[1905.10679] Improved object recognition using neural networks trained to mimic the brain's statistical propertiesopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

The current state-of-the-art object recognition algorithms, deep convolutional neural networks (DCNNs), are inspired by the architecture of the mammalian visual system, and are capable of human-level performance on many tasks. However, even these algorithms make errors. As they are trained for object recognition tasks, it has been shown that DCNNs develop hidden representations that resemble those observed in the mammalian visual system. Moreover, DCNNs trained on object recognition tasks are currently among the best models we have of the mammalian visual system. This led us to hypothesize that teaching DCNNs to achieve even more brain-like representations could improve their performance. To test this, we trained DCNNs on a composite task, wherein networks were trained to: a) classify images of objects; while b) having intermediate representations that resemble those observed in neural recordings from monkey visual cortex. Compared with DCNNs trained purely for object categorization, D

3 mentions: @hardmaru@osoleve
Date: 2020/08/11 02:21

Referring Tweets

@hardmaru Apparently this article is behind a paywall (preprint: t.co/01qCUmhllG). For some reason, I can access it through a VPN connected from some random country. Here is a link to the PDF that should work: t.co/RSRQYMl1hb

Related Entries

Read more [1912.11570] SketchTransfer: A Challenging New Task for Exploring Detail-Invariance and the Abstract...
0 users, 1 mentions 2020/02/12 03:51
Read more Network Randomization: A Simple Technique for Generalization in Deep Reinforcement Learning | OpenRe...
0 users, 1 mentions 2020/02/26 02:20
Read more [1912.04161] Reinforcement Learning with Convolutional Reservoir Computingcontact arXivarXiv Twitter
0 users, 1 mentions 2020/03/12 02:21
Read more Model-Predictive Policy Learning with Uncertainty Regularization for Driving in Dense Traffic | Open...
0 users, 1 mentions 2020/03/12 06:51
Read more Tutorials - Overleaf, Online LaTeX Editor
1 users, 3 mentions 2020/09/12 06:52