ECE 6254 - Syllabus Overview

Matthieu R Bloch

January 7, 2020

Waiting list: I want to get in!

  • There is a big demand for the class (~70 people on the waiting list)

  • If you have a slot but are unsure about sticking around, please try to make up your mind by Thursday

  • To help you make an educated decision:
    • There is a self-assesment posted to test your background in probabilities and linear algebra
    • This class is not about programming, although there will be programming assignments
    • This class is not about learning about to build and train a deep neural network

Logistics

Class time and venue: Tuesday and Thursday 12pm-1:15pm, Klaus 1443

Image

Instructor: Prof. Matthieu Bloch

  • Office: TSRB 442
  • Email: matthieu.bloch@ece.gatech.edu
  • Office hours: 12pm-1:15pm on Wednesday, available by Blue Jeans

Websites

  • Piazza: for Q&A (Register!)
  • Canvas: for assignment posting and submission
  • Slack: for live Q&A (Sign up, link on canvas!)
  • Course website: https://bloch.ece.gatech.edu/ece6254sp20

Teaching assistants

  • To be announced at the end of the week once enrollment stabilizes

  • Guidelines for interactions
    • TAs are here to help, interact with them the same way you interact with me
    • Always cc me if you email them directly (hint: you shouldn’t)
    • Office hours will be announced early next week

Image
http://www.phdcomics.com

What to expect in ECE 6254

Image
https://xkcd.com/1838
  • ECE 6254 is about principles of learning theory and learning algorithms

    • We will use probability and linear algebra as our main tools
    • We will prove quite a few things formally (theorems, lemmas)
    • We will not develop cool apps based on Deep Neural Nets
    • Exams and homework will have theoretical components
  • All that being said…

    • We will also use simulations to understand concepts and algorithms
    • Homework will have an experimental component (Python required)
    • ECE 6254 is a fun course and you will learn a lot of useful concepts
  • ECE 6254 focuses primarily on foundations, but I will cover recent topics too

  • If you’re unsure about taking the class, the self-assessment is here to help!

Textbooks

Recommended textbooks

Image
$28 on amazon
Image
Free legal PDF
Image
$67 on amazon
Image
$52 on amazon
Image
$51 on amazon


  • The course notes will be largely self contained, but having other references helps

Electronic communication policy

Image
https://xkcd.com/1873
  • General guidelines
    • Email the Dean of Students if your personal situation requires special academic consideration
    • Use Piazza for technical questions
      • You can be anonymous to your peers, not to the instructors
    • Be courteous in your electronic interactions
      • Avoid judgmental language, e.g., “The answer is obvious.”
      • Try to be constructive
  • Assuming your situation does not fall in the above category
    • Include [ECE 6254] in the subject of the email
    • If you email the TAs, cc me.

Writing emails

Image
Image

Grading

  • Self-assessment test (5%, posted tonight!)
    • Review of concepts from calculus, linear algebra, probability theory, and programming.
    • Open-book/internet test but no collaboration permitted.
  • Homework (25%)
    • Due approximately every week (~10 assignments overall).
    • Both mathematical and programming problems.
    • You are encouraged to typeset in \(\LaTeX\).
  • Midterm exam (25%)
    • Tentatively scheduled March 3, 2020 during class.
  • Final exam (25%)
    • April 27 2020 11:20 AM - 2:10 PM in classroom
  • Final project (20%)
    • Teams of 3-5 students, in-depth study of a topic of your choosing.
    • Three deliverables: proposal, report, poster.

Assignments policy

Image
https://xkcd.com/1658
  • Two stage deadline policy
    • Soft official deadline with 3% bonus (conditions apply, read the fine print)
    • Hard deadline 48 hours after soft-deadline; no late homework accepted after hard deadline
  • Abide by the Georgia Tech honor code
    • Reference all your sources
    • Do not plagiarize other sources (python code, homework solutions, etc.)
    • Do not upload course material on other websites
    • When in doubt regarding what constitutes plagiarism, ask!
  • Assignments are individual but light collaboration permitted and encouraged
    • Piazza is here for that purpose
    • Small study groups are ok
  • Projects will be collaborative in teams of 3-5 students
    • Form your teams by February 7, 2020

How to succeed in ECE 6254?

Image
https://xkcd.com/1411
  • Don’t fake it until you make it
    • Go over the lecture notes and suggested reading
    • Work on assigned homework problem - struggling a bit is part of the process
    • Start assigments early
  • What will my grade be?
    • I don’t exactly know. Cutoffs are usually curved and you should do your best.
    • \(\P{\text{Grade}=A|\text{largely above average on every assignment}}\to 1\)
    • \(\P{\text{Grade}=C|\text{turn in sloppy homework}}\to 1\)
Image
Sample standing

Final thoughts

  • I believe in accountability, integrity, and fairness
    • I will hold you to the same standards
  • Don’t be shy and don’t hesitate to approach me!
    • I have zero tolerance for whining and complaining but I’m usually friendly
  • Top five comments I’ve received from CIOS
    • Sometimes the French accent makes it hard to understand what he says.
    • Your handwriting was terrible.
    • I hated your use of slides.
    • I payed $3,000 to take this class and even though you didn’t see all $300,000 in tuition and fees collected for this class directly in your paycheck, the students are your clients as well. Make sure they’re getting what they’ve paid for.
    • I would not be surprised if the division between A’s and B’s in the class was with the French student’s getting most of the A’s in the class and the other students getting the B’s.

Topics covered in class

  • Supervised learning
    • Theory of generalization
    • Classification and regression
    • Kernel methods
    • Gradient descent
  • Unsupervised learning
    • Dimensionality reduction
    • Density estimation
  • Reinforcement learning
    • Regret analysis
  • Additional cool topics as times permits
    • Decision trees
    • Ensemble methods (Random forests)
    • Deep learning
    • Fairness, Accountability, Transparency, and Ethics (FATE) in Machine Learning