Agnimitra Dasgupta

Social:      

Jump to: Interests, Experience, Talks, Publications, Conferences, Teaching, Mentoring, Education, Awards

Research Interests

Uncertainty quantification, scientific machine learning, inverse problems, reliability and rare-events estimation, multifidelity methods, generative modeling

Research Experience

2023-now Postdoctoral Research Associate, Department of Aerospace and Mechanical Engineering
University of Southern California
Advisor: Prof. Assad A. Oberai
2017-2023 Research Assistant, Department of Civil Engineering
University of Southern California
Advisor: Prof. Erik A. Johnson
2020, 2021 Givens Associate, Mathematics & Computer Science Division
Argonne National Laboratory
Mentor: Dr. Zichao 'Wendy' Di
2015-2017 Research Assistant, Department of Civil Engineering
Indian Institute of Science
Advisor: Prof. Debraj Ghosh

Publications

A. Dasgupta, D. V. Patel, D. Ray, E. A. Johnson and A. A. Oberai, A hybrid dimension-reduced and variational framework for solving high-dimensional inverse problems, Computer Methods in Applied Mechanics & Engineering, 2023. (accepted) [arxiv]
A. Dasgupta and E. A. Johnson, REIN: Reliability estimation via importance sampling with normalizing flows, Reliability Engineering & System Safety, 2023. [link]
D. Ray, J. Murgoitio-Esandi, A. Dasgupta and A. A. Oberai, Solution of physics-based inverse problems using conditional generative adversarial networks with full gradient penalty, Computer Methods in Applied Mechanics and Engineering, 2023. (in press) [link]
A. Dasgupta and E. A. Johnson, Model falsification from a Bayesian viewpoint with applications to parameter inference and model selection of dynamical systems, ASCE Journal of Engineering Mechanics, 2023. (in press) [doi]
A. Dasgupta and D. Ghosh, Failure Probability Estimation of Linear Time Varying Systems by Progressive Refinement of Reduced Order Models, SIAM/ASA Journal on Uncertainty Quantification, Volume 7, Issue 3, Pages 1007–1028, 2019. [link]
In preparation
A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, A new perspective on model falsification as a binary classifier, in preparation
A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, Hybrid Bayesian approaches for the inference of spatially varying constitutive parameters of linear isotropic materials from noisy response data, in preparation

Conferences (with reviewed papers)

A. Dasgupta, J. Murgoitio-Esandi, H. Ramaswamy, D. Ray and A. A. Oberai, Conditional score-based generative models for solving physics-based inverse problems, Workshop on Deep Learning and Inverse Problems, 37th Annual Conference on Neural Information Processing Systems (NeuRIPS), New Orleans, LA, December 2023. (upcoming) [workshop] [paper]
A. Dasgupta and E. A. Johnson, REIN: Rare event simulation via importance sampling with normalizing flows, ASCE Engineering Mechanics Institute Conference, Atlanta, GA, June 2023. Awarded Best Student Competition Paper by the EMI Probabilistic Methods Committee.
E. A. Johnson and A. Dasgupta, Learning importance sampling distributions via normalizing flows for estimating rare-event failure probabilities, 2023 New Zealand Society for Earthquake Engineering Conference, University of Auckland, April 2023. [paper]
A. Dasgupta, C. Graziani and Z. W. Di, Simultaneous reconstruction and uncertainty quantification for tomography, 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Greece, June 2023. [paper] [arxiv]
A. Dasgupta, D. V. Patel, D. Ray, E. A. Johnson and A. A. Oberai, GAN-Flow: A dimension-reduced variation framework for physics-based inverse problems, Workshop on Machine Learning and the Physical Sciences, 36th Annual Conference on Neural Information Processing Systems (NeuRIPS), New Orleans, LA, December 2022. [workshop] [paper] [poster]
A. Dasgupta and E. A. Johnson, Model falsification from a Bayesian viewpoint with applications to system indentification and model selection, 8th World Conference on Structural Control and Monitoring, UCF, Orlando, FL, June 2022.
A. Dasgupta and Z. W. Di, Uncertainty quantification for ptychography using normalizing flows, Machine Learning and the Physical Sciences, Workshop at the 35th Annual Conference on Neural Information Processing Systems (NeurIPS), virtual event, 2021. [workshop] [paper] [poster]

Conferences (with reviewed abstracts)

A. Dasgupta, Solving large-scale physics-based inverse problems using coupled generative adversarial networks and normalizing flows, SoCal Solids Conference, USC, Los Angeles, CA, August 2023.
A. Dasgupta and E. A. Johnson, Rare-events simulation using normalizing flows, ASCE Engineering Mechanics Institute Conference, GaTech, Atlanta, GA, June 2023.
A. Dasgupta and E. A. Johnson, Normalizing flows based importance sampling for estimating rare event probabilities, SIAM Conference on Mathematics of Data Science, San Diego, CA, September 2022.
A. Dasgupta and E. A. Johnson, An approximate Bayesian perspective on model falsification with applications to parameter inference and model selection of dynamical systems, ASCE Engineering Mechanics Institute Conference, JHU, Baltimore, MD, June 2022.
A. Dasgupta and Z. W. Di, Ptychographic Inversion and Uncertainty Quantification Using Invertible Neural Networks SIAM Conference on Uncertainty Quantification, Atlanta, GA, April 2022. [slides]
A. Kannan, A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, Investigation of bounds on sample size for reliable model falsification, SIAM Conference on Computational Science and Engineering, virtual event, March 2021.
A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, Deep learning of surrogate models from un-falsified physics, SIAM Conference on Computational Science and Engineering, virtual event, March 2021.
A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, Characterization of spatial heterogeneity in material properties using a probabilistic hybrid approach, ASCE Engineering Mechanics Institute Conference, Caltech, Pasadena, CA, June 2019.
A. Dasgupta, S. De, K. Teferra, E. A. Johnson, S. F. Wojtkiewicz and L. Graham-Brady, Probabilistic validation of material models, ASCE Engineering Mechanics Institute Conference, MIT, Cambridge, MA, May 2018.
S. De, T. Yu, A. Dasgupta, E. A. Johnson and S. F. Wojtkiewicz, Probabilistic model validation of a full-scale four-story base-isolated building, ASCE Engineering Mechanics Institute Conference, MIT, Cambridge, MA, May 2018.
A. Dasgupta, S. De, E. A. Johnson and S. F. Wojtkiewicz, Probabilistic model validation of large-scale systems using reduced order models, SIAM Conference on Uncertainty Quantification, Orange County, CA, April 2018.
A. Dasgupta and D. Ghosh, Progressively refining reduced order models for estimating failure probabilities of dynamical systems, SIAM Conference on Uncertainty Quantification, Orange County, CA, April 2018

Talks & Tutorials

Score-based diffusion generative models: Implementation and results, Uncertainty Quantification Summer School, University of Southern California, August, 2023. [github]
Forward and inverse uncertainty quantification using generative models, Virtual Seminar, Johns Hopkins University, July, 2022. Invited by Prof. Michael Shields and Prof. Lori Graham-Brady.
A hybrid probabilistic approach to characterizing material heterogeneity, USC Women in Science & Engineering (WiSE) STEM Bytes Seminar, June, 2021.

Teaching Experience

Guest Lecturer, University of Southern California, 2023
  • AME 508 Machine Learning and Computational Physics, Fall 2023

    • CE529 Finite Element Analysis,   Fall 2017, 2019, 2021
  • CE458 Computational Strcutural Analysis,   Spring 2018, 2022
  • CE402 Computer Methods in Engineering,   Spring 2019

    • Mentoring Experience

    USC Center for Undergraduate Research in Viterbi Engineering (CURVE), Fall 2023 - Spring 2024
    (co-mentored with other graduate students)
  • Serena Li (B.S. 2026, expected), Renal cell carcinoma classification from contrast enhanced CT images using deep learning, co-mentored with Mehrnegar Aminy.
  • Joseph Liu (B.S. 2024, expected), Initializing wildfire forecasts using conditional diffusion models, co-mentored with Bryan Shaddy.
  • Haofeng (Ace) Xu (B.S. 2022, expected), Anomaly detection using auto-encoders, co-mentored with Javier Murgoitio-Esandi.

    • USC Viterbi CED Summer Institute (DEI initiative), Summer 2018
  • Mark Parent (B.S. 2022) and Nathan Medina, Base isolation mechanisms in earthquakes
  • Mayra Rodriguez and Emmanuel Salgado, Inverse problems: An application to estimation of material parameters

    • USC Viterbi Summer Research Programme, Summer 2020
  • Archana Kannan, Exploring required sample sizes for reliable model class falsification using likelihood bounds, Summer–Fall 2020. M.S., 2020.
  • Julia Woomer, Exploring required sample sizes for reliable model class falsification using error bounds, Summer-Fall 2020. B.S. ME, 2023 (expected)

    • Education

    Ph.D. Civil University of Southern California 2023
    M.S. Electrical University of Southern California 2020
    M.E. Civil Indian Institute of Science 2017
    B.E. Civil Jadavpur University 2015

    Honors & Awards (selected)

    • ASCE EMI Probabilistic Methods Committee Student Paper Competition, 2023.
    • Outstanding Research Assistant Award, USC Viterbi School of Engineering, 2023.
    • Outstanding Research Assistant Award, Sonny Astani Department of Civil and Environmental Engineering, USC, 2022.
    • Das Family Travel Award of $1000 to attend EMI, 2022.
    • SIAM Student Travel Award for attending SIAM UQ 2022 courtesy of NSF.
    • Provost's Ph.D. Fellowship, University of Southern California, 2017-2019.
    • Prof. N S Lakshmana Rao Medal, Civil Engineering, IISc, 2018.
    • 2-year scholarship for graduate study, MHRD, India, 2015-2017.
    • Gold Medalist (and other awards), Jadavpur University, 2015.
    • Silver Medal from N.C.E. Bengal, India, 2014.