The sections below lists all the school projects completed during my undergrad and graduate studies.

Project Pensive - Vector Institute: Language Model-driven Recommender Systems, Sep 2021 - Jan 2021
[ Presentation ] [ Code ]

  • Implemented a Language Model-driven Recommender System (LMRec) for conversational recommender systems using the Yelp and Reddit dataset.
  • Designed and implemented two user-controllable filters (for diversification & extreme content) for online recommendation tasks.
  • Evaluated model performance using multiple metrics such as F1-score, Mean Average Precision (MAP), and NDCG.

Deep Learning Course Project: Image Style Transfer, Jan 2020 - May 2020
[ Presentation ] [ Code ]

  • Implemented and compared two image-to-image translation frameworks, named Cycle-Consistent Adversarial Networks (CycleGAN) and Unsupervised Image-to-Image Translation Networks (UNIT) to complete the task of style transfer.
  • Conducted a small-scale sensitivity analysis on CycleGAN and UNIT and concluded experimental findings through both quantitative and qualitative analysis, where UNIT outperformed CycleGAN on two style transfer datasets at the cost of computing power and training time.

Stochastic Processing Course Project: Two-stage Stochastic Optimization for bike sharing systems, Sep 2020 - Dec 2020
[ Code ] [ Document ]

  • Implemented a two-stage stochastic optimizationg model which determines the optimizal number of bikes per stations and optimal transshipment of bikes for the bike sharing problem.

Machine Learning Course Project: Neural Network for Recommender Systems, Sep 2020 - Dec 2020
[ Presentation ] [ Report ] [ Code ]

  • Implemented a Variational Autoencoder (VAE) for a Neural Network Recommender System.
  • Improved the recommendation performance with a further enhanced hybrid framework, increasing 10% of the Mean Average Precision (MAP) measurement.
  • Evaluated model reliability and effectiveness using sensitivity analysis and metrics such as NDCG and MAP.

First-Place Capstone Project: In-car Conversational Recommender System, Sep 2019 - May 2020
[ Video ] [ Poster ] [ Code ]

  • Designed and built a robust personalized in-car restaurant recommender system for iNAGO Inc. with core features such as recommendation explanations that were demo in CES 2019.
  • Analyzed Yelp dataset by leveraging skills and knowledge in Information Retrieval and Sentiment Analysis to extract information from 6 million review data and to construct informative description for recommendations.
  • Programmed the explanation and critiquing processes for the system which enable sequential recommendations through interactive conversational interactions.

AI Course Project: Automotive Vehicle Make Recognition System, Jan 2020 - May 2020
[ Video ] [ code ]

  • Built a deep supervised transfer learning model with AlexNet and VGG-19 for classification of vehicle make.
  • Preprocessed 16k+ image data with 200 different class labels from the Stanford open-source dataset.
  • Performed both quantitative and qualitative analysis with a confusion matrix and feature extraction techniques.

Information System Project: PetSelect Information Website, Sept 2017 - Dec 2017

  • Carried out and implemented the idea of creating an informative web application for users to gain knowledge about different pet breeds in a team of 7.
  • Managed resource database using MS Access and developed the entire information system by implementing the interactions betwen Java, Java Server Pages, and SQL in Eclipse.
  • Examined and refined the web application by trouble shooting the system, debugging codes, and updating web content.