Implement Human Activity Recognition in PyTorch using hybrid of LSTM, Bi-dir LSTM and Residual Network Models

Code for my Master's thesis on Multi‑Task Self‑Supervised Learning for label‑efficient learning. Modular PyTorch framework combining contrastive + pretext tasks with dynamic loss weighting, and centralized/federated training (HAR/STL‑10) to learn compact, robust representations.

This project is to build a model that predicts the human activities such as Walking, Walking Upstairs, Walking Downstairs, Sitting, Standing or Laying using readings from the sensors on a smartphone carried by the user.

The project I produced for the assignment for 'Course 3: Getting and Cleaning Data, of the Data Science Specialization from Johns Hopkins University on Coursera'.

Mini-Project Given during the Assignment 1 of ML Course of IITGN ES-335

High-performance Human Activity Recognition (HAR) pipeline in Julia using PCA for dimensionality reduction (82% compression) and an Optimized Ensemble Model (ANN+SVM+kNN), achieving 93.3% accuracy on unseen test subjects.