Robust Estimation of Tree Structured Ising Models
Ashish Katiyar, Vatsal Shah, and Constantine Caramanis. "Robust Estimation of Tree Structured Ising Models." arXiv preprint arXiv:2006.05601 (2020).
CV
Education
- Ph.D in Electrical Engineering, University of Texas at Austin, 2017-present, Supervisor: Prof. Constantine Caramanis
- MS in Electrical Engineering, Texas A&M University, College Station, 2015-2017, Supervisor: Prof. Aniruddha Datta
- B. Tech in Electrical Engineering, Indian Institute of Technology, Jodhpur, 2008-2012
Major Academic Projects
- Robust Estimation of Graphical Models, January 2018-Present
- Proving limited unidentifiability of learning tree structured Ising Models and Gaussian Graphical Models from noisy samples.
- Developing an algorithm to recover the equivalence class of trees with O(logn) noisy samples in O(n^3) running time.
- Thompson Sampling with Bounded Means, October 2019-January 2020
- Developing a variant of Thompson Sampling algorithm for Bernoulli Bandits with knowledge of non-trivial mean bounds.
- Learning Communication Strategy for Two User Interference Channel, October 2018-December 2018
- Developing an anti autoencoder feedforward neural network to learn efficient constellation for communication over an interference channel.
- A Bayesian Approach to determine the Composition of Heterogeneous Cancer Tissue, January 2016-May 2017
- Developed an algorithm to enable low cost, scalable estimation of the composition of heterogeneous cancer tissue using a bayesian estimate of the number of cells from different cancer types.
- Digital Radio Mondiale (DRM/DRM+) Transmitter Development
- A team of 6 members implemented DRM radio standard on LabVIEW and generated final RF signal using NI PXIe-5673 RF vector signal generator.
- Supervisor: Prof. Sandeep Yadav
Industry experience
- Summer 2020: Facebook - Research Intern
- Developing a scalable labeling algorithm robust to human errors in the human labeling framework.
- Summer 2019: InterDigital - AI Researcher
- Developing efficient uncertainty estimation algorithm for fully connected Bayesian neural network.
- Using it for active learning on MNIST data to achieve nearly optimal accuracy using only 10% samples.
- 2012-2015: Defence Avionics Research Establishment, India - Scientist ‘B’
- Developing Drivers and Test Software.
Academic Internship
- Summer 2011:
- University: Indian Institute of Technology, Kanpur, India
- Error Correction Algorithm development for Wireless Sensor Networks.
- Supervisor: Prof. Aditya K. Jagannatham
- Summer 2010:
- University: University of Technology of Troyes, France
- GUI development to interface temperature and pressure sensor.
- Supervisor: Prof. David Hewson
Skills
- Python
- MATLAB
- C/C++
Publications
Robust Estimation of Tree Structured Gaussian Graphical Models
Ashish Katiyar, Jessica Hoffmann, Constantine Caramanis, "Robust Estimation of Tree Structured Gaussian Graphical Models" Proceedings of the 36th International Conference on Machine Learning, PMLR 97:3292-3300, 2019
A Bayesian approach to determine the composition of heterogeneous cancer tissue
Ashish Katiyar, Anwoy Mohanty, Jianping Hua, Sima Chao, Rosana Lopes, Aniruddha Datta and Michael L. Bittner(2018). "A Bayesian approach to determine the composition of heterogeneous cancer tissue" BMC Bioinformatics. 2018;19(Suppl 3)1
Bayesian data and channel joint maximum-likelihood based error correction in wireless sensor networks
Ashish Katiyar, and Aditya K. Jagannatham (2011) "Bayesian data and channel joint maximum-likelihood based error correction in wireless sensor networks" Proceedings of the 1st International Conference on Wireless Technologies for Humanitarian Relief pp 141-146
Honors
- General Proficiency Medal
- Awarded by The Hon. President of India for best academic performance in Electrical Engineering.
- B.Tech Project Medal
- Awarded by The Hon. President of India for Best B.Tech Project work among all the disciplines in the institute.