A Practical Approach to Sizing Neural Networks
This provides a practical tool for researchers and practitioners to optimize neural network sizing, reducing computational costs and improving generalization, though it is incremental in applying existing information-theoretic models.
The paper tackles the problem of determining the maximum size of a neural network for a given dataset to avoid overfitting and inefficiency, presenting analytical rules and a heuristic method validated experimentally to estimate capacity requirements.
Memorization is worst-case generalization. Based on MacKay's information theoretic model of supervised machine learning, this article discusses how to practically estimate the maximum size of a neural network given a training data set. First, we present four easily applicable rules to analytically determine the capacity of neural network architectures. This allows the comparison of the efficiency of different network architectures independently of a task. Second, we introduce and experimentally validate a heuristic method to estimate the neural network capacity requirement for a given dataset and labeling. This allows an estimate of the required size of a neural network for a given problem. We conclude the article with a discussion on the consequences of sizing the network wrongly, which includes both increased computation effort for training as well as reduced generalization capability.