(a) Time-Critical Edge Applications in Autonomous Vehicles (Dr. Kecheng Yang)
With the advances in robotics, autonomous systems have become increasingly pervasive and are able to provide considerable computing power at the edge. Meanwhile, a significant portion of data and applications in such autonomous systems are time-sensitive. This project focuses on the infrastructures and the schedulers in particular in such systems to improve their abilities to support time-critical applications. The Robot Operating System (ROS) has been widely used to develop autonomous systems, and its successor ROS2 was proposed for enhanced real-time support and performance. Students will experience how the applications are implemented on top of ROS or ROS2, investigate the computing resource allocation schemes, and perform system-level optimization to improve the real-time performance of the applications. The outcomes will be evaluated by both simulation and actual implementation on autonomous model cars.
(b) Wearable Sensing System for Personalized Fall Detection at the Edge (Dr. Anne H.H. Ngu)
Fall is a major public health problem leading to loss of life independence, premature mortality, and major hospitalization cost. Wearable devices, especially smartwatches that pair with smartphones, are increasingly a platform of choice for deploying digital health applications. This project focuses on creating a personalized fall detection system on low-cost smartwatches that can detect both “hard” and “soft” falls. In particular, students will experiment with different machine learning personalization strategies, different model validation strategies and scalable mechanism for data collection for personalization. Students will be exposed to skills such as Android programming, machine learning with Python, non-SQL database and edge computing using smartwatches and phones.
(c) AI-based Auxiliary Diagnostic Systems for Early Detection of Chronic and Systemic Diseases (Dr. Mylene C.Q. Farias)
As the global population grows and ages, the prevalence of chronic and systemic diseases increases, which demands strategies to develop preventive public health initiatives. The diagnosis of chronic and systemic diseases is generally based on medical imaging exams, such as X-rays, computational tomography, and magnetic resonance imaging scans. The aim of this project is to create auxiliary diagnostic systems based on modern deep learning models, which cannot only help diagnose these but also differentiate between areas with and without pathologies. Early auxiliary diagnostic systems can be tailored to identify specific conditions like cardiovascular diseases, osteoporosis, and cancer. For instance, the examination of maxillofacial dental panoramic X-rays can reveal early signs of osteoporosis and cardiovascular diseases. Similarly, high-quality magnetic resonance imaging scans can be employed for the early detection of various types of cancer, such as those affecting the brain, liver, and breast. Therefore, the development of auxiliary diagnostic systems holds promise in enhancing early diagnosis of chronic and systemic diseases and contributing to more effective healthcare interventions.
(d) Humanizing Large Language Models (Dr. Heena Rathore)
Are you intrigued by the mind of a machine? If yes, this project pushes the boundaries of what AI can understand and feel. We will work on projects that delve into Large Language Models (LLM’s) ability to demonstrate empathy, navigate moral dilemmas, and more. Imagine designing an LLM that can detect subtle emotional cues in text, fostering deeper connections between humans and technology. Alternatively, consider building an LLM that grapples with complex ethical problems, offering nuanced perspectives on them. This project will enable human-AI interactions in an effective manner and reshape our understanding of LLMs.
(e) Machine Learning Compilers for Scalable Graph Computation (Dr. Apan Qasem)
Graph algorithms are at the core of data-intensive applications in many computational domains including cybersecurity, medical informatics, business analytics and deep learning. As such, efficient graph processing is of critical concern. While recent advancements in massively parallel GPU graph analytics have demonstrated remarkable results, the irregular structure of graphs remains a significant challenge in unleashing the full capabilities of the underlying hardware. Irregularity in real-world graphs can make performance unpredictable and non-portable across different inputs and architectures.
Our research project will focus on harnessing the power of machine learning compilers to optimize the performance and energy efficiency of critical graph algorithms. The primary objective is to conduct an in-depth study using real-world graph datasets to identify and understand the dominant performance traits of these algorithms. We will utilize the insights gained from the study to develop novel code transformations that can yield performance and energy across a range of architectures and problem domains. Participants in this project will be working with the CompilerGym framework from Meta and should have some familiarity with Python. Background in compilers and machine learning is a plus but not required.