Automation of Scientific Research
Course Number: 02-450
Automated scientific instruments are used widely in research and engineering. Robots dramatically increase the reproducibility of scientific experiments, and are often cheaper and faster than humans, but are most often used to execute brute-force sweeps over experimental conditions. The result is that many experiments are “wasted” on conditions where the effect could have been predicted. Thus, there is a need for computational techniques capable of selecting the most informative experiments.
This course will introduce students to techniques from Artificial Intelligence and Machine Learning for automatically selecting experiments to accelerate the pace of discovery and to reduce the overall cost of research. Real-world applications from Biology, Bioengineering, and Medicine will be studied. Grading will be based on homeworks and two exams. The course is intended to be self-contained, but students should have a basic knowledge of biology, programming, statistics, and machine learning.
Students who complete the course successfully will be able to:
- Understand and explain core concepts and experimental methods in molecular and cell biology
- Understand the core concepts and algorithms used in automated science and engineering
- Modify and apply software to automate several research and engineering tasks
- Apply knowledge of automated science algorithms to select and justify techniques for addressing various discovery and design challenges
- Evaluate and interpret the results of the chosen approach
- Understand, explain and critique published papers that employ automation for scientific research and engineering