SHING-HEI

Here are some of my personal projects:

Bayesian Neural Network (BNN) | Code | (Spring 2024)

I programmed a BNN with two layers and a factorized Gaussian posterior, and tested it on a small dataset for binary classification. See the code for the results.

Gradient-based Grasp Planner | Demo | Code | (Summer 2023)

I programmed a grasp planner that iteratively refine the grasp pose of a robot gripper by gradient descent to maximize/minimize the predicted deformation on a deformable object represented by its partial-view point cloud.

In the image, the green partial-view point cloud is the object, the red point cloud is the initial gripper. The black point cloud is the optimized gripper that minimizes deformation

Sentiment Classification | Code | (Spring 2023)

I programmed and analyzed classical ML models such as Adaboost and SVM for classifying sentiment of movie reviews.

Recommender System for Companies and Investors | Demo | Code | (Summer 2022)

How can we reduce the workload of raising funds and company search for founders and investors ?

My recommender system recommends companies to investors using predictions generated by clustering on feature vectors and fully-connected networks.

BlackJack Q-Learning | Demo | Code | (Fall 2022)

I programmed it from scratch. The code learns to play blackjack via tabular q-learning.

C++ Blackjack Training App | Demo | Code | (Fall 2022)

The player can play multiple games until he runs out of money to bet. At the end of each round, the player will be notified which move he should have played and the estimated likelihood of winning would be updated.

Polynomial Interpolation for Robot Surgery | Code | (Fall 2023)

Cutting is one of the most common and essential tasks in surgery. Valuable autonomous surgical assistants need to be able to cut tissue precisely and smoothly. How can we generate a smooth trajectory for a robot to trace and cut a retracted tissue? The code generates continuous curves that interpolates points that the robot should reach. Check out the visualization in the code.

Each image on the left shows the trajectory (red) generated by my cubic spline interpolation at the bottom of the retracted tissue.

C++ Sprite Editor | Demo | Code | (Fall 2022)

A sprite editor that allows drawing pixel arts, playing animation consisting of multiple frames in a loop to make a sprite and importing a sprite in JSON format.

Spreadsheet Application | Demo | (Spring 2022)

Programmed in C#. The Spreadsheet Application has two parts: GUI and Model. The Spreadsheet Application can evaluate formulae, track dependencies of variables in the formulae, detect invalid formulae, and parse an XML file that encodes a spreadsheet.

Convolutional and Fully-Connected Neural Networks | Code | (Summer 2022)

I built the neural networks for image classification. Back propagation is done by raw tensor operations instead of ML libraries.

Agar.io Simulation | Demo | (Spring 2022)

Programmed in C#. This program simulates an online multiplayer game Agar.io. The main challenges are interaction between server and clients and designing control mechanisms that match the display region’s coordinates.

Page updated

Google Sites

Report abuse