Email: abhinavsagar4 at gmail dot com
Address: 64 Greens Radius Developers, Santacruz, Mumbai, India
Hi! I just completed my undergrad at VIT Vellore.
My research areas are bayesian deep learning, generative models, variational inference etc
on the theoretical side and medical imaging, autonomous driving etc on the application side.
More broadly, I am interested in deep learning and computer vision with a bayesian approach taking uncertainty into account.
The communities I follow are NeurIPS, ICLR, CVPR, ICCV, ECCV, MICCAI, MIDL, WACV and UAI.
I spent the summer of 2019 in Tessact where I worked on product recommendation using variational autoencoders.
I spent the summer of 2018 in Tata Group where I trained a neural network on power grid electricity consumption data to predict the load 24 hours ahead of the actual generation.
Besides research, I enjoy travelling, playing guitar and cooking (also eating).
If you would like to do a research collaboration, please send me an email.
[6th Jan 2020] Appointed as teaching assistant for CSE4020 (Machine Learning) with Professor Gayathri P.
[14th Sep 2019] Speaking on Automated Machine Learning at RMZ Millenia Business Park in Chennai, India.
[3rd Jun 2019] Attending the Nordic Probabilistic AI School in Trondheim, Norway with full travel grant.
[8th Apr 2019] Got selected for Computer Vision internship at Tessact in Mumbai, India.
[6th Oct 2018] Speaking on Ethics of Artificial Intelligence at Channa Reddy Auditorium in Vellore, India.
[15th May 2018] Got selected for Deep Learning internship at Tata Group in Jamshedpur, India.
Research Papers and Preprints
Generate High Resolution Images With Generative
Uncertainty Quantification using Variational
Inference for Biomedical Image Segmentation
Semantic Segmentation With Multi Scale Spatial
Attention For Self Driving Cars
Stochastic Bayesian Neural Networks
Bayesian Multi Scale Neural Network for Crowd
RUHSNet: 3D Object Detection Using Lidar Data in
HRVGAN: High Resolution Video Generation using
Monocular Depth Estimation Using Multi Scale
Neural Network And Feature Fusion
Generate Novel Molecules With Target Properties
Using Conditional Generative Models