Krishnanunni C G

POB 3SEo2B

201 E 24th St

Austin, Texas 78712

I am a PhD scholar in the Department of Aerospace Engineering & Engineering Mechanics, UT Austin working at the interface of machine learning and mechanics. I am a member of the PHO-ICES group, Oden Institute for Computational Engineering and Sciences and work with my advisor Prof. Tan Bui-Thanh. I received a bachelors degree in Civil Engineering from National Institute of Technology, Calicut in 2017 and a masters degree in Structural Engineering from Indian Institute of Technology, Madras in 2020.

I come from a mechanics background where I undertook projects on signal processing, optimization, and mathematical modelling in solid mechanics with Prof. B. N. Rao at IIT Madras and Prof. Mohammed Ameen at NIT Calicut. I was fortunate to be awarded a research fellowship to work with Prof. Phoolan Prasad at the Indian Institute of Science, Bangalore who exposed me to rigorous mathematics behind nonlinear hyperbolic waves and in particular allowed me to appreciate the beauty of mathematics in mechanics. Currently, my broad interest lies in anything creative and mathematically beautiful.

Current Research Focus

Machine learning methods for science and engineering have gained significant popularity in recent years (Scientific Machine Learning). In contrast, my research takes the opposite direction: I draw inspiration from techniques in mechanics and numerical analysis to develop novel machine learning algorithms. Specifically, my PhD thesis addresses the problem of neural network architecture adaptation by leveraging concepts from topology optimization and adaptive mesh refinement strategies in the finite element method. This results in a principled and efficient alternative to conventional neural architecture search (NAS) approaches (More details available here).

Currently, I am interested in generative modeling for prior calibration in Bayesian inverse problems, as well as in developing derivative-free optimization methods for black-box optimization.

news

Jul 1, 2025	I will be coorganizing the 3rd annual workshop on Scientific Machine Learning hosted by the Oden Institute of Computational Engineering and Sciences, UT Austin (Sept. 25 and Sept. 26).
Jun 25, 2025	Our new paper “Lilan: A linear latent network approach for real-time solutions of stiff, nonlinear, ordinary differential equations” is available in arXiv.
Jun 1, 2025	I will be coorganizing the SIAM TXLA 2025 Sectional meeting hosted by the Oden Institute of Computational Engineering and Sciences, UT Austin.
May 25, 2025	Our new paper on a “topological derivative approach for deep neural network architecture adaptation” is available in arXiv.
Mar 15, 2025	Our new paper on “An adaptive and stability-promoting layerwise training approach for sparse deep neural network architecture” has been accepted for publication in Computer Methods in Applied Mechanics and Engineering.

selected publications

2025

arXiv

Topological derivative approach for deep neural network architecture adaptation

Krishnanunni C G, Tan Bui-Thanh, and Clint Dawson

arXiv preprint arXiv:2502.06885, 2025

Bib HTML

@article{krishnanunni2025topological,
  title = {Topological derivative approach for deep neural network architecture adaptation},
  author = {C G, Krishnanunni and Bui-Thanh, Tan and Dawson, Clint},
  journal = {arXiv preprint arXiv:2502.06885},
  year = {2025},
}

arXiv

Lilan: A linear latent network approach for real-time solutions of stiff, nonlinear, ordinary differential equations

William Cole Nockolds, Krishnanunni C G, and Tan Bui-Thanh

arXiv preprint arXiv:2501.08423, 2025

Bib HTML

@article{nockolds2025constant,
  title = {Lilan: A linear latent network approach for real-time solutions of stiff, nonlinear, ordinary differential equations},
  author = {Nockolds, William Cole and C G, Krishnanunni and Bui-Thanh, Tan},
  journal = {arXiv preprint arXiv:2501.08423},
  year = {2025},
}

2022

arXiv

Layerwise Sparsifying Training and Sequential Learning Strategy for Neural Architecture Adaptation

CG Krishnanunni, and Tan Bui-Thanh

arXiv preprint arXiv:2211.06860, 2022

HTML