6.867 Machine Learning
Fall 2017
image restricted to class participants
Image taken from https://www.toptal.com/machine-learning/machine-learning-theory-an-introductory-primer
Instructors: Devavrat Shah, David Sontag, Suvrit Sra
TAs: Nirav Bhan, Hongge Chen, Hyunghoon Cho, Michael S Fleder, Curtis George Northcutt, Ardavan Saeedi, Manjot Sangha, Lawrence J Sun, Zhi Xu, Tiange (Alice) Zhan, Chengkai Zhang, Liang Zhou
Lecture:
Tues/Thurs: 2.30-4
(26-100)
Recitation: Fri: 9am (26-322), 9am (36-144), 10am (36-144), 10am (26-322), 11am (36-144), 11am (26-322), 12pm (56-154), 12pm (26-322), 1pm (56-154), 1pm (26-322), 2pm (26-322), 2pm (56-154), 3pm (26-322), 3pm (56-154), 4pm (26-322), 4pm (56-154)
Office Hours: Monday/Wednesday: 4-6
(36-156)
Course Information:
Piazza:
We will use Piazza for making announcements, and managing
discussions:
https://piazza.com/class#fall2017/6867
Description:
Principles, techniques, and algorithms in machine learning from the point of view of statistical inference; representation, generalization, and model selection; and methods such as linear/additive models, active learning, boosting, support vector machines, non-parametric Bayesian methods, hidden Markov models, and Bayesian networks. The course will give the student the basic ideas and intuition behind modern machine learning methods as well as a bit more formal understanding of how, why, and when they work. The underlying theme in the course is statistical inference as it provides the foundation for most of the methods covered. Recommended prerequisite: 6.036.
Books:
• [B] Pattern Recognition and Machine Learning, Bishop;
Springer, 2007.
• [EH] Computer Age Statistical Inference, Efron and Hastie;
Cambridge University Press, 2016.
• [JWHT] An Introduction to Statistical Learning: , James, Witten,
Hastie, Tibshirani; Springer, 2013. Easy read, very pragmatic.
Announcements
HW 1 paper reviews for groups (clarification)
Please note that all the students who submitted a report, have been assigned 3 papers to review. That means even if your HW partner has submitted the report, you still need to register on CMT and review the papers that have been assigned to you. Please use your MIT email and try resetting your password if you see 'email already in use'.Announced on 30 September 2017 12:55 p.m. by Ardavan Saeedi
HW1 reviewer assignment
Hi all,We have assigned each individual student (who submitted HW1) three papers to review. Reviews must be entered on the CMT site by 11:59PM on Tuesday, October 3. Please login and change your role to a reviewer. If you have any question/issues regarding CMT or the papers assigned to you (e.g. paper not anonymized), please contact ardavans@mit.edu.
The 6.867 staff
Announced on 29 September 2017 1:51 p.m. by Ardavan Saeedi
HW 1 Final Reminders
Final reminders:
- Register here: https://cmt3.research.microsoft.com/MITML2017
- Make sure you can make a submission. If you see "email already being used" try resetting your password. (DO NOT LEAVE THIS TO THE LAST MINUTE).
- Due to the size of the class we will not accept any submissions from listeners and non-registered students.
- If you are submitting as a group you should:
- Only submit one file per group (only for this assignment).
- Add the name of the person who is in the group but not uploading the file as a coauthor (not on the paper but in CMT).
- Add the email of your HW partner who is not uploading the file in the conflict of interest section.
- Double-check your paper before submission and make sure it’s anonymized (you will lose 20% if your paper is not anonymized).
- After the deadline the submission will be closed and you won’t be able to submit via CMT (Deadline: 11:59 PM on Thursday, September 28).
- There is a 20% automatic deduction from your grade if you submit after the deadline.
For the unlikely event of a late submission or other issues regarding submission, contact ardavans@mit.edu.
Announced on 28 September 2017 11:04 a.m. by Ardavan Saeedi
Project milestone 0 reminder
Project milestone 0 is due today. Please remember to submit through the google form before 11:59 pm. Again, each student must submit using the form. If you are a listener, please do not submit.Announced on 26 September 2017 5:04 p.m. by Zhi Xu
HW1 submission reminder (less than 3 days to deadline)
Some additional points:
- If you are submitting as a group you should:
- Only submit one file per group (only for this assignment).
- Add the name of the person who is in the group but not uploading the file as a coauthor (not on the paper but in CMT).
- Add the email of your HW partner who is not uploading the file in the conflict of interest section.
Pre-submission (REMINDER)
- Register here: https://cmt3.research.microsoft.com/MITML2017
- Make sure you can make a submission. If you have issues with the registration or submission, contact ardavans@mit.edu (DO NOT LEAVE THIS TO THE LAST MINUTE).
- Note that due to the size of the class we will not accept any submissions from listeners and non-registered students.
Submission (Deadline: 11:59 PM on Thursday, September 28)
- Double-check your paper before submission and make sure it’s anonymized (you will lose 20% if your paper is not anonymized).
- After your final submission, download the submitted paper and make sure it is the anonymized correct version.
- If you are submitting as a group, add the email of your HW partner who is not uploading the file in the conflict of interest section.
- After the deadline the submission will be closed and you won’t be able to submit via CMT.
- There is a 20% automatic deduction from your grade if you submit after the deadline.
Post-submission
- We will try our best to send all papers out for reviews in 24 hours. Deadline for submitting the reviews is 11:59PM on Tuesday, October 3. Use the same link as above to login and then change your role to “Reviewer” in order to review papers.
For the unlikely event of a late submission, contact ardavans@mit.edu.
Announced on 26 September 2017 3:03 p.m. by Ardavan Saeedi
