Logistics
Table of contents
- Course Format
- Course Time/Location
- Course Websites
- Office Hour
- Email Etiquette
- Textbook
- Assignments, Projects, and Grading
- Campus Policies
- Resources for Support and Learning
- Course Schedule
Course Format
Students can attend the lecture in the classroom or via zoom broadcasting. The lectures will be recorded and uploaded to canvas.
Course Time/Location
- Time: Tues/Thurs, 8-9:20 am
- Location:
- Classroom: Warren Lecture Hall 2114
- Zoom: https://ucsd.zoom.us/j/92870101766
- Subscribe to calendar - needs UCSD Google account
Course Websites
Office Hour
- Prof. Office Hour: Tuesday 10:00-10:30 AM
- TA Office Hour: Wednesday 4:00-5:00 PM
Email Etiquette
- Please include “DSC 291” in the title to make sure we receive your email.
- For emails received between Mon-Fri, 10-4 pm, we will try to answer within the same day; else, we will try to answer on the next workday.
Textbook
There is no official textbook for this course. The primary contents are from research papers. Free resource to start in deep learning:
[1] Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep learning. MIT press, 2016. https://www.deeplearningbook.org/
Assignments, Projects, and Grading
Students are expected to complete 3 homework assignments, 1 course project and 1-2 scribing. HWs allows a group of 1-2 students/group. Course project allows 2-3 students/group.
- Lectures will be scribed via Latex.
- HWs are expected to be typed in latex (except the coding part).
- The course project will be formatted in latex, and templates will be provided.
- For tutorials on Python, PyTorch, and latex, please refer to this list.
Summary of Grading Criteria
| Assignment | Points | Weight | Due Date |
|---|---|---|---|
| HW 1 | 100 | 15% | 2.5 weeks after release |
| HW 2 | 100 | 15% | 2.5 weeks after release |
| HW 3 | 100 | 15% | 2.5 weeks after release |
| Lecture Scribe | 100 | 15% | 3 days after lecture |
| Course Project | 100 | 40% | midterm checkpoint due May 23, 1 pm final demo/reports due Jun 15, 2 pm |
| 100% |
Late Policy
We understand there might be some circumstances where you may need more time to complete assignments and the late policy is exactly designed for this scenario for you to plan accordingly. Please do not email us for extensions, as we will grade according to your submission time.
- The penalty for HW1-3, Lecture Scribe and Course Project midterm checkpoint is 10% per day.
- For the Course project final demo/reports, no late submissions will be accepted as the final grades need to be submitted to the Registrar in time.
Grading Scale
- A+ = 98-100%
- A = 93-97%
- A- = 90-92%
- B+ = 86-89%
- B = 83-85%
- B- = 80-82%
- C = 70-79%
- D = 60-69%
- F = below 59%
Campus Policies
- UC San Diego Principles of Community
- UC San Diego Policy on Integrity of Scholarship
- Religious Accommodation
- Nondiscrimination and Harassment
- UC San Diego Student Conduct Code
Resources for Support and Learning
There are a variety of resources available to students at UC San Diego, which are designed to address needs and enhance the student experience. We encourage you to check them out.
| Learning and Academic Support | |
|---|---|
| Ask a Librarian: Library Support Chat or make an appointment with a librarian to focus on your research needs Course Reserves, Connecting from Off-Campus and Research Support Find supplemental course materials First Gen Student Success Coaching Program Peer mentor program that provides students with information, resources, and support in meeting their goals Office of Academic Support & Instructional Services (OASIS) Intellectual and personal development support | Writing Hub Services in the Teaching + Learning Commons One-on-one online writing tutoring and workshops on key writing topics Supplemental Instruction Peer-assisted study sessions through the Academic Achievement Hub to improve success in historically challenging courses Tutoring – Content Drop-in and online tutoring through the Academic Achievement Hub Tutoring – Learning Strategies Address learning challenges with a metacognitive approach |
| Support for Well-being and Inclusion | |
| Basic Needs at UCSD Any student who has difficulty accessing sufficient food to eat every day, or who lacks a safe and stable place to live is encouraged to contact: foodpantry@.ucsd.edu | basicneeds@ucsd.edu | (858) 246-2632 Counseling and Psychological Services Confidential counseling and consultations for psychiatric service and mental health programming Triton Concern Line Report students of concern: (858) 246-1111 Office for Students with Disabilities (OSD) Supports students with disabilities and accessibility across campus | Community and Resource Centers Office of Equity, Diversity, and Inclusion As part of the Office of Equity, Diversity, and Inclusion, the campus community centers provide programs and resources for students and contribute toward the evolution of a socially just campus (858) 822-3542 | diversity@ucsd.edu Get Involved Student organizations, clubs, service opportunities, and many other ways to connect with others on campus Undocumented Student Services Programs and services are designed to help students overcome obstacles that arise from their immigration status and support them through personal and academic excellence |
Course Schedule
A detailed course schedule can be found in the Course Timeline page with slides and recording links (will be updated throughout the quarter). An overall schedule can be found below:
| Week | Title / Topic | Due dates |
|---|---|---|
| 1 | Intro & Machine Learning Basics | |
| 2 | Machine Learning Basics | |
| 3 | Robustness – overview & attack | HW 1 out, Project instruction out |
| 4 | Robustness – attack & verification | |
| 5 | Robustness – verification | |
| 6 | Robustness – verification & defense | HW 1 due, HW 2 out |
| 7 | Midterm presentation | |
| 8 | Midterm presentation / Interpretability | Midterm report due, HW 2 due, HW 3 out |
| 9 | Interpretability – post-hoc explanations | |
| 10 | Interpretability – interpretable models | |
| 11 | Final project demo | HW 3 due, Project final report, slides, codes due |