Learning Abstract Classes using Deep Learning

TL;DR

This paper investigates the ability of a deep learning model, GoogLeNet, to learn and generalize abstract concepts like orientation, which are easy for humans but challenging for neural networks.

Contribution

It evaluates CNN performance on abstract class differentiation and transferability, highlighting the gap between human and machine understanding of abstract concepts.

Findings

CNN can differentiate abstract classes with some accuracy

Transfer learning to unseen objects shows limited generalization

Humans outperform CNNs in understanding abstract concepts

Abstract

Humans are generally good at learning abstract concepts about objects and scenes (e.g.\ spatial orientation, relative sizes, etc.). Over the last years convolutional neural networks have achieved almost human performance in recognizing concrete classes (i.e.\ specific object categories). This paper tests the performance of a current CNN (GoogLeNet) on the task of differentiating between abstract classes which are trivially differentiable for humans. We trained and tested the CNN on the two abstract classes of horizontal and vertical orientation and determined how well the network is able to transfer the learned classes to other, previously unseen objects.