A deep learning project in which we improve upon the Deep4SNet deep learning CNN model to counter the voice clones that have been camouflaged by the SiF-DeepVC method.

View GitHub

This project is about creating a toy model of Google Maps that can predict personalized travel time for a given trip. The project consists of three stages: map matching, traffic speed estimation, and travel time prediction. We used a road network data from San Francisco and a mobility trace data from the Cabspotting project. We implemented various methods such as Hidden Markov Model, XGBoost, and K-means to achieve the desired functionalities. The final performance of the travel time estimation is measured in mean absolute error (MAE). Our model performance resulted in 199.499 seconds; about 3 minutes of error.

This is the first time I've applied geospatial data in software development. It sparked my interest in real-world data because of the endless practicalities it can provide. Shoutout to my teammate, Xiyang Hu, for the endless nights we stayed holed up in the library, coding and researching methods.

View Documentation

View GitHub

An asynchronous multiplayer game in which you compete with other players to reach the longest distance ran as a cute Capybara. You can chat with other players and see each other's scores (distance) on the leaderboard.

This is my first time experiencing the full stack, and a full game development process. Shoutout to my teammate, Carlos Garro, for coding the physics and controls of the game. I had so much fun drawing the Capy and the environment. The backend server set up was annoying, but worth it. The frontend design process was fun, I really enjoyed writing the frontend side. It felt exhilarating when everything was working as intended at the end.

View GitHub

Play

In a team of 5, we analyzed the official geospatial COVID-19 data from various Chicago regions dating from January 2020 to October 2022. We carried out the prediction and analysis of multiple models on the data. We used the Time Series model to estimate the pattern of COVID infections in each region in mid-age, monthly, and seasonally. For the prediction of the number of weekly infections, we also used several different models to make judgments. Logistic regression, Naive Bayes, decision tree, and K-NN. In each model, we used 10-fold validation to split the data to get the highest accuracy of the model. We take the data predicted by different models and compare them to see the differences. At the same time, we also used the map data of Chicago to show the data analysis in a dynamic way. The audience of such analysis was real estate agencies in Chiacgo on location desirability.

This is was my introduction to data science course, so it was the first time I worked on a large data project cooperatively. It was fun coming up with theories of the numbers as a group. We shared opinions and doubts, and were able to create this amazing project. Shoutout to my teammates, Tongshu Wu, Mario Martino, Jacklyn Clauss, and Gabriella Rub for the fun and educating time.

View Documentation

Covid Growth by Zip

This is a collection of all the little games and tools I created during my academic journey to learn web development. Shoutout to my teacher, Professor James Femister!

Please be advised that they have been ported from Node.js to static due to GitHug Pages limitations.

Play

There's more!

Just gimmie some time to add them! lol