Graph Neural Networks Projects

The project aims to explore how Machine Learning can improve Multi-Agent Path Finding. It focuses on enabling multiple agents to move from their starting points to goals without collisions. The research examines how ML enhances the efficiency and coordination of agents in complex environments. It studies environment representation, path planning, and execution of solutions. The goal is to understand and highlight how ML can transform multi-agent navigation in large-scale automated systems like warehouses.

This project focuses on improving how data moves across a network. It uses machine learning to predict traffic patterns for several time periods ahead. The method helps route data more efficiently, reducing sudden network changes. It also lowers the cost of monitoring the network while keeping performance close to the best possible.

This project focuses on making computer networks smarter and more efficient. It finds the best path for network services to travel through virtual functions while considering network limits. The system learns from past data using advanced AI techniques to quickly adapt to changes. Tests show it works almost as well as exact solutions but much faster.

This project focuses on improving wireless communication in millimeter wave systems. It uses a smart learning method to design the transmitter signals directly from received signals. This approach works well even when users move or the network changes. The method reduces errors, speeds up communication, and works for different network setups without extra training.

This project is about teaching computers to analyze very large medical images for disease detection. Normally, labeling these images takes too much time. The researchers created a new method that looks at the images at different zoom levels and learns how parts of the image relate to each other. Their approach makes predictions more accurate and works better than previous methods on standard datasets.

This project develops a new method to measure blood pressure without using a cuff. It finds which body signals actually affect blood pressure and then uses a smart computer model to track changes beat by beat. The method works accurately for people of different ages and health conditions. It can help monitor blood pressure continuously and more reliably than current approaches.

This project develops a smart system called DeepSIG to identify types of radio signals. It combines three different AI models to look at signals in multiple ways, like sequences, images, and graphs. The system learns from all these views together to make more accurate predictions. Tests show it works better than using any single method, especially when only a few signals are available.