Subhayan De
University of Colorado Boulder
26 Papers
58 Citations
Subhayan De is an academic researcher from University of Colorado Boulder. The author has contributed to research in topics: Stochastic gradient descent & Computer science. The author has an hindex of 6, co-authored 19 publications. Previous affiliations of Subhayan De include University of Southern California.
Chat about Author
Papers
Topology optimization under uncertainty using a stochastic gradient-based approach
TL;DR: An optimization approach that generates a stochastic approximation of the objective, constraints, and their gradients via a small number of adjoint (and/or forward) solves, per iteration is proposed.
51
On transfer learning of neural networks using bi-fidelity data for uncertainty propagation
TL;DR: In this article, the authors explore the application of transfer learning techniques using training data generated from both high- and low-fidelity models and explore two strategies for coupling these two datasets during the training procedure.
51
•Posted Content
On transfer learning of neural networks using bi-fidelity data for uncertainty propagation
TL;DR: This paper explores the application of transfer learning techniques using training data generated from both high- and low-fidelity models to alleviate the issue for uncertainty propagation and focuses on accuracy improvement achieved by transfer learning over standard training approaches.
43
Bi-fidelity modeling of uncertain and partially unknown systems using DeepONets
Subhayan De,Malik Hassanaly,Matthew Reynolds,Ryan N. King,Alireza Doostan +4 more
TL;DR: In this article , the authors proposed a bi-fidelity modeling approach for complex physical systems, where they model the discrepancy between the true system's response and a low fidelity response using a deep operator network, a neural network architecture suitable for approximating nonlinear operators.
•Posted Content
Topology Optimization under Uncertainty using a Stochastic Gradient-based Approach
TL;DR: In this paper, a stochastic approximation of the objective, constraints, and their gradients via a small number of adjoint (and/or forward) solves, per iteration, is proposed.