Projects

Here are some of the personal, in-class, and research projects I made back in the days.
Note: this page is only maintained till 2023. For further projects, please see my research page.

Split-it

A webapp that splits the bill for a group of friends. It parses the receipt image into food items and prices using OCR and ChatGPT, and achieves SOTA parsing performance.

BOVW

Bag of Visual Words classifier for image classification. Implemented using cv2 and sklearn, with applications in Reinforcement Learning.

Deblur

Debarring Text Images via L0 Regularized Intensity and Gradient Prior: a Python implementation

Decoding

Bayesian decoding of subsitution cipher with Markov Chain Monte Carlo

Fork

Tinder, but for food: a webapp that helps a group of friends decide on a restaurant to dine at through ML-based recommendation and automatic selection

Maps

a webapp that allows users to find shortest routes between 2 points on the Providence map using a GUI

Deep Manager

a Deep Reinforcement Learning agent that manages a portfolio in the stock market and can make profits; trained in TensorFlow with advantage REINFORCE

Park@Brown

a mathematical model to compute the best parking spots allocations strategy at Brown University

IMU + object recognition

An object recognition post-processing pipeline that uses Inertial Measurement Unit to strengthen detection accuracy by 50%. Used on Vision Prosthetic in Paradiso Lab.

Controbo

A Unity API that enables controlling robot navigation using pointing gestures to corroborate natural language commands.

Cha-Ching!

an autonomous Java trading bot that trades intelligently in a simulated stock market

RL Pong

a Reinforcement Learning agent (using policy gradient) that learns to play the Atari game Pong through trial-and-error

RL Cartpole

a Reinforcement Learning agent (Deep Q Network) that learns to balance a pole on a cart through trial-and-error

Connect4

an AI connect4 player using Minimax algorithm and alpha-beta pruning; implementation won first place in Brown CS18 class tournament.

Search

a Scala search engine that uses term frequency and pageRank algorithm to index a wikipedia corpus and return most relevant documents

Markov Chain

group research paper on the application of discrete-time Markov Chains in the problems of randomizing

Fibonacci

a group research paper on the interesting number-theoretical properties of the Fibonacci sequence

Turing System

an analysis of the usage of Turing System in species competition problems