|
Mangal Prakash Senior Scientist, AI/ML Johnson & Johnson (Janssen R&D) High Wycombe, UK mangalcrj@gmail.com [GitHub] [Twitter] [Google Scholar] [Resume/CV] |
My research interests span across computer vision and machine learning with particular emphasis on deep learning, graphical models, and deep probabilistic generative models. I am a Senior Scientist, AI/ML at Johnson & Johnson (Jansen R&D) in High Wycombe, UK working on building novel Machine Learning tools for drug discovery. Prior to this, I was a AI Research Scientist at Exscientia in Oxford, UK where I lead a small team in the area of Molecular Property Prediction. I was a PhD candidate at TU Dresden/Max-Planck Institute of Molecular Cell Biology and Genetics, advised by Dr. Florian Jug. In my PhD, I developed methods and software using deep learning and graphical models for image denoising, object segmentation and tracking in biomedical images. I obtained MSc degree in Electrical Engineering from University of Minnesota, Twin Cities with specialization in distributed control systems and its application to smart grids. I completed my B.Tech degree in Electrical Engineering from National Institute of Technology, Durgapur, India.
News
[Feb 2023] My collaborative paper Afadin and zyxin contribute to coupling between cell junctions and contractile actomyosin networks during apical constriction is accepted at PLOS Genetics.[March 2022] My collaborative paper Extracellular mechanical forces drive endocardial cell volume decrease during zebrafish cardiac valve morphogenesis is accepted at Developmental Cell.
[Jan 2022] My paper Interpretable Unsupervised Diversity Denoising and Artefact Removal, in collaboration with Google Research, is accepted as a Spotlight presentation at ICLR 2022.
[Jan 2022] My collaborative paper Extracellular mechanical forces drive endocardial cell volume decrease during cardiac valve morphogenesis, is under review at Developmental Cell.
[Sep 2021] My collaborative paper Tissue scale properties of the extracellular matrix regulates nuclear shape, organisation and fate in the embryonic midline sutures, is now available on bioRrxiv.
[April 2021] My paper Removing Pixel Noises and Spatial Artefacts with Generative Diversity Denoising Methods, in collaboration with Google Research, is now available on arXiv.
[Jan 2021] My paper Fully Unsupervised Diversity Denoising with Convolutional Variational Autoencoders has been accepted at ICLR 2021.
[Oct 2020] My collaborative paper Regionalized tissue fluidization is required for epithelial gap closure during insect gastrulation has been published in Nature Communications.
[Aug 2020] My paper DenoiSeg: Joint Denoising and Segmentation has been selected for oral presentation at ECCV 2020 workshop on BioImage Computing.
[July 2020] I jointly taught a course on content aware image restoration hosted by NEUBIAS Academy.
[June 2020] My latest work DivNoising: Diversity Denoising with Fully Convolutional Variational Autoencoders is available on arXiv.
[May 2020] My paper A Primal-Dual Solver for Large-Scale Tracking-by-Assignment has been published at AISTATS 2020.
[Jan 2020] My paper Fully Unsupervised Probabilistic Noise2Void has been selected for oral presentation at ISBI 2020.
[Jan 2020] My paper Leveraging Self-Supervised Denoising for Image Segmentation has been selected for poster presentation at ISBI 2020.
[Jan 2020] I gave a talk about our work "Leveraging Self-Supervised Denoising for Image Segmentation" at Quantitative Bio Imaging Conference at University of Oxford, UK. Check the project page below for slides.
Conference Publications
|
Interpretable Unsupervised Diversity Denoising and Artefact Removal Mangal Prakash, Mauricio Delbracio, Peyman Milanfar, Florian Jug International Conference on Learning Representations (ICLR), 2022 (Spotlight). [Paper] [GitHub] |
|
Fully Unsupervised Diversity Denoising with Convolutional Variational Autoencoders Mangal Prakash, Alexander Krull, Florian Jug International Conference on Learning Representations (ICLR), 2021. [Paper] [GitHub] |
|
A Primal-Dual Solver for Large-Scale Tracking-by-Assignment Stefan Haller, Mangal Prakash, Lisa Hutschenreiter, Tobias Pietzsch, Carsten Rother, Florian Jug, Paul Swoboda, Bogdan Savchynskyy Proceedings of the 24th International Conference on Artificial Intelligence and Statistics (AISTATS) 2020. [Paper] [Webpage] [GitHub] [Talk] |
|
DenoiSeg: Joint Denoising and Segmentation Mangal Prakash, Tim-Oliver Buchholz, Deborah Schmidt, Alexander Krull, Florian Jug BioImage Computing Workshop, European Conference on Computer Vision (ECCV) 2020 (Oral). [Paper] [Webpage] [GitHub] [Talk] |
|
Fully Unsupervised Probabilistic Noise2Void Mangal Prakash, Manan Lalit, Pavel Tomancak, Alexander Krull, Florian Jug IEEE International Symposium on Biomedical Imaging (ISBI) 2020 (Oral). [Paper] [Webpage] [GitHub] [Slides] |
|
Leveraging Self-Supervised Denoising for Image Segmentation Mangal Prakash, Tim-Oliver Buccholz, Manan Lalit, Pavel Tomancak, Florian Jug, Alexander Krull IEEE International Symposium on Biomedical Imaging (ISBI) 2020. [Paper] [GitHub] [Talk] |
|
Probabilistic Noise2Void: Unsupervised Content-Aware Denoising Alexander Krull, Tomas Vicar, Mangal Prakash, Manan Lalit, Florian Jug In Frontiers of Computer Science, 2020. [Paper] [GitHub] |
|
Distributed Stopping Criterion for Ratio Consensus Mangal Prakash, Saurav Talukdar, Sandeep Attree, Sourav Patel, Murti V. Salapaka In Allerton Conference on Communications, Control and Computing, 2018 (Oral). [Paper] |
|
|
Distributed apportioning in a power network for providing demand response services Sourav Patel, Sandeep Attree, Saurav Talukdar, Mangal Prakash, Murti V. Salapaka In IEEE International Conference on Smart Grid Communications, 2017 (Oral). [Paper] |
|
|
Reconstruction of networks of cyclostationary processes Saurav Talukdar, Mangal Prakash, Donatello Materassi, Murti V. Salapaka In IEEE Decision and Control, 2015 (Oral). [Paper] |
|
Journal Publications and Preprints
|
Afadin and zyxin contribute to coupling between cell junctions and contractile actomyosin networks during apical constriction Mark Maddock Slabodnick, Sophia C Tintori, Mangal Prakash, Christopher D Higgins, Alicia H Chen, Timothy D Cupp, Terrence Wong, Emily Bowie, Florian Jug, Bob Goldstein PLOS Genetics, 2023. [Paper] |
|
|
Extracellular mechanical forces drive endocardial cell volume decrease during cardiac valve morphogenesis Helene Vignes, Christina Vagena-Pantoula, Mangal Prakash, Caren Norden, Florian Jug, Julien Vermot Developmental Cell, 2022. [Paper] |
|
|
Tissue scale properties of the extracellular matrix regulates nuclear shape, organisation and fate in the embryonic midline sutures Diana Alves Afonso, Allyson Quinn Ryan, Adrian Lahola-Chomiak, Mangal Prakash, Florian Jug, Carl D. Modes, Jacqueline M. Tabler BioRxiv, 2021. [Paper] |
|
|
Regionalized tissue fluidization by an actomyosin cable is required for epithelial gap closure during insect gastrulation Akanksha Jain, Vladimir Ulman, Argo Mukherjee, Mangal Prakash, Lokesh Pimpale, Stefan Munster, Robert Hasse, Kristen A. Panfilio, Florian Jug, Stefan W. Grill, Pavel Tomancak, Anastasios Pavlopoulos Nature Communications, 2020. [Paper] |
|
|
Distributed Stopping Criterion for Consensus in the Presence of Delays Mangal Prakash, Saurav Talukdar, Sandeep Attree, Vikas Yadav, Murti V. Salapaka In IEEE Transactions on Control of Network Systems, 2019. [Paper] |
|
Awards
University of Minnesota College of Science and Engineering Fellowship 2014Teaching
Dresden Deep Learning Hackathon d3hack2019, TU DresdenTeam mentor
9th Sep to 13th Sep 2019
Signals and Systems (EE 3015), University of Minnesota
Graduate Teaching Assistant (TA) with Prof. Paul Imbertson
Fall 2015
State Space Control System Design (EE 4233), University of Minnesota
Graduate Teaching Assistant (TA) with Prof. Andrew Lamperski
Spring 2016