I am an Assistant Professor at the Department of Electronics & Electrical Communication Engineering, IIT Kharagpur, India. Prior to this, I had a short stint as an Assistant Professor (Lecturer) in ML & AI at the Department of Computer Science, University of Bath, UK (Sep. 2022 to Apr. 2023).
I work in the area of deep learning and inverse problems in imaging. My research objective is to develop novel, computationally efficient, and provably convergent algorithms for challenging image analysis problems by combining model- and data-driven approaches.
I was a postdoc with Prof. Carola-Bibiane Schönlieb at the Cambridge Image Analysis (CIA) group, DAMTP, University of Cambridge, where I was a member of the AI and image analysis team for the all-in-one cancer imaging project conducted by a group of mathematicians, computer scientists, and radiologists at the University of Cambridge.
Before joining Cambridge, I worked with Prof. Ozan Öktem as a postdoc in the imaging group at the Dept. of Mathematics, KTH, Sweden. I completed my Ph.D. in September 2018, at the Indian Institute of Science (IISc.), Bangalore, under the supervision of Prof. Chandra Sekhar Seelamantula.
Visit my Google Scholar profile to see my complete list of published articles. My detailed CV can be found here.
Prospective students: I am looking for students with a strong mathematical background and programming skills for both Ph.D. and master's projects. If you are a student at IIT Kharagpur and would like to work with me, drop me an email for further discussion.
Prospective collaborators: I have been very fortunate to have the opportunity to work with outstanding students and researchers from varied academic backgrounds, especially during and after my postdoc. If you are broadly interested in the theory and applications of machine learning in the context of imaging (and beyond, when it comes to theory) and believe that we can mutually benefit from a collaboration, please feel free to drop me an email.
Timeline
25 March 2024: We have an accepted paper on "Convergent regularization in inverse problems and linear plug-and-play denoisers" in Foundations of Computational Mathematics (FoCM).
23 December 2023: Paper on Langevin Monte Carlo with flow-based priors (with convergence analysis) for posterior sampling in Bayesian inverse problems accepted to the SIAM Journal on Imaging Sciences.
27 November 2023: Paper on provably convergent plug-and-play quasi-Newton methods accepted to the SIAM Journal on Imaging Sciences.
22 October 2023: Paper accepted (for oral presentation) at the NeurIPS-2023 Workshop on Deep Learning and Inverse Problems.
19 October 2023: Paper accepted in the IEEE Transactions on Medical Imaging.
5 & 7 September 2023: Two invited mini-symposium talks on adversarial regularization and plug-and-play methods at the 11th Applied Inverse Problems (AIP) Conference.
19 August 2023: New preprint on Bilevel learning with inexact line-search and dynamic selection of accuracy, with provable convergence to a stationary point.
9 August 2023: New preprint on learned mirror descent with stochasticity, acceleration, and equivariance is on arxiv.
18 July 2023: New preprint on provably convergent plug-and-play regularizers is available on arxiv.
12 May 2023: Joined the Department of E&ECE at IIT Kharagpur as an Assistant Professor (Grade-I).
9 May 2023: New preprint on Provably Convergent Plug-and-Play Quasi-Newton Methods is on arxiv.
4 April 2023: Invited talk at the British Applied Math Colloquium (BAMC) @ the mini-symposium on "Advances in Applied Numerical Linear Algebra and its applications".
30 March 2023: Invited talk on data-driven optimization at the Math Foundations of AI seminar, QMUL.
1 March 2023: New paper on data-driven mirror descent accepted to the SIAM Journal on Mathematics of Data Science. This is joint work with Hong Ye Tan (primary author), Dr. Junqi Tang, and Prof. Carola Schönlieb.
3 February 2023: Invited talk at the RNTW01 workshop, Isaac Newton Institute, University of Cambridge.
24-26 January 2023: Presented an invited poster at the ICMS workshop in Edinburgh.
2 December 2022: Invited talk on "Data-Driven Mirror Descent with Input-Convex Neural Networks" at the Numerical Analysis seminar series at Bath.
2 November 2022: Gave a talk on Data-driven methods for imaging inverse problems: algorithms and theoretical guarantees in the bi-weekly seminar series at the University of Bath Centre for Mathematics and Algorithms for Data.
27 September 2022: Invited talk on data-driven regularization for imaging inverse problems: Algorithms and guarantees at the SIAM conference on maths for data science (MDS22).
13 September 2022: Invited talk on deep learning for image reconstruction in X-ray CT at the workshop on advanced image reconstruction methods at UCL, UK.
1 September 2022: Joined the Department of Computer Science, University of Bath, as an Assistant Professor (Lecturer) in ML and AI.
27 August 2022: We have an accepted paper on Learned image reconstruction methods with theoretical guarantees in the IEEE Signal Processing Magazine.
26 July 2022: Invited talk on deep learning for image reconstruction in X-ray CT at the CMIH Academic Engagement Event, University of Cambridge.
14 June 2022: New preprint on learned mirror-descent using input-convex neural networks.
11 June 2022: New preprint out on provable data-driven methods for imaging inverse problems.
26 May 2022: Invited talk on data-driven adversarial regularization at the mini-symposium on Inverse Problems with Data-Driven Methods and Deep Learning at IPMS-2022.
16 April 2022: Invited talk on Machine learning for inverse problems in imaging at the Winter School of AI and Robotics, IIT Kharagpur.
21-25 March 2022: Co-organized a mini-symposium on the Provable Properties of Learned Reconstruction Approaches at SIAM IS22.
21 March 2022: Presented our work on adversarial regularizers at the SIAM IS22 mini-symposium on Deeply Learned Regularization for Inverse Imaging Problems.
7 March 2022: New paper entitled INSIDEnet: Interpretable NonexpanSIve Data-Efficient network for denoising in grating interferometry breast CT accepted to Medical Physics.
28 February - 4 March 2022: Co-organized the LMS invited lecture series on the mathematics of deep learning.
22 October 2021: Our new paper on learning convex regularizers satisfying the variational source condition is accepted to the NeurIPS-2021 Workshop on Deep Learning and Inverse Problems. preprint
11 October 2021: Our new paper on styleGAN inversion is now on arxiv.
28 September 2021: Our paper on unrolled adversarial regularization, which combines iterative unrolling with data-driven regularization, is accepted to NeurIPS-2021.
8 June 2021: Our recent paper on combining end-to-end reconstructions with data-driven regularization is now available: paper, code
30 March 2021: Our recent work on Adversarially learned iterative reconstruction for imaging inverse problems is now on arXiv. Find the python code that reproduces the results reported in the paper.
11 March 2021: Gave a talk on data-driven regularization for inverse problems (with Prof. Carola-Bibiane Schönlieb) at the Oberwolfach Workshop on deep learning and inverse problems.
9 March 2021: Gave a talk titled An introduction to deep learning for inverse problems (with Dr. Christian Etmann) at pre-GAMM 2021.
14 August 2020: Presented our recent work on unsupervised deep learning for inverse problems at MCQMC-2020 organized by the University of Oxford (video).
6 August 2020: Our recent work on data-driven convex regularization is now on arXiv (preprint). A PyTorch-based implementation of the algorithm is available here.
23 July 2020: Presented my Ph.D. work in SPCOM-2020 (video).
26 August to 30 September 2019: Visiting researcher at the Cambridge Image Analysis group.
26 September 2018: Defended Ph.D. thesis (supervisor: Prof. Chandra Sekhar Seelamantula, Dept. of Electrical Engineering, Indian Institute of Science (IISc.), Bangalore). Prof. Vikram M. Gadre (Dept. of Electrical Engineering, IIT-Bombay) and Prof. Animesh Kumar (Dept. of Electrical Engineering, IIT-Bombay) were my thesis examiners.