AI-assisted Knowledge Graph Design

Keywords: Machine Learning, Knowledge Discovery in Database (KDD)

____________

A knowledge graph-based framework to assist inexperienced engineers in developing new products by learning best practices from existing Computer-aided Designs (CADs) using Graph Neural Networks (GNNs).

____________

GitHub | PPT | Conference Presentation

Real-time Energy Disaggregation Using LSTM

Keywords: Machine Learning, LSTM, Signal Processing

____________

An LSTM-based machine learning module that automatically disaggregates devices' power signatures in real-time and predicts machine states. This module could be combined with HCI for energy and signal analysis of smart manufacturing.

____________

[Ongoing Research]

Real-time Traffic Monitoring with Computer Vision

Keywords: Computer Vision, Object Detection on 3D Point Cloud

____________

A project sponsored by the California Department of Transportation (Caltrans) to design a computer vision-based monitoring system for real-time traffic control and pedestrian safety analysis. The system utilizes real-time object detection on 3D Velodyne LiDAR point clouds. We are also working on creating and optimizing 3D Velodyne LiDAR dataset (like KITTI) for the proposed system.

____________

[Ongoing Research]

The Smart Connected Worker

Keywords: Machine Learning, Computer Vision, Intelligent Manufacturing, IoT, HCI

____________

An already deployed real-time monitoring system of manufacturing workflow that integrates state-of-the-art machine learning techniques with the workplace scenarios of advanced manufacturing systems.

____________

GitHub | Publication

Real-time Machine State Monitoring and Fault Detection

Keywords: Machine Learning, Computer Vision, Object Detection

____________

An application of Computer Vision and Artificial Intelligence (AI) models to replace discrete labor-intensive methods to monitor the machine state and predict the errors and risks for advanced manufacturing.

____________

GitHub | Publication | PPT | Conference Presentation

Real-time Human-Machine Interaction Monitoring

Keywords: Machine Learning, Computer Vision, Text Recognition, HCI

____________

A CRAFT-based finger detection and text recognition model for the real-time human-machine interaction control of a 3D printer.

____________

GitHub

Heterogeneous Graph Transformer

Keywords: Machine Learning, Transformers/Attention, GNNs, CSP/SAT Solving

____________

A Heterogeneous Graph Transformer (HGT) that combined GNN with transformers to create efficient SAT solvers. I investigated, analyzed and debugged numerous baseline models, and designed a suitable experiment that verified the performance of the proposed Heterogeneous Graph Transformer model.