Automation of Scientific Research
Course Number: 02-750
Automated science and engineering combines Robotics, Machine Learning, and Artificial Intelligence to accelerate the pace of discovery and rational design. This course introduces students to the Machine Learning and Artificial Intelligence algorithms that enable this emerging paradigm. Emphasis is placed on techniques for sequential analysis (i.e., model discovery and hypothesis generation), design of experiments, and optimization to maximize the return on research capital. Specific approaches will include Active Learning, Reinforcement Learning, and Bayesian Optimization. Examples of automated science and engineering from the literature will be studied. Grading will be based on homeworks, quizzes, and two exams.
Robotic scientific instruments are already used to decrease costs and increase reproducibility. Automated science and engineering take this one step further by leveraging Artificial Intelligence and Machine Learning to interpret data and select experiments in a closed-loop fashion. This emerging paradigm is motivated by the fact that most systems are too complex for humans to truly understand. Artificial Intelligence and Machine Learning can manage this complexity and find the most efficient paths to discovery and rational design by avoiding the costs of performing experiments where the outcome can already be predicted accurately.
- Quizzes (10%)
- Five quizzes will be given online via canvas. See the calendar below for due dates.
- Homeworks (50%)
- Graduate student section (02-750): Five graded assignments will be given. See the calendar below for due dates.
- Undergraduate student section (02-450): Four graded assignments will be given. See the calendar below for due dates.
- Undergrads may elect to do the fifth homework for up to 5% extra credit towards their final grade
- Each student will receive a credit of 3 grace days to be applied to assignments. You do not need to ask permission to use late days, they will be deducted automatically.
- Lateness policy: After your grace days have been exhausted, a 25% penalty will be applied for each day beyond the official due date.
- Cheating policy: All work must be your own. Unauthorized collaboration or plagiarism will result in a negative grade (e.g., a homework worth 100 points will be factored in as a -100 points towards your final grade) and will be reported to your academic advisor and dean.
- First exam (20%)
- An in-class exam will be given the final lecture before Spring break.
- Second exam (20%)
- An in-class exam will be given the final lecture of the semester.