Jointly Trained Image and Video Generation using Residual Vectors

Yatin Dandi, Aniket Das, Soumye Singhal, Vinay P. Namboodiri, Piyush Rai

2020 Winter Conference on Applications of Computer Vision (WACV '20)

Generalized Adversarially Learned Inference

Yatin Dandi, Homanga Bharadhwaj, Abhishek Kumar, Piyush Rai

AAAI Conference on Artificial Intelligence (AAAI-21), NeurIPS 2020 Workshop: Self-Supervised Learning - Theory and Practice

Work done under Prof. Piyush Rai, IIT Kanpur , and Abhishek Kumar, Google research.

Model-Agnostic Learning to Meta-Learn

Arnout Devos*, Yatin Dandi*

Pre-registration workshop, NeurIPS (2020). Full paper published in Proceedings of Machine Learning Research (PMLR).

Work done under Prof. Matthias Grossglauser at the INDY lab, EPFL.

Implicit Gradient Alignment in Distributed and Federated Learning

Yatin Dandi*, Luis Barba*, Martin Jaggi

AAAI Conference on Artificial Intelligence (AAAI-22), FL-ICML 2021 : International Workshop on Federated Learning for User Privacy and Data Confidentiality in Conjunction with ICML 2021.

Work done under Prof. Martin Jaggi at the MLO lab, EPFL.

NeurInt-Learning Interpolation by Neural ODEs (Spotlight)

Avinandan Bose*, Aniket Das*, Yatin Dandi, Piyush Rai

The Symbiosis of Deep Learning and Differential Equations: DLDE Workshop, NeurIPS 2021.

Work done under Prof. Piyush Rai at IIT Kanpur, India.

Data-heterogeneity-aware Mixing for Decentralized Learning

Yatin Dandi, Anastasia Koloskova, Martin Jaggi, Sebastian U. Stich

The Symbiosis of Deep Learning and Differential Equations: DLDE Workshop, NeurIPS 2021.

Work done under Prof. Martin Jaggi at the MLO lab, EPFL.

Other Projects

head over to my CV for details

Causal Inference, Out of Distribution Generalization and Meta Learning: A Unified Perspective

Research project under Prof. Matthias Grossglauser. Studied the theory of causal inference, surveyed recent works on causal inference and out of distribution generalization for several classes of machine learning models, analyzed the relationship between out of distribution generalization, meta-learning and causal inference, introduced a new paradigm of "out of task distribution generalization".

On the Effect of Noise induced by Gradient Stochasticity on Optimizing 2-player Differentiable Games

Research project under Tatjana Chavdarova . Analyzed the continuous time limit of stochastic variants of first order algorithms for differentiable games using the theory of Stochastic Differential Equations (SDEs), derived first order approximations for second order algorithms such as SGA (Symplectic Gradient Adjustment) for n-player games.

Disentangled Representation Learning using Generative Models

Research project under Prof. Piyush Rai. Studied various approaches for learning disentangled representations of sequential data such as using using new adversarial loss terms, factorized hierarchical priors and exploiting the probabilistic model and architecture of the LSTM based autoencoder to promote disentanglement, implemented a Variational Autoencoder model for disentangling of time invariant content and dynamics in sequential data (Mandt et al.) using Pytorch and experimented with modifications in the probabilistic model.

Deep Reinforcement Learning for Atari Games

Implemented Deep Q-Learning and Policy Gradient methods for Atari Games using PyTorch and OpenAI Gym along with various classical RL methods using Numpy such as Dynamic Programming (Policy and Value iteration), Monte Carlo (Epsilon-greedy and off-policy), TD Learning (Q-Learning and SARSA) and Q-Learning with Function Approximation. Presently studying state of the art variants of actor-critic methods.

Image Captioning with visual attention

Project under Programming Club, IIT Kanpur. Studied various encoder-decoder based architectures for image captioning and implemented the model described in Show, Attend and Tell (Xu et al.2015) using Tensorflow. Used MS COCO dataset for training and evaluation.

Computational Models for Inference of Social Dynamics

Research project under Prof. Nisheeth Srivastava. The aim is to study the effect of variation of multiple physical parameters in animated situations to determine the underlying cause of inference of emotion and social situations without language.