Structured sparsity: Regular patterns for efficient execution
• Accuracy preservation: Maintains high performance with fine-tuning

Performance Results

• Memory saving: 22.4% through single-shot pruning
• Accuracy:
  • Without fine-tuning: 70.3% (3.0% drop)
  • With fine-tuning: 72.8% (0.5% drop)
  • Original SViT: 73.3%
• Dataset: ImageNet-1K

Implementation Pattern

import torch

class PSViTPruner:
    def __init__(self, svit_model, target_reduction=0.224):
        self.model = svit_model
        self.target_reduction = target_reduction
        
    def uniform_channel_pruning(self, layer, pruning_ratio):
        """Stage 1: Uniform channel-wise filter pruning"""
        weight = layer.weight.data
        num_channels = weight.shape[0]
        prune_count = (num_channels * pruning_ratio)
        
        
        channel_importance = torch.norm(weight, p=, dim=(, , ))
        prune_channels = torch.argsort(channel_importance)[:prune_count]
        
        
        layer.weight.data[prune_channels] = 
         prune_channels
    
     ():
        
        sensitivities = {}
         name, layer  .model.named_modules():
             (layer, ):
                
                sensitivity = .measure_layer_sensitivity(layer)
                sensitivities[name] = sensitivity
         sensitivities
    
     ():
        
        pruned_layers = {}
         name, layer  .model.named_modules():
             (layer, ):
                
                sensitivity = sensitivities[name]
                pruning_ratio = .compute_pruning_ratio(sensitivity)
                pruned_layers[name] = .uniform_channel_pruning(layer, pruning_ratio)
         pruned_layers
    
     ():
        
        original_accuracy = .evaluate_model()
        
        
        .prune_temporarily(layer, ratio=)
        pruned_accuracy = .evaluate_model()
        
        
        .restore_layer(layer)
        
         original_accuracy - pruned_accuracy