6.S198 Deep Learning Practicum


Undergraduate Level
Units: 3-3-6
Prereq: 6.034 or 6.036; Permission of instructor required
Instructors: Hal Abelson (hal@mit.edu), Natalie Lao (natalie@mit.edu)
Schedule:  MW3-4:30, room 56-154
This course prepares students to carry out projects that use deep learning in such areas as language processing and image recognition.  In the first part of the course, we’ll survey basic-deep learning frameworks, including convolutional neural networks, embedding, recurrent neural networks, and generative adversarial networks.  For each framework, we’ll introduce the fundamental concepts, explore open-ended demo applications and carry out programming assignments that modify and extend the demos. For the second part of the course, students will work in teams to design and implement original projects that use these techniques.  Each project team will have a mentor who is a machine learning expert from industry. We’ll also discuss some policy and societal issues in deep learning, such as algorithmic fairness and interpretability.
Programming work will be done in Javascript using a new (spring 2018) open-source platform called tensorflow.js, which is a version of Tensorflow that runs in the browser.
6.S198 counts as a course 6 advanced departmental lab (DLAB2) and an independent inquiry (II) subject.
Suggested  prerequisites are a working knowledge of Javascript and some exposure to machine learning (e.g., as in 6.034 or 6.036).
Enrollment is limited, permission of instructor required.
See the course home page at mit.edu/6.s198 for information on how to enroll.