abstracts[] |
{'sha1': 'fd9b42f2008dd3969402b43a3763a0bd078529f9', 'content': 'Deep convolutional neural networks (CNNs) have demonstrated impressive\nperformance on many visual tasks. Recently, they became useful models for the\nvisual system in neuroscience. However, it is still not clear what are learned\nby CNNs in terms of neuronal circuits. When a deep CNN with many layers is used\nfor the visual system, it is not easy to compare the structure components of\nCNN with possible neuroscience underpinnings due to highly complex circuits\nfrom the retina to higher visual cortex. Here we address this issue by focusing\non single retinal ganglion cells with biophysical models and recording data\nfrom animals. By training CNNs with white noise images to predict neuronal\nresponses, we found that fine structures of the retinal receptive field can be\nrevealed. Specifically, convolutional filters learned are resembling biological\ncomponents of the retinal circuit. This suggests that a CNN learning from one\nsingle retinal cell reveals a minimal neural network carried out in this cell.\nFurthermore, when CNNs learned from different cells are transferred between\ncells, there is a diversity of transfer learning performance, which indicates\nthat CNNs are cell-specific. Moreover, when CNNs are transferred between\ndifferent types of input images, here white noise v.s. natural images, transfer\nlearning shows a good performance, which implies that CNN indeed captures the\nfull computational ability of a single retinal cell for different inputs. Taken\ntogether, these results suggest that CNN could be used to reveal structure\ncomponents of neuronal circuits, and provide a powerful model for neural system\nidentification.', 'mimetype': 'text/plain', 'lang': 'en'}
|
contribs[] |
{'index': 0, 'creator_id': None, 'creator': None, 'raw_name': 'Qi Yan', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 1, 'creator_id': None, 'creator': None, 'raw_name': 'Yajing Zheng', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 2, 'creator_id': None, 'creator': None, 'raw_name': 'Shanshan Jia', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 3, 'creator_id': None, 'creator': None, 'raw_name': 'Yichen Zhang', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 4, 'creator_id': None, 'creator': None, 'raw_name': 'Zhaofei Yu', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 5, 'creator_id': None, 'creator': None, 'raw_name': 'Feng\n Chen', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 6, 'creator_id': None, 'creator': None, 'raw_name': 'Yonghong Tian', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 7, 'creator_id': None, 'creator': None, 'raw_name': 'Tiejun Huang', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
{'index': 8, 'creator_id': None, 'creator': None, 'raw_name': 'Jian K. Liu', 'given_name': None, 'surname': None, 'role': 'author', 'raw_affiliation': None, 'extra': None}
|
ext_ids |
{'doi': None, 'wikidata_qid': None, 'isbn13': None, 'pmid': None, 'pmcid': None, 'core': None, 'arxiv': '1811.02290v1', 'jstor': None, 'ark': None, 'mag': None, 'doaj': None, 'dblp': None, 'oai': None, 'hdl': None}
|
files[] |
{'state': 'active', 'ident': 'fva5l4yuwfgplej5k5ocfkrpya', 'revision': 'f9bc8578-2ba4-4cb0-b6d5-0a8ee6bc2753', 'redirect': None, 'extra': None, 'edit_extra': None, 'size': 2061257, 'md5': '739ff795a30ddcef50421df9b94a7bc5', 'sha1': 'b83548678e274f24b54cbf32228d7b2ee740c07e', 'sha256': '086356afd6cca07261ca861f80d6f249981bebf09a27ce009550dbd320203571', 'urls': [{'url': 'https://arxiv.org/pdf/1811.02290v1.pdf', 'rel': 'repository'}, {'url': 'https://web.archive.org/web/20200923065107/https://arxiv.org/pdf/1811.02290v1.pdf', 'rel': 'webarchive'}], 'mimetype': 'application/pdf', 'content_scope': None, 'release_ids': ['y7g2wnjhwbcuxjtf6puy7aahpi'], 'releases': None}
|