Training for object recognition with increasing spatial frequency: A comparison of deep learning with human vision.
Training for object recognition with increasing spatial frequency: A comparison of deep learning with human vision.
复制标题
空间频率递增的物体识别训练：深度学习与人类视觉的比较。
DOI：
10.1167/jov.21.10.14
复制
发表时间：
2021-09-01
影响因子：
1.8
通讯作者：
Op de Beeck H
中科院分区：
文献类型：
--
作者：
Avberšek LK;Zeman A;Op de Beeck H
人类视觉的个体发育和视觉输入的实时神经处理表现出惊人的相似性-对空间频率的敏感性，以粗到细的方式进行。在人类早期发育过程中，对更高空间频率的敏感性随着年龄的增长而增加。在成年期，当人类接收新的视觉输入时，通常首先处理低空间频率，然后再处理更高的空间频率。我们调查了这种由粗到细的进展在多大程度上可能会影响人工视觉中的视觉表征，并将其与成人的表征进行了比较。我们通过在训练过程中逐渐增加空间频率信息，模拟了深度卷积神经网络（CNN）中图像处理的粗到细进展。我们比较了标准训练和粗到细训练后的CNN性能与来自行为和神经成像实验的广泛数据集。与人类相比，使用标准协议训练的CNN对低空间频率信息非常不敏感，在对此类对象图像进行分类方面表现出非常差的性能。通过使用我们的粗到细方法训练CNN，我们将CNN在ImageNet数据集低通滤波图像上的分类准确率从0%提高到32%。从粗到精的训练也使CNN对混合图像中的低空间频率更敏感，这些图像在不同频带中具有冲突的信息。当在包含完整空间频率信息的图像上比较不同训练的网络时，我们没有看到代表性差异。总的来说，这种计算，神经和行为的研究结果的整合显示的相关性的暴露和处理的输入与空间频率内容的变化，在某些方面的高层次的对象表示。
The ontogenetic development of human vision and the real-time neural processing of visual input exhibit a striking similarity—a sensitivity toward spatial frequencies that progresses in a coarse-to-fine manner. During early human development, sensitivity for higher spatial frequencies increases with age. In adulthood, when humans receive new visual input, low spatial frequencies are typically processed first before subsequent processing of higher spatial frequencies. We investigated to what extent this coarse-to-fine progression might impact visual representations in artificial vision and compared this to adult human representations. We simulated the coarse-to-fine progression of image processing in deep convolutional neural networks (CNNs) by gradually increasing spatial frequency information during training. We compared CNN performance after standard and coarse-to-fine training with a wide range of datasets from behavioral and neuroimaging experiments. In contrast to humans, CNNs that are trained using the standard protocol are very insensitive to low spatial frequency information, showing very poor performance in being able to classify such object images. By training CNNs using our coarse-to-fine method, we improved the classification accuracy of CNNs from 0% to 32% on low-pass-filtered images taken from the ImageNet dataset. The coarse-to-fine training also made the CNNs more sensitive to low spatial frequencies in hybrid images with conflicting information in different frequency bands. When comparing differently trained networks on images containing full spatial frequency information, we saw no representational differences. Overall, this integration of computational, neural, and behavioral findings shows the relevance of the exposure to and processing of inputs with variation in spatial frequency content for some aspects of high-level object representations.
登录
查看更多内容
影响因子：
3.8
作者：
Jozwik KM;Kriegeskorte N;Storrs KR;Mur M
通讯作者：
Mur M
影响因子：
--
作者：
Bullier, J
通讯作者：
Bullier, J
影响因子：
5.3
作者：
Adeli, Hossein;Vitu, Francoise;Zelinsky, Gregory J.
通讯作者：
Zelinsky, Gregory J.
影响因子：
4.3
作者：
Kubilius J;Bracci S;Op de Beeck HP
通讯作者：
Op de Beeck HP
DOI：
10.1073/pnas.0507062103
发表时间：
2006-01-10
影响因子：
11.1
作者：
Bar, M;Kassam, KS;Halgren, E
通讯作者：
Halgren, E