TY - GEN
T1 - Spline Filters For End-to-End Deep Learning
AU - Balestriero, Randall
AU - Cosentino, Romain
AU - Glotin, Herv
AU - Baraniuk, Richard
N1 - Funding Information:
Richard Baraniuk and Randall Balestriero were supported by DOD Vannevar Bush Faculty Fellowship grant N00014-18-1-2047, NSF grant CCF-1527501, ARO grant W911NF-15-1-0316, AFOSR grant FA9550-14-1-0088, ONR grant N00014-17-1-2551, DARPA REVEAL grant HR0011-16 C-0028, and an ONR BRC grant for Randomized Numerical Linear Algebra. This work was partially supported by EADM MADICS and SABIOD.org
Publisher Copyright:
© Copyright 2018 by the Authors. All rights reserved.
PY - 2018
Y1 - 2018
N2 - We propose to tackle the problem of end-to-end learning for raw waveforms signals by introducing lcamablc continuous time-frequency atoms. The derivation of these filters is achieved by first, defining a functional space with a given smoothness order and boundary conditions. From this space, we derive the parametric analytical filters. Their differentiability property allows gradient- based optimization. As such, one can equip any Deep Neural Networks (DNNs) with these filters. This enables .us to tackle in a front-end fashion a large scale bird detection task based on the freefieldlOlO dataset known to contain key challenges, such as high dimensional inputs (> 100000) and the presence of multiple sources and soundscapcs.
AB - We propose to tackle the problem of end-to-end learning for raw waveforms signals by introducing lcamablc continuous time-frequency atoms. The derivation of these filters is achieved by first, defining a functional space with a given smoothness order and boundary conditions. From this space, we derive the parametric analytical filters. Their differentiability property allows gradient- based optimization. As such, one can equip any Deep Neural Networks (DNNs) with these filters. This enables .us to tackle in a front-end fashion a large scale bird detection task based on the freefieldlOlO dataset known to contain key challenges, such as high dimensional inputs (> 100000) and the presence of multiple sources and soundscapcs.
UR - http://www.scopus.com/inward/record.url?scp=85057273546&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85057273546&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85057273546
T3 - 35th International Conference on Machine Learning, ICML 2018
SP - 636
EP - 645
BT - 35th International Conference on Machine Learning, ICML 2018
A2 - Krause, Andreas
A2 - Dy, Jennifer
PB - International Machine Learning Society (IMLS)
T2 - 35th International Conference on Machine Learning, ICML 2018
Y2 - 10 July 2018 through 15 July 2018
ER -