[2005.14436v1] Machine learning in spectral domainopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

Deep neural networks are usually trained in the space of the nodes, by adjusting the weights of existing links via suitable optimization protocols. We here propose a radically new approach which anchors the learning process to reciprocal space. Specifically, the training acts on the spectral domain and seeks to modify the eigenvectors and eigenvalues of transfer operators in direct space. The proposed method is ductile and can be tailored to return either linear or non linear classifiers. The performance are competitive with standard schemes, while allowing for a significant reduction of the learning parameter space. Spectral learning restricted to eigenvalues could be also employed for pre-training of the deep neural network, in conjunction with conventional machine-learning schemes. Further, it is surmised that the nested indentation of eigenvectors that defines the core idea of spectral learning could help understanding why deep networks work as well as they do.

1 mentions: @q9ac
Date: 2020/06/29 00:52

Referring Tweets

@q9ac 逆空間におけるDeep learning.計算コストが減るらしい. 逆空間に住んでいる生き物との交信が求められている t.co/yMjIiDtZ71

Related Entries

Read more [1911.01646v1] Dynamics of two-level atom interaction with single-mode fieldcontact arXivarXiv Twitt...
0 users, 1 mentions 2020/01/20 18:52
Read more [1911.03526v1] Multiphoton-excited DUV photolithography for 3D nanofabricationcontact arXivarXiv Twi...
0 users, 1 mentions 2020/01/22 09:50
Read more [1911.03134v1] Critical review of quantum plasmonic models for finite-size mediacontact arXivarXiv T...
0 users, 1 mentions 2020/01/23 23:20
Read more [2001.08788v1] Topological metric detects hidden order in disordered mediacontact arXivarXiv Twitter
0 users, 1 mentions 2020/01/27 17:21
Read more [1705.06913v5] An elementary rigorous proof of bulk-boundary correspondence in the generalized Su-Sc...
0 users, 1 mentions 2020/02/11 23:20
Read more [1911.11127v2] Three-body problem for Langevin dynamics with different temperaturescontact arXivarXi...
0 users, 1 mentions 2020/02/23 20:21