| HAL-Inria | Publications, software ... of Inria's scientists |
Reports
Recurrent neural network weight estimation through backward tuning
Recurrent neural network weight estimation through backward tuning
Abstract : We consider another formulation of weight estimation in recurrent networks,
proposing a notation for a large amount of recurrent network units that helps
formulating the estimation problem. Reusing a “good old” control-theory principle,
improved here using a backward-tuning numerical stabilization heuristic, we obtain
a numerically stable and rather efficient second-order and distributed estimation,
without any meta-parameter to adjust. The relation with existing technique is discussed
at each step. The proposed method is validated using reverse engineering
tasks.
Cited literature [61 references]
https://hal.inria.fr/hal-01610735
Contributor : Thierry Viéville Connect in order to contact the contributor
Submitted on : Thursday, October 5, 2017 - 9:11:21 AM
Last modification on : Friday, January 21, 2022 - 3:10:03 AM
Contributor : Thierry Viéville Connect in order to contact the contributor
Submitted on : Thursday, October 5, 2017 - 9:11:21 AM
Last modification on : Friday, January 21, 2022 - 3:10:03 AM
File
RR-9100.pdf
Files produced by the author(s)