A Quantitative Approach to Predicting Representational Learning and Performance in Neural Networks

TL;DR

This paper introduces a pseudo-kernel based method to analyze and predict how neural networks develop representations and perform on tasks based on initial conditions and training strategies.

Contribution

It presents a novel analytical tool that predicts representational learning and multitask performance from initial network states and training curriculum.

Findings

The method accurately predicts the impact of weight initialization scale.

It forecasts how training curriculum affects multitask performance.

Validated on simple and complex neural network scenarios.

Abstract

A key property of neural networks (both biological and artificial) is how they learn to represent and manipulate input information in order to solve a task. Different types of representations may be suited to different types of tasks, making identifying and understanding learned representations a critical part of understanding and designing useful networks. In this paper, we introduce a new pseudo-kernel based tool for analyzing and predicting learned representations, based only on the initial conditions of the network and the training curriculum. We validate the method on a simple test case, before demonstrating its use on a question about the effects of representational learning on sequential single versus concurrent multitask performance. We show that our method can be used to predict the effects of the scale of weight initialization and training curriculum on representational learning and downstream concurrent multitasking performance.