Jan 20, 2022: Doubly Adaptive Scaled Algorithm for Machine Learning Using Second-Order Information, joint work with Majid Jahani, Sergey Rusakov, Peter Richtárik, Michael W. Mahoney, and Martin Takáč, was accepted by ICLR 2022.
I told Chelsea it is where unicorns lived …
Presenting Paper @ICMLA 2020
On December 15, 2020, I presented the paper of Finite Difference Neural Network: Fast Prediction of Partial Differential Equations at 19TH IEEE International Conference on Machine Learning and Applications. Video Presentation: Available on Youtube Co-authors: Dr. Nur Sila Gulgec, Prof. Albert S. Berahas, Prof. Shamim N. Pakzad, and Prof. Martin…
Hessian-free Optimization for Fast Inference – Deep Learning on Learning PDEs
On November 13, 2020, I spoke at INFORMS annual meeting – Nonsmooth and Nonconvex Optimization on the work of Finite Difference Neural Networks. Video Presentation: link-to-Youtube
Training DNN for Molecular Systems – Challenges in Optimization.
On November 18, 2020, I spoke at AIChE Annual Meeting about the work – Improved Training on Improved Training of Graph-Embedding Based Neural Network Energy Functions for Catalysis. Video Presentation: link-to-Youtube Co-Authors: Prof. Srinivas Rangarajan, Prof. Martin Takac. Abstract: Complex reaction networks comprising of thousands of intermediates and reactions are ubiquitous in…