Continual Diffusion

James Seale Smith is a PhD student at Georgia Institute of Technology specializing in machine learning. He focuses on continual adaptation of strongly pre-trained models, with a particular emphasis on protecting data privacy and minimizing computing resources. His research questions revolve around continuously adapting pre-trained models to emerging and difficult concepts, customizing pre-trained models to personal user data, and efficiently re-training and replacing large-scale models. Some of his recent accomplishments include papers accepted at major conferences, co-organizing events, and serving on boards and committees related to machine learning. His interests also include lifelong learning, knowledge distillation, federated learning, and low-label learning. Overall, James Seale Smith"s work revolves around developing solutions and techniques for continual learning in the fields of computer vision and natural language processing.