How to Train Your Multi-Exit Model? Analyzing the Impact of Training Strategies

TL;DR

This paper analyzes how different training strategies affect multi-exit neural networks, introduces metrics for analysis, and proposes a mixed training approach that improves performance and efficiency.

Contribution

It introduces a set of metrics to analyze training dynamics and proposes a novel mixed training strategy for multi-exit models, demonstrating its advantages.

Findings

Conventional joint and disjoint training strategies are suboptimal.

The proposed mixed training strategy improves performance and efficiency.

Comprehensive evaluations validate the effectiveness of the new approach.

Abstract

Early exits enable the network's forward pass to terminate early by attaching trainable internal classifiers to the backbone network. Existing early-exit methods typically adopt either a joint training approach, where the backbone and exit heads are trained simultaneously, or a disjoint approach, where the heads are trained separately. However, the implications of this choice are often overlooked, with studies typically adopting one approach without adequate justification. This choice influences training dynamics and its impact remains largely unexplored. In this paper, we introduce a set of metrics to analyze early-exit training dynamics and guide the choice of training strategy. We demonstrate that conventionally used joint and disjoint regimes yield suboptimal performance. To address these limitations, we propose a mixed training strategy: the backbone is trained first, followed by the training of the entire multi-exit network. Through comprehensive evaluations of training strategies across various architectures, datasets, and early-exit methods, we present the strengths and weaknesses of the early exit training strategies. In particular, we show consistent improvements in performance and efficiency using the proposed mixed strategy.