Budget-Aware Pruning for Multi-Domain Learning

TL;DR

This paper introduces a novel method for pruning multi-domain deep learning models, reducing their size and computational cost while maintaining performance across multiple tasks, suitable for resource-constrained environments.

Contribution

It proposes a budget-aware pruning approach that encourages shared filters across domains, enabling efficient multi-domain models with fewer parameters and lower complexity.

Findings

Achieves significant model size reduction while maintaining accuracy.

Supports multiple domains with fewer parameters than single-domain models.

Outperforms baseline models in resource-limited scenarios.

Abstract

Deep learning has achieved state-of-the-art performance on several computer vision tasks and domains. Nevertheless, it still has a high computational cost and demands a significant amount of parameters. Such requirements hinder the use in resource-limited environments and demand both software and hardware optimization. Another limitation is that deep models are usually specialized into a single domain or task, requiring them to learn and store new parameters for each new one. Multi-Domain Learning (MDL) attempts to solve this problem by learning a single model that is capable of performing well in multiple domains. Nevertheless, the models are usually larger than the baseline for a single domain. This work tackles both of these problems: our objective is to prune models capable of handling multiple domains according to a user defined budget, making them more computationally affordable while keeping a similar classification performance. We achieve this by encouraging all domains to use a similar subset of filters from the baseline model, up to the amount defined by the user's budget. Then, filters that are not used by any domain are pruned from the network. The proposed approach innovates by better adapting to resource-limited devices while, to our knowledge, being the only work that is capable of handling multiple domains at test time with fewer parameters and lower computational complexity than the baseline model for a single domain.