CS 342 - Neural networks - Fall 2019
meets
MWF 10am-11am UTC 3.102
instructor Philipp Krähenbühl ( philkr (at) utexas.edu )
office hours M 11am-noon GDC 4.824
Please only visit during office hours or with appointment (or if I didn’t reply your email after the third try)
TA
Xingyi Zhou ( zhouxy (at) cs.utexas.edu )
office hours W 1:30PM-2:30PM GDC 1.302 (Basement) Desk 5
Brady Zhou ( bzhou (at) cs.utexas.edu )
office hours W 11:30AM-12:30PM GDC 1.302 (Basement) Desk 1
Ishan Nigam ( ishann (at) cs.utexas.edu )
office hours M 11:30AM-12:30PM GDC 1.302 (Basement) Desk 1
This course covers the basic building blocks and intuitions behind designing, training, tuning, and monitoring of deep networks. We will cover both the theory of deep learning, as well as hands-on implementation sessions in pytorch. We will also cover a series of application areas of deep networks in: computer vision, sequence modeling in natural language processing, deep reinforcement learning, generative modeling, and adversarial learning. In the homework assignments, we will develop a vision system for a racing simulator, SuperTuxKart, from scratch.
Prerequisites: Python, basic ML background
Textbooks: None
Class overview
We will use Piazza for questions and canvas homework. CS 342 follows the flipped classroom model, and delivers all course material online. The schedule below shows the due dates for all modules. Open problems are optional.
Each weak, we start with a quiz on the course material Monday, and then work through the results of the quiz and any questions in class. Wednesday and Friday, we work on a medium sized coding project related to the course material. The TAs and I will do the coding, but you’ll tell us what do code up. We’ll split the class into even groups, and each group will solve the problem individually. We’ll train any larger networks over night following the class, and then compare solutions.
Schedule
Date | What | Material | |
---|---|---|---|
Wed | Aug 28 | Introduction | |
Fri | Aug 30 | Coding - Estimating PI | ipynb html team 1 team 2 team 3 |
Sun | Sep 01 | up to and including Background.Summary | |
Mon | Sep 02 | No class - Labor day | |
Wed | Sep 04 | Coding - Part I | ipynb html team 1 team 3 team 4 |
Fri | Sep 06 | Coding - Part II | team 1 team 2 team 3 team 4 |
Sun | Sep 08 | up to and including First example.Summary | |
Mon | Sep 09 | Quiz | |
Wed | Sep 11 | Coding - Part I | ipynb html team 1 team 2 team 3 |
Fri | Sep 13 | Coding - Part II | team 1 team 2 team 3 team 4 |
Sun | Sep 15 | up to and including Deep networks.Summary | |
Mon | Sep 16 | Quiz | |
Wed | Sep 18 | Coding - Part I | ipynb html team4 |
Fri | Sep 20 | Coding - Part II | team 1 team 2 team 3 team 4 |
Fri | Sep 20 | Homework 1 due 11:59pm | |
Sun | Sep 22 | up to and including Convolutional Networks.Building efficient convolutional networks | |
Mon | Sep 23 | Quiz | |
Wed | Sep 25 | Coding - Part I | ipynb html Phlipp |
Fri | Sep 27 | Coding - Part II | team 1 team 2 team 3 team 4 |
Sun | Sep 29 | up to and including Making it work.Initialization in pytorch | |
Mon | Sep 30 | Quiz | |
Wed | Oct 02 | Coding - Part I | ipynb html team4 |
Fri | Oct 04 | Coding - Part II | team 1 team 2 team 3 team 4 |
Fri | Oct 04 | Homework 2 due 11:59pm | |
Sun | Oct 06 | up to and including Making it work.Open Problem: Pruning and compression | |
Mon | Oct 07 | Quiz | |
Wed | Oct 09 | Coding - Part I | ipynb html |
Fri | Oct 11 | Coding - Part II | team 1 team 2 team 3 team 4 |
Sun | Oct 13 | up to and including Making it work.Summary, a practical guide to deep network optimization | |
Mon | Oct 14 | Quiz | |
Wed | Oct 16 | Coding - Part I | ipynb html |
Fri | Oct 18 | Coding - Part II | final ipynb final html |
Fri | Oct 18 | Homework 3 due 11:59pm | |
Sun | Oct 20 | up to and including Computer Vision.Case Study: RetinaNet | |
Mon | Oct 21 | Quiz | |
Wed | Oct 23 | Coding - Part I | ipynb html |
Fri | Oct 25 | Coding - Part II | |
Sun | Oct 27 | up to and including Computer Vision.Summary | |
Mon | Oct 28 | Quiz | |
Wed | Oct 30 | Coding - Part I | ipynb html |
Fri | Nov 01 | Coding - Part II | |
Fri | Nov 01 | Homework 4 due 11:59pm | |
Sun | Nov 03 | up to and including Sequence modeling.Summary | |
Mon | Nov 04 | Quiz | |
Wed | Nov 06 | Coding - Part I | ipynb html |
Fri | Nov 08 | Coding - Part II | ipynb |
Fri | Nov 08 | Homework 5 due 11:59pm | |
Sun | Nov 10 | up to and including Reinforcement learning.Gradient free optimization | |
Mon | Nov 11 | Quiz | |
Wed | Nov 13 | Coding - Part I | ipynb html |
Fri | Nov 15 | Coding - Part II | |
Fri | Nov 15 | Homework 6 due 11:59pm | |
Sun | Nov 17 | up to and including Embedding learning.Summary | |
Mon | Nov 18 | Quiz | |
Wed | Nov 20 | Coding - Part I | ipynb html |
Fri | Nov 22 | Coding - Part II | |
Sun | Nov 24 | up to and including Adversarial attacks.Summary | |
Mon | Nov 25 | Quiz | |
Wed | Nov 27 | No class - Thanksgiving | |
Fri | Nov 29 | No class - Thanksgiving | |
Sun | Dec 01 | No lectures | |
Mon | Dec 02 | Research presentation - Philipp | |
Wed | Dec 04 | Project presentations I | |
Fri | Dec 06 | Project presentations II | |
Fri | Dec 06 | Final project due 11:59pm | |
Sun | Dec 08 | up to and including Final words.Open Problem: Bias, fairness, and ethics in deep learning | |
Mon | Dec 09 | Final project results |
Assignments and final project
- homework 1 (8%)
- homework 2 (8%)
- homework 3 (8%)
- homework 4 (8%)
- homework 5 (8%)
- homework 6 (8%)
- final project (30%)
All homework are due at 11:59pm, but you’ll have one additional hour to upload your assignment to canvas (in case your internet connection is really slow). Please do not email me for further extensions.
Collaboration / Academic Honesty Policy
Collaboration is not allowed for homework or Quizzes. Every homework and quiz needs to be solved individually. We will check for duplicates. For the final project groups of up to 3 students are allows. Different groups are allowed to discuss ideas and share data, but no code.
The online course format allows for multiple methods of identity verification, collusion, collaboration and plagiarism monitoring and detection. A violation of the course policy may include (but is not limited to) the following:
- Providing your UT EID to any other person
- Collaborating or sharing information with another person regarding the material on any quiz, assessment or assignment, before, during and/or after any quiz, assessment or assignment
- Recording any quiz, assessment or assignment material in any format
- Failing to properly cite language, ideas, data, or arguments that are not originally yours
- The public (such that it can be viewed by more than one person) posting of any form of a test bank or group of questions from any assignment
- Consulting forbidden materials or sources of information
The University of Texas at Austin Academic Integrity principles call for students to avoid engaging in any form of academic dishonesty on behalf of yourself or another student. Grade-related penalties are routinely assessed (“F” in the course is not uncommon), but students can also be suspended or even permanently expelled from the University for scholastic dishonesty.
If you have any questions about what constitutes academic dishonesty, please refer to the Dean of Students website or contact the instructor for this course.
You must agree to abide by the Honor Code of the University of Texas. You will not work with or collaborate with others in any way while completing any of the graded course assignments.
Late Policy
Quizzes: N/A
Homework:
- 1 day: -20%
- 2 days: -50%
- 3 days: -100%
Final Project:
- 1 day: - 100%
You have a total of 3 slip-days for your homework. There are no slip-days for the final project or quizzes. A homework that is 3 days late will be graded at -100% irrespective of slip-days, as we will release the solution on day 3.
Assignments, Assessment, Evaluation
- Quizzes (10%): Quizzes (linear, may miss 2).
- Coding (12%): Attendance and involvement (may miss 2, afterwards 1% deduction per class missed).
- Homework (48%): There are six homework assignments in this class (8% per assignment).
- Final Project (30%): There is one final project in this class.
Documented Disability Statement
The University of Texas at Austin guarantees that students with disabilities have access to appropriate accommodations. You may request an accommodation letter from the Division of Diversity and Community Engagement, Services for Students with Disabilities https://diversity.utexas.edu/disability/.
If you have approved accommodations for the course, please contact us to arrange them. Please do this as soon as possible, so that you can have the benefit of the accommodations throughout the duration of the course.
Course Etiquette
We expect that you will treat online discussions as though you are having a civil, respectful discussion with your fellow classmates in the same classroom. Please refrain from using profanity or any euphemisms for profanity. Please do not bait other commenters or personally attack them. Please do not use sarcasm in a way that can be misinterpreted negatively. And please do not make the same point over and over again. In short, please just respect the right of your colleagues to ask questions and discuss their opinions about the subject matter of our course on the discussion board. Violators of these discussion rules will simply be shut out from all class communications—email, Piazza, and office hours.
Behavior Concerns Advice Line
If you are worried about someone who is acting differently, you may use the Behavior Concerns Advice Line to discuss by phone your concerns about another individuals behavior. This service is provided through a partnership among the Office of the Dean of Students, the Counseling and Mental Health Center (CMHC), the Employee Assistance Program (EAP), and The University of Texas Police Department (UTPD). Call 512-232-5050 or visit https://besafe.utexas.edu/behavior-concerns-advice-line
Course Conent
Course creators: Philipp Krähenbühl and Chao-Yuan Wu
Instructor: Philipp Krähenbühl
TA Support: Dian Chen and Thanh An Nguyen
Using this material in your own course
Feel free to build on this course material. We only ask that you attribute us appropriately. For course material:
- If you use an entire slide deck, simply leave the attribution on the first slide
- If you use a single slide, put an attribution somewhere on the slide
For homework:
- Put the original license and attribution somewhere in the starter code.
License:
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
The code for all homework assignments (and coding examples in class) is released under the MIT license.
Getting access to this repo
If you’re teaching a similar class and would like to gain raw access to the course material, shoot me an email and I can add you. To make things as easy as possible use the subject Access to deep learning class repo
and make sure to give me your github id. Please briefly describe why you’d need access including a link to the course you’re going to teach. I’d ask you to not share any solutions online and keep the attributions in place.
All material, except for the homework solutions can be directly accessed from this webpage.
This syllabus is subject to change. Last updated: November 25, 2019