Success is no accident. It is hard work, perseverance, learning, studying, sacrifice and most of all, love of what you are doing or learning to do.
Projects and Experiments:
Active SLAM and Path Planning for Mobile Robot Navigation
Active SLAM and Path Planning for Mobile Robot Navigation
Built an agent to explore and map a partially mapped indoor robot environment. Used Active SLAM approach for goal state selection in a real-world environment. Incorporated recovery mechanisms and a finetuning parameter for better accuracy and reduced computational time.
Reinforcement Learning - Proximal Policy Optimization Algorithm
Reinforcement Learning - Proximal Policy Optimization Algorithm
Built an agent that learns to play Atari’s Pacman game using Reinforcement learning. Used OpenAI gym environment and its observation/reward model to compare PPO algorithm with different Reinforcement Learning algorithms. Introduced new strategies to improve the performance such as, Max-pooling, RandomWalk, etc. to help learn faster.
Stock Market Prediction using Sentiment Analysis of Tweets
Stock Market Prediction using Sentiment Analysis of Tweets
Implemented web crawlers to extract Tweet data and Stock Market Price. NLP techniques like Tokenization, Stop-wordremoval, Stemming are used to perform language modeling and Sentiment Analysis on Twitter data. Compared predictions of stock price fluctuations from different Machine learning techniques such as SVM, Regressions, Deep Networks, etc. The ML models predicted the prices most accurately when Topic Modeling was used. Considering past 5 days of tweet sentiments, we were able to predict change in stock prices with close to 70% accuracy.