My Projects
GitHub repositories that I've built.
Implimentation of basic RL algorithms in PyTorch
Re implimentation of the paper - Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization
CNN implementation in Tensorflow for CIFAR-10 image classification.
We build a visual search system and evaluate it. SIFT descriptors as image features and vocabulary trees as the database structures.
Introduction to ML. Implementation of decision trees , SVM , Bayes Classifiers, Bagging-Boosting, Ensembles.
Implementation of a bot that plays a game of Duckhunt, TicTacToe , TicTacToe3d , Checkers. Ant Colony optimization and genetic algorithms for search problem dealing with uncertainty and limited resources.
Implementation of regression , non parametric regression , EM on Phylogenetic tree ,VI for earthquake epicenter detection , DGM and DP for Machine Learning. All implementations require derivations ( in the report)
Implementation of a k layer NN for classification of CIFAR10 dataset, Convolution Neural Network for surname ethnicity classification , RNN for text generation without DL libraries.
A deep convolution network for classification of chest X-ray images. Uses VGG16 as the base for transfer learning. Data Augmentation in the form of lung segmentation by U-Net network. Visual explanations using GradCAM.
Audio feature extraction , Hidden Markov Models with Gaussian Emissions for ASR , Phoneme Recognition with Deep Neural Networks
Image Feature detection using SIFT , SURF. Feature matching and performance evaluation with change in Rotation and Scale of image.
A Seq2Seq LSTM Encoder-Decoder framework that borrows the methodology of skipgram and continuous bag-of-words from Word2vec to learn contextual word embedding directly from speech data.
Parallel implementation of Mandelbrot, Gauss Seidal iterations, Bitonic sort and Odd Even sort using MPI in C.
Implementations of PGM in ML - Bayes Net, VI and Seq MCMC , SLAM, MCMC, Message Passing , LDA for topic clustering , VI for GMM
Implementation of dimensionality reduction technique called Gaussian Process Latent Variable Model and comparison with PCA, kernel PCA, MDS, t SNE on Oilflow,Vowels ,HAR UCI ,Wine UCI ,USPS Digits datasets.
My Interests
Topics that I find interesting and want to learn more about.
Machine Learning
Artificial Intelligence
Reinforcement Learning