Projects
Computational Biology Research
Conducted a study on how to remove noise from genomics data with the help of unsupervised learning. Achieved 22.31% reduction in noise by implementing denoising autoencoder for noisy genomics data.
Yoga Pose Estimation and Training System
Implemented an interactive end-to-end Yoga training system with the help of Computer Vision and Deep Learning, achieving over 90% accuracy in pose estimation. Utilized MoveNet and custom Deep Neural Network model, processing webcam live feed for dynamic feedback on yoga postures.
Python package to implement TOPSIS
Implemented a python package to employ Technique for Order Preference by Similarity to Ideal Solution or TOPSIS. Generates an output table containing Correlation, Square of Correlation, Root Mean Square Value, Accuracy, TOPSIS score, and Rank with provided input of weights, and a csv dataset.
Training a machine to play the classic snake game
Developed a custom built AI to play the snake game and produced a dataset from scratch consisting of snake moves with 50,000 entries. Further, trained ANN, SVM and Random Forest on the generated dataset and conducted an ablation study on performance of ML models.
Cash Control Android App
Developed ‘Cash Control,’ an Android-based money-management app for tracking both online and offline expenses. Integrated geofencing to prompt users to log expenses upon reaching a ‘home location’ and used Theil-Sen regression for predictive analysis of future spending by category
