Topographic deep artificial neural networks reproduce the hallmarks of the primate inferior temporal cortex face processing network

Topographic deep artificial neural networks reproduce the hallmarks of the primate inferior temporal cortex face processing network
复制标题

地形深度人工神经网络再现了灵长类颞下皮层面部处理网络的特征

DOI:
10.1101/2020.07.09.185116
复制
发表时间:
2020
期刊:
bioRxiv
影响因子:
--
通讯作者:
J. DiCarlo
J. DiCarlo
中科院分区:
--
文献类型:
--
作者:
Hyodong Lee;Eshed Margalit;K. Jozwik;M. Cohen;N. Kanwisher;Daniel L. K. Yamins;J. DiCarlo

文献摘要

参考文献

被引文献

相似文献

猴子下颞叶(IT)皮层的一个显著特征是IT面部处理网络。其特点包括:与非面部对象相比,对面部对象反应更大的“面部神经元”、在每个IT解剖水平的病灶中的这些神经元的强空间聚类(“面部斑块”)以及这些病灶的优先互连。虽然一些深度人工神经网络(ANN)可以很好地预测IT神经元的反应,包括面部神经元,但它们并不能解释这些面部网络特征。在这里,我们问他们是否可以解释一个简单的,代谢动机除了目前的人工神经网络腹侧流模型。具体来说,我们设计并成功训练了地形深度ANN(TDANN)来解决现实世界的视觉识别任务(如之前的工作),但除此之外,我们还优化了每个网络,以最大限度地减少其IT层中神经元布线长度的代理。我们报告说,这种双重优化后,TDANN的模型IT层再现了IT人脸网络的特点:面部神经元的存在,面部神经元的集群,定量匹配那些在IT人脸补丁,这些补丁之间的连接,以及出现的人脸视点不变性沿着网络层次结构。我们发现,这些现象出现了一系列的自然经验,但不是高度非自然的训练。两者合计,这些结果表明,IT的脸处理网络可能是一个基本的层次解剖沿着腹侧流,选择压力的视觉系统,以完成一般的对象分类,并选择压力,以尽量减少轴突布线长度的结果。
A salient characteristic of monkey inferior temporal (IT) cortex is the IT face processing network. Its hallmarks include: “face neurons” that respond more to faces than non-face objects, strong spatial clustering of those neurons in foci at each IT anatomical level (“face patches”), and the preferential interconnection of those foci. While some deep artificial neural networks (ANNs) are good predictors of IT neuronal responses, including face neurons, they do not explain those face network hallmarks. Here we ask if they might be explained with a simple, metabolically motivated addition to current ANN ventral stream models. Specifically, we designed and successfully trained topographic deep ANNs (TDANNs) to solve real-world visual recognition tasks (as in prior work), but, in addition, we also optimized each network to minimize a proxy for neuronal wiring length within its IT layers. We report that after this dual optimization, the model IT layers of TDANNs reproduce the hallmarks of the IT face network: the presence of face neurons, clusters of face neurons that quantitatively match those found in IT face patches, connectivity between those patches, and the emergence of face viewpoint invariance along the network hierarchy. We find that these phenomena emerge for a range of naturalistic experience, but not for highly unnatural training. Taken together, these results show that the IT face processing network could be a consequence of a basic hierarchical anatomy along the ventral stream, selection pressure on the visual system to accomplish general object categorization, and selection pressure to minimize axonal wiring length.
DOI: 10.1146/annurev-neuro-070815-013934
发表时间: 2016-07-08
影响因子: 13.9
作者:
Freiwald W;Duchaine B;Yovel G
通讯作者: Yovel G
DOI: 10.1093/cercor/bhm196
发表时间: 2008-07-01
期刊: CEREBRAL CORTEX
影响因子: 3.7
作者:
Op de Beeck, Hans P.;Deutsch, Jennifer A.;DiCarlo, James J.
通讯作者: DiCarlo, James J.
DOI: 10.1093/cercor/1.1.1
发表时间: 1991-01-01
期刊: CEREBRAL CORTEX
影响因子: 3.7
作者:
Felleman, Daniel J.;Van Essen, David C.
通讯作者: Van Essen, David C.
DOI: 10.1016/j.mri.2005.12.025
发表时间: 2006-05-01
影响因子: 2.5
作者:
Kourtzi, Zoe;Augath, Mark;Kiorpes, Lynne
通讯作者: Kiorpes, Lynne
DOI: 10.1146/annurev-vision-102016-061214
发表时间: 2017-09-15
影响因子: 6
作者:
Grill-Spector K;Weiner KS;Kay K;Gomez J
通讯作者: Gomez J