How important is weight symmetry in backpropagation?
Name
1510.05067.pdf
Size
250.36 KB
Format
Adobe PDF
Checksum (MD5)
b46d36d6ffcc9a3d7a8f2f60463ecb41
Author(s)
Liao, Qianli
Leibo, Joel Z
Poggio, Tomaso A
Date Issued
February 2016
Journal
Thirtieth AAAI Conference on Artificial Intelligence
Publisher
Association for the Advancement of Artificial Intelligence
Citation
Liao, Qianli, Joel Z. Leibo and Tomaso Poggio. "How Important is Weight Symmetry in Backpropagation." Thirty-Second AAAI Conference on Artificial Intelligence, February 12-17, 2016, Phoenix, Arizona, Association for the Advancement of Artificial Intelligence, February 2016. © 2016 Association for the Advancement of Artificial Intelligence
Version
Author's final manuscript
Abstract
Gradient backpropagation (BP) requires symmetric feedforward and feedback connections-the same weights must be used for forward and backward passes. This "weight transport problem" (Grossberg 1987) is thought to be one of the main reasons to doubt BP's biologically plausibility. Using 15 different classification datasets, we systematically investigate to what extent BP really depends on weight symmetry. In a study that turned out to be surprisingly similar in spirit to Lillicrap et al.'s demonstration (Lillicrap et al. 2014) but orthogonal in its results, our experiments indicate that: (1) the magnitudes of feedback weights do not matter to performance (2) the signs of feedback weights do matter-the more concordant signs between feedforward and their corresponding feedback connections, the better (3) with feedback weights having random magnitudes and 100% concordant signs, we were able to achieve the same or even better performance than SGD. (4) some normalizations/stabilizations are indispensable for such asymmetric BP to work, namely Batch Normalization (BN) (Ioffe and Szegedy 2015) and/or a "Batch Manhattan" (BM) update rule.
MIT Department
Massachusetts Institute of Technology. Department of Brain and Cognitive Sciences
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
McGovern Institute for Brain Research at MIT
Terms of Use
Creative Commons Attribution-Noncommercial-Share Alike
Persistent DSpace Link
DOI of Published Version
https://www.aaai.org/ocs/index.php/AAAI/AAAI16/paper/view/12325
