This project experiments with the use of set representations like Zonotopes, Taylor Models, Starsets, Polynomial Zonotopes etc to model uncertainty in the position of obstacles and tries to incorporate this in path and motion planning by simulating it’s position at various points in time.

The aim of this project was to understand the working of Mirror Descent Policy Optimization by implementing it and experimenting with different models. The experimentation was done on on- and off- policy DDPG and SAC over different environments.

The aim of this project was to take a data science approach to the pandemic by trying to pinpoint index slices which appear to represent mutation invariant regions in the SARS-CoV2 gene. Initial experimentation was done with publicly available data from sources like Kaggle. After that, we took in real-life data from GISAID.

This project was an implementation project of the Research paper (named same as the title). The model was tested initially on MNIST dataset and then, publicly available map data.

The aim of this project was to implement the research paper named “PiTree” but to also experiment with the impact of using different decision trees. The project was primarily implemented in Python. The final results included the results of various decision trees on factors like QoE, buffer rate, latency etc.

The project focused on automating the optimization of various parameters of different ensemble and boosting methods.