# Online Filter Clustering and Pruning for Efficient Convnets

## TL;DR

This paper introduces an online filter clustering and pruning method for convolutional neural networks that groups similar filters during training, enabling more effective pruning and acceleration of DNNs.

## Contribution

It proposes a novel online clustering loss to group similar filters during training, facilitating more efficient filter pruning within residual blocks.

## Key findings

- Achieves competitive accuracy on CIFAR10 and CIFAR100.
- Effectively prunes filters by clustering similar ones during training.
- Reduces model complexity while maintaining performance.

## Abstract

Pruning filters is an effective method for accelerating deep neural networks (DNNs), but most existing approaches prune filters on a pre-trained network directly which limits in acceleration. Although each filter has its own effect in DNNs, but if two filters are the same with each other, we could prune one safely. In this paper, we add an extra cluster loss term in the loss function which can force filters in each cluster to be similar online. After training, we keep one filter in each cluster and prune others and fine-tune the pruned network to compensate for the loss. Particularly, the clusters in every layer can be defined firstly which is effective for pruning DNNs within residual blocks. Extensive experiments on CIFAR10 and CIFAR100 benchmarks demonstrate the competitive performance of our proposed filter pruning method.