Advanced topics in deep learning. Introduction to deep neural networks, model drift, and adversarial learning.
Lecture Time: TR 6pm - 7:15pm
Location: online: Blackboard Collaborate
Amanda Fernandez
Email: Amanda dot Fernandez at utsa dot edu
Office: NPB 3.324
Hours: TBA, or by appointment
http://www.cs.utsa.edu/~fernandez
TBA
Email:
Office:
Hours TBA, and by appointment
Exams: Thursday Februrary 25th and Thursday April 22nd
There will not be a final examination for this course.
All exams are held in the same classroom as the lecture.
No materials or electronics will be permitted at exams. Collaboration on exams is strictly prohibited and in violation of academic integrity policies.
There are no make-up exams available, unless coordinated with your instructor 1 week in advance of the midterm. Students must attend the exams for the section in which they are enrolled. Contact the instructor in advance if other arrangements need to be made.
Deep Learning, by Ian Goodfellow, Yoshua Bengio, and Aaron Courville
This textbook is free to view online, or you may purchase a copy here.
We will not follow the book chapter-by-chapter, but rather use it as a reference. See the tentative schedule for a listing of chapters and other resources related to each topic.
For CS 4263: CS 3753, CS 3793. CS 4233, or CS4253
For CS 5483: CS 5163
The course will be using Blackboard (http://utsa.blackboard.com/) for grade reporting, program submissions and tentatively for online quizzes.
All course announcements will be made through Blackboard, including notices of change of office hours, class cancellation, and reminders. It is your responsibility to connect Blackboard notifications with your preferred email, and to check that email frequently in order to stay up to date accordingly.
A calendar of course events will be maintained on Blackboard - you are encouraged to sync your personal calendar with Blackboard in order to keep track of deadlines and events.
If you will miss a lecture do not contact your instructor unless you will be missing several classes, an exam, or your presenteation due to a university-sanctioned excuse. Instead, check Blackboard for updates and follow up with a classmate for missed content.
There will be several quizzes administered in class throughout the semester. The dates for each will be advertised well in advance, in order to provide time to plan.
Late quizzes will not be accepted. As all quiz dates will be posted in advance, no extensions will be provided without a university-sanctioned excuse.
At the start of the semester, graduate students will be assigned a recent paper of relevance to each week in the course. If enrollment allows, students will be permitted to form teams of 2 to complete this exercise. Students will present the paper in class, briefly with visuals, then guide the class in a discussion.
Undergraduate students will participate in the discussion and bring a list of questions for every paper.
Students will investigate an advanced topic in deep learning, applying concepts learned in the course, and implement a deep learning model. They will present their model and findings in class.
Requirements will differ between undergradute and graduate students. Graduate students will work individually, while undergradute students will be encouraged to form teams (size dependent upon enrollment).
All projects will include the following deliverables:
- Proposal - document and brief presentation of proposed deep learning project for the semester.
- Update 1 - updated proposal indicating related works and proposed approach.
- Update 2 - updated report indicating implementation details.
- Update 3 - updated report including preliminary results.
- Presentation - in class
- Final Report - finalized version of report writeup, include evaluation and results.
Further details and deadlines will be provided on Blackboard and discussed in class.
Student Disability Services (SDS) at UTSA promotes equal access for all university programs and activities for students with disabilities.
If you are registered with SDS and choose to utilize their services in this course, it is your responsibility to respond to the instructor when contacted in order to make a plan for accommodations.
Please note that the Test Center is not open during the evening - exams taking place during evening courses will need to be coordinated at least 2 weeks in advance with the instructor.
Common syllabus information and links can be found at http://provost.utsa.edu/syllabus.asp.
For questions or concerns regarding the drop date, withdrawal dates, or academic holidays, please refer to the academic calendar from the UTSA Registrar: https://www.utsa.edu/registrar/
The integrity of a university degree depends on the integrity of the work done for that degree by each student. The University expects every student to maintain a high standard of individual honor in their scholastic work.
You must write your own code. Because patterns of cheating do not always become apparent until after several assignments have been completed, you should be aware all of your submissionss are available to your instructor on Blackboard.
Further information on UTSA's policies regarding academic dishonesty can be found in UTSA's Student Code of Conduct, Section 203.