Albert's Projects
An Alexa skill which states facts in the history of AI. Built using the Alexa Skills Kit.
Classifying dog breeds in images using transfer learning and CNNs. Makes use of the pre-trained Inception-V3 model used in the ImageNet competition.
Using adversarial search and Minimax/Alpha Beta pruning strategies, one can create agents capable of creating competitive board games. Heuristics can be added to augment the strength of the agent and evaluate the value of a move under time constraints.
Implemented planning graphs to solve problems in logistics using search heuristics. Logical planners can be adapted to solve any type of problem given a notion of states, actions, constraints and goals.
Using Hidden Markov Models to translate gestures in American Sign Language to Text. Used BIC, DIC and cross validation to train optimal models for each word.
Using LSTM modules to generate predictions of Apple stock prices, and sentence generation based on Sherlock Holmes.
Using constraints and strategies to narrow down the number of options, one can solve hard Sudoku puzzles very easily.
Comparing the performance of RNN networks to convert audio files into phonetic translations.
Re-implementation of ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
Ascii image art generator with Rust
Reinforcement learning using PPO on the 3D Ball environment for Unity ML-Agents. Using MPI to do distributed training across separate processes
Audio classification of domestic cats sounds (hungry, angry, purring, etc) using raw waveforms. PyTorch implementation of M5 architecture.
Implementing the game of Go using OpenGL
Style transfer from brown tabby cats to tortoiseshell using predominantly U-Net based generators. Includes TF record serialization for images and some tensorboard logging.
This is a compilation of my solutions to the DailyProgrammer challenges. These challenges address a number of topics such as search, game playing and Mathematics.
Creating internet memes using Generative Adversarial Networks (GANs).
Diffusion model for MRI image generation
This is my personal website.
JAX/Flax implementation of U-2 Net for salient object detection.
Sandbox experiments for playing around with nanoGPT
Solve puzzles. Learn CUDA.
Experiment on rotation equivariance using Group Convolution U-Nets applied to segmentation of butterfly images
Vision Transformer implemented with JAX and dm-haiku
Using Alpha-Beta pruning and Minimax algorithms to play the Jaipur board game with both players possessing perfect information.