CS6140 Machine Learning, FALL 2015
*check/reload this page every day
- Please complete the TRACE evaluation for CS6140: In order to
access your survey progress, please log in to your myNEU
account; you can access the TRACE link under the Self-Service
- You can also write directly to me (personally or
anonymously) about the course, if you feel more comfortable
that way. Either before or after the grades are due, your
opinion about the course or the instructor/TAs will not affect
the grade in any way.
- HW6 and HW7 dont have late days penalties. You will get the
points (full or partial) if you demo the code by Sat 12/19. By aware that last day is
usually very busy, so please demo ahead of time if you are ready
(also demoing in last session gives you no time to make the
changes TAs might ask for)
- HAPPY THANKSGIVING ! Here is the schedule for week 11/23
- No class or office hours on Wed 11/25
- No office hours Thu 11/26 - enjoy the turkey.
- Office hours as usual on Fri 11/27 and Sat 11/28.
- We are interested in a qualified (math, programming,
communication) student to work on a project involving running a
state of the art Deep Neural Network on proper GPU
hardware. The paper we are interested to follow is about
doing multi-label classification:
- This project can wave HW7 requirement for the course,
but it is definitely more demanding than HW7, so in any
case it doesnt give an easier route to a good grade. I also
expect the work on this project to go beyond the end of the
- HW4 extended to Sat 11/7. The HW schedule will probably
- HW4 due Sat 11/7
HW5 due Fri 11/20
HW6 due Fri 12/4
HW7 due Sun 12/13
- HWs and notes will be available a little earlier, so
students can start the next HW if they finish the current one
before the deadline.
- Added a problem (regular credit) on "Gradient Boosted Trees
for Regression". It shouldn't take more than 2 hours,
assuming your regression tree code from HW1 works.
- Make-up class : SATURDAY NOV 21
NOON-3pm, Behrakis room 220.
- Monday 10/12 COLUMBUS DAY - no office hours. Happy Holiday!
- Office hours room 362 WVH
- Mon, Tue, Thu, Fri 5pm-7pm
- Sat noon-2pm
- Please use the piazza
discussion forum for all questions regarding material,
assignments, due dates, data issues, programming issues, etc.
That is, do not use the direct email to TAs or Instructors for
Personal/private matters, such as availability, delays,
grades, term projects or other advanced material, etc, should be
discussed by email.
- A friendly advanced warning on particularly time
consuming exercises: you might want to start these as
early as possible, and make sure to first read all the
- HW2 Neural Network (the autoencoder problem)
- HW4 The ECOC implementation
- HW6 The SMO solver implementation for SVM
- HW1 posted,
due Tue September 22
- Set up Dropbox
Sign up for Dropbox if you dont have an account already
Create a *code only* folder inside Dropbox on your working
machine (in preferred location) named CS6140_X_Yabcd with
X=First name initial, and Yabcd=your last name. For example if
one's name is Michael Jordan, the folder would be called
This folder is for code, results, reports, pdfs, notes - BUT
NOT FOR DATA. Keep data on a separate location on your
Share this folder via Dropbox with "email@example.com". Note
that we are not using this address for email, so we wont read
any emails sent to it.
- Please complete the student
information and put it as a file in your dropbox folder
- Discussion forum :
sure it directs your browser to
- Welcome to CS 6140- Machine
Learning. Bookmark this page! General structure of the
- Seven modules (2 weeks each) with readings, lecture notes,
slides, explanations (video), recorded lectures, and of course
- Assignments (homeworks) are focused on programing ML
techniques, training on data, test/evaluate, and
- Mandatory: You are expected to participate at office hours;
either to discuss your difficulty with the assignment, or to
demo the code and results for grading.
- Each assignment is graded on the spot by a TA when you have
it working. We wont be picky with the grading, but some
assignments can be quite challenging. Often the TA will point
to something that needs more work before you can get full