Department of Psychology, University of Liverpool, UK.
School of Psychology, University of Birmingham, UK.
Vision Res. 2022 Aug;197:108058. doi: 10.1016/j.visres.2022.108058. Epub 2022 Apr 26.
In this paper we consider recent advances in the use of deep convolutional neural networks to understanding biological vision. We focus on claims about the plausibility of feedforward deep convolutional neural networks (fDCNNs) as models of image classification in the biological system. Despite the putative similarity of these networks to some properties of the biological vision system, and the remarkable levels of performance accuracy of some fDCNNs, we argue that their plausibility as a framework for understanding image classification remains unclear. We highlight two key issues that we suggest are relevant to the evaluation of any form of DNN used to examine biological vision: (1) Network transparency under analysis - that is, the challenge of understanding what networks do, and how they do it. (2) Identifying appropriate benchmarks for comparing network performance and the biological system using both quantitative and qualitative performance measures. We show that there are important divergences between fDCNNs and biological vision that reflect fundamental differences in computational architectures, and representational structures, supporting image classification in these networks and the biological system.
本文考虑了深度学习卷积神经网络在理解生物视觉方面的最新进展。我们关注的是关于前馈深度卷积神经网络（fDCNN）作为生物系统中图像分类模型的合理性的说法。尽管这些网络与生物视觉系统的某些性质具有潜在的相似性，并且一些 fDCNN 的性能准确性达到了惊人的水平，但我们认为，它们作为理解图像分类的框架的合理性仍然不清楚。我们强调了两个关键问题，我们认为这些问题与用于研究生物视觉的任何形式的 DNN 的评估都相关：（1）分析下的网络透明度——即理解网络的工作原理和方式的挑战。（2）使用定量和定性性能指标，为比较网络性能和生物系统找到合适的基准。我们表明，fDCNN 和生物视觉之间存在重要的差异，这些差异反映了计算架构和表示结构的根本差异，这些差异支持这些网络和生物系统中的图像分类。
