Toward Building a Cost-Effective and Robust Machine Learning System

Explore the challenges and solutions for developing cost-effective and robust machine learning systems in this 57-minute seminar by Cuong Nguyen, Assistant Professor at Florida International University. Delve into key strategies for reducing costs in data collection, model training, and evaluation while maintaining high performance. Learn about active learning techniques to optimize data gathering, and discover how continual and transfer learning can minimize model training expenses. Examine the critical role of transferability estimation and model selection in efficiently choosing the best source model for transfer to a target task, enabling ML model development with limited data and computational resources. Gain insights from Nguyen's extensive experience in probabilistic machine learning and artificial intelligence, spanning roles at Amazon Web Services, National University of Singapore, and Cambridge University.

Seminar Series: Toward building a cost-effective and robust machine learning system