The constant advancement of laboratory automation techniques has greatly increased the speed and scope of experimentation while reducing experiment cost. Even with these advances, many experimental systems are too large to explore exhaustively. The alternative is to collect a feasible subset and use it to train a predictive model for the rest. The question then becomes how to choose this subset. The field of active machine learning provides methods that can be used in a closed-loop experimental system to iteratively choose experiments that are most likely to yield an accurate predictive model. In this course, participants will experience the development of an “automated science” AI-driven process from definition of an experimental space and process goals, to automating data collection, building a predictive model and using the model to choose new experiments. The process will be focused on playing a simple children’s board game (e.g, Battleship) using an integrated system containing a liquid handling robot, a plate reader, a robotic arm and plate management capabilities. It is strongly encouraged that participants in this course have programming experience in Python or Matlab.
- Basic assay design
- Using simulation in design and testing of modeling and active learning components
- Set up of integrated robotic system for AI-driven experimentation
- Management of communication between AI and hardware
Joshua D. Kangas, Ph.D., Assistant Teaching Professor, Computational Biology Department, School of Computer Science, Carnegie Mellon University.
Robert F. Murphy, Ph.D., Ray and Stephanie Lane Professor of Computational Biology and Professor of Biological Sciences, Biomedical Engineering, and Machine Learning Head, Computational Biology Department, School of Computer Science, Carnegie Mellon University.