vITAL Lab
As the director of the Information Theory and Machine Learning (vITAL) research lab at USC, I lead a team dedicated to pioneering research in information theory and machine learning. Our lab focuses on tackling foundational problems in these fields, particularly looking at issues like trustworthy and scalable federated learning, ensuring security and privacy in large-scale distributed systems, exploring the potentials of transfer learning, and innovating in coded computing. These research directions are crucial as they address the challenges and opportunities presented by modern technological advancements and the need for secure, efficient computational methods in distributed settings. Currently, we focus on four main directions:
In our work at the vITAL lab, we focus on developing trustworthy and scalable federated learning systems. Federated learning is a machine learning setting where the goal is to train a model across multiple decentralized edge devices or servers holding local data samples, without exchanging them. This technique is particularly important for preserving privacy and reducing the need to centralize data, which can be sensitive or proprietary.
In our research at the vITAL lab, addressing security and privacy in large-scale distributed systems is a core focus. As these systems grow in complexity and scale, they become more vulnerable to security breaches and privacy issues. The nature of distributed systems—where data and computational tasks are spread across multiple locations—poses unique challenges for maintaining confidentiality, integrity, and availability.
In the realm of our research at the vITAL lab, transfer learning is a significant area of focus due to its potential to dramatically improve the efficiency and effectiveness of machine learning models. Transfer learning involves taking a model developed for one task and reusing it on a second, related task. This is particularly valuable in scenarios where labeled data is scarce or expensive to obtain.
In our research at the vITAL lab, coded computing is a transformative approach that we are pioneering to enhance the efficiency and fault tolerance of distributed computing systems. Coded computing involves integrating coding theory with distributed computing tasks to address key challenges such as latency, errors, and node failures in large-scale systems.
Current Members
The current members of our lab are a vibrant group of researchers and students who are deeply engaged in pushing the boundaries of what's possible in our key research areas. They bring fresh perspectives and innovative ideas that are crucial for our ongoing projects and future breakthroughs.
Previous Members
Previous members of our lab have successfully leveraged their experience here to make significant contributions in academia and industry, furthering the field of information theory and machine learning. They have carried forward the rigorous research standards and innovative spirit fostered at the vITAL lab into their professional endeavors.
