E$^2$CM: Early Exit via Class Means for Efficient Supervised and Unsupervised Learning

Alperen Gormez, Erdem Koyuncu

Keywords: class means, early exit, efficient neural networks

Abstract: State-of-the-art neural networks with early exit mechanisms often need considerable amount of training and fine-tuning to achieve good performance with low computational cost. We propose a novel early exit technique, E$^2$CM, based on the class means of samples. Unlike most existing schemes, E$^2$CM does not require gradient-based training of internal classifiers. This makes it particularly useful for neural network training in low-power devices, as in wireless edge networks. In particular, given a fixed training time budget, E$^2$CM achieves higher accuracy as compared to existing early exit mechanisms. Moreover, if there are no limitations on the training time budget, E$^2$CM can be combined with an existing early exit scheme to boost the latter's performance, achieving a better trade-off between computational cost and network accuracy. We also show that E$^2$CM can be used to decrease the computational cost in unsupervised learning tasks.

One-sentence Summary: We propose a better early exit technique for deep neural networks, which achieves the same performance with less computation.