Zobrazeno 1 - 8
of 8
pro vyhledávání: '"Kumar, Siddharth Krishna"'
Autor:
Kumar, Siddharth Krishna
This paper critically examines the fundamental distinctions between gradient methods applied to non-differentiable functions (NGDMs) and classical gradient descents (GDs) for differentiable functions, revealing significant gaps in current deep learni
Externí odkaz:
http://arxiv.org/abs/2401.08426
Autor:
Kumar, Siddharth Krishna
It is well known that a determined adversary can fool a neural network by making imperceptible adversarial perturbations to an image. Recent studies have shown that these perturbations can be detected even without information about the neural network
Externí odkaz:
http://arxiv.org/abs/1807.10335
Autor:
Kumar, Siddharth Krishna
A proper initialization of the weights in a neural network is critical to its convergence. Current insights into weight initialization come primarily from linear activation functions. In this paper, I develop a theory for weight initializations with
Externí odkaz:
http://arxiv.org/abs/1704.08863
Autor:
DAKSHNAMOORTHY, Easu, RYNTATHIANG, Ralph H., SIVAKUMAR, Sarang, VINOD KUMAR, Siddharth Krishna
Publikováno v:
Mechanika; 2024, Vol. 30 Issue 2, p183-187, 5p
Publikováno v:
Proceedings of the National Academy of Sciences of the United States of America, 2016 Aug 01. 113(32), E4581-E4581.
Externí odkaz:
https://www.jstor.org/stable/26471361
Publikováno v:
Proceedings of the National Academy of Sciences of the United States of America, 2016 Jan 01. 113(1), E61-E70.
Externí odkaz:
https://www.jstor.org/stable/26467329
Publikováno v:
Proceedings of the National Academy of Sciences of the United States of America; 8/9/2016, Vol. 113 Issue 32, pE4581-E4581, 1p
Publikováno v:
Proceedings of the National Academy of Sciences of the United States of America [Proc Natl Acad Sci U S A] 2016 Feb 09; Vol. 113 (6), pp. E813. Date of Electronic Publication: 2016 Feb 01.