We are no long accepting registrations or requests for this course.

This will be a project-based IAP course that aims to develop new AI for a series of problems. Students will work closely with MIT faculty/staff in small teams and will be provided with data, project ideas, and computing resources. There will be numerous opportunities for successful and interested students to continue their research with ongoing research projects after the IAP course through UROPs etc.

AI challenges such as ImageNet, CIFAR, Graph Challenge, Moments in Time have resulted in major advances in image recognition, graph processing, and video action recognition. These and many other challenge problems are characterized by: 1) open datasets, 2) clear problem statements and 3) baseline implementations. Inspired by these challenges, through the USAF-MIT AI Accelerator, we are developing challenge problems to bring AI innovations to domains such as:

1) Datacenter Monitoring: Develop AI to monitor datacenter energy consumption

2) Magnetic Navigation: Develop AI for aerial vehicles for navigation in GPS denied environments by leveraging novel ML techniques alongside physical modelling.

3) SEVIR: Develop classification/object detection AI to detect tornadoes in radar data 

4) CogPilot: Develop AI techniques using multimodal physiology signals for pilot performance monitoring

For this IAP course, students will select a data challenge from above and will be developing AI algorithms collaboratively.