Generative AI MathPrompter++ MathPrompter++ improves the reasoning capabilities of LLMs on numerical problems with robustness by embedding zero-shot CoT in a former Microsoft Research paper - MathPrompter ICLR 2023. The method improves the accuracy while reducing the hallucinate rate. Furthermore, I compiled a new dataset DiverseMath that serves as a better benchmark for reasoning in numerical problems. Stable Diffusion Models A brief introduction to image synthesis applications, stable diffusion models and its applications. Image Colorization An image colorization application that converts black and white images to colored ones using pix2pix model. Also includes interactive digit recognition webapp. 3D Reconstruction Perception for Home Robot Experimenting with various NeRF and Gaussian Splatting based SLAM algorithms to build a real-time 3D reconstruction of the environment. The methods used include NeRF-SLAM, Nice SLAM, Mip-NeRF and Splatam. Inverse Rendering with 2D Gaussian Splatting Developed a novel framework in CUDA for inverse rendering using 2D Gaussian splatting. Achieved state-of-the-art results gaining 15% lower normal map error, and obtained more consistent relighting results. Stochastic Primal-Dual Network for 3D Tomographic reconstruction Developed a stochastic version of primal-dual algorithm for reconstructing image volumes from CT scans. Implemented a python framework with custom Tensorflow layers and a library using Astra to aid the project. Statistical Image Processing Low Rank Matrix Recovery A in-depth analysis of problem formulations for low-rank matrix recovery such as Robust Principal Component Pursuit, and algorithms such as ALM and Dual Optimization. Extended the approach to applications such as Foreground-Background separation in videos. Saturation Noise in Compressed Sensing An in-depth analysis of compressive signal recovery methods in the presence of Gaussian noise and Saturation effects. Conducted experiments on synthetic signals, images and audio. Also derived a detailed proof for performance guarantees of the novel Likelihood Maximization method. Permutation Noise in Compressed Sensing Analysis of permutation noise in linear measurement applications. Has a prominent application in group testing for Covid 19. Proposed a hypothesis testing based correction algorithm along with theory and experiments. Self/Research VSCode Journal Plugin VS Code all-in-one journal plugin, combining automated TODO extraction, daily notes, task syncing, file tagging, and natural language support for seamless task management and note-taking Object Detection using YOLOv8 An exploration of different libraries such as `ultralytics` and `fiftyone` to train YOLO on COCO for object detection to understand ML pipelines. Recipe Rating Prediction A project with data exploration, feature engineering, recommendation models based on Random Forest and Logistic Regression to predict recipe ratings, addressing challenges like data imbalance and sparse interactions. Instance Segmentation using U-Net Developed a U-Net for instance segmentation using custom dataloaders and pre-processing techniques on PyTorch on the COCO dataset. Virtual Keyboard Quantum Money Motivation for Quantum Money. Description of different quantum money schemes - secret and public key based. Detailed analysis of various adaptive attacks on secret-key schemes and briefly touches upon the soundness of public-key schemes.
