Representational Distance Learning for Deep Neural Networks

TL;DR

This paper introduces representational distance learning (RDL), a method for training deep neural networks by aligning their internal representational spaces with those of a reference model, improving performance and enabling biological plausibility.

Contribution

The paper proposes RDL, a novel transfer learning technique that aligns internal representational spaces of neural networks using RDMs, applicable across different architectures.

Findings

RDL is competitive with existing transfer learning methods on MNIST and CIFAR-100.

RDL significantly improves classification performance over baseline networks.

RDL allows for architectural differences between student and teacher models.

Abstract

Deep neural networks (DNNs) provide useful models of visual representational transformations. We present a method that enables a DNN (student) to learn from the internal representational spaces of a reference model (teacher), which could be another DNN or, in the future, a biological brain. Representational spaces of the student and the teacher are characterized by representational distance matrices (RDMs). We propose representational distance learning (RDL), a stochastic gradient descent method that drives the RDMs of the student to approximate the RDMs of the teacher. We demonstrate that RDL is competitive with other transfer learning techniques for two publicly available benchmark computer vision datasets (MNIST and CIFAR-100), while allowing for architectural differences between student and teacher. By pulling the student's RDMs towards those of the teacher, RDL significantly improved visual classification performance when compared to baseline networks that did not use transfer learning. In the future, RDL may enable combined supervised training of deep neural networks using task constraints (e.g. images and category labels) and constraints from brain-activity measurements, so as to build models that replicate the internal representational spaces of biological brains.