About

Course Description

Over the past several years, thanks for the development of new training rules, massive computing capabilities, and enormous training datasets, deep learning systems have redefined the state-of-the-art in object identification, face recognition, and speech recognition. Examples of modern tools include Deep Mind’s AlphaGo, Facebook’s Deep Face, and Baidu’s Deep Speech. This course will explore deep learning, multistage machine learning methods that learn representations of complex data. During a 14-week course, students will learn and implement deep learning architectures for computer vision and natural language processing applications. The models covered in the course range from Deep-Convolutional-Networks-based systems such as VGG net and Residual Net to recurrent architectures such as Recurrent Neural Networks and LSTM-based models.

Course Instructor

Teaching Assistants

ankit
Dr. Ankit Patel Huaijin (George) Chen Wanjia (Robin) Liu Josue Ortega Caro Weili Nie
abp4@rice.edu huaijin.chen@rice.edu wl22@rice.edu caro@bcm.edu wn8@rice.edu

Mission

  • Jumpstart your ability to use Deep Learning in your research
  • Less Theory, More Doing: This is not a math class (though we will cover some aspects of theory of Deep Learning near the end)

Course Overview

  • Time/Location: Wednesday 2:00 – 4:30 pm – Herzstein 212
  • Teaching Assistants: Huaijin (George) Chen <huaijin.chen@rice.edu>, Wanjia (Robin) Liu <wl22@rice.edu>, Josue Ortega Caro <caro@bcm.edu>, Weili Nie <wn8@rice.edu>
  • Course Website: http://elec576.rice.edu
  • Office Hours:
    • Due to the large enrollment, we kindly ask the student go through the steps in “Course Help” section below before requesting an office hour meeting
    • Dr. Ankit Patel: Wednesday 9:30 – 10:30 am DCH 2050 (Send an Email in advance)
    • Huaijin (George) Chen: time and location TBD (See the “Course Help” section first)
    • Wanjia (Robin) Liu: Monday 1:00 – 2:00 pm DCH 3063 (See the “Course Help” section first)

Course Help

  1. Your first point of contact should always be Piazza. If your questions are not asked before, post a question with as many details as you can. This allows you to get answers not only from the TAs or the instructor but also from your 100+ fellow classmates. More importantly, your classmates may benefit from your questions and the answers.
  2. To incentify the use of Piazza, we will award extra credits to those active Piazza members who provide valuable answers or ask good questions. So please be sure to not to use the anonymous mode if you would like to receive the extra credits.
  3. If you cannot get answers on Piazza, please then send an email with the link to the Piazza posts, as well as the best time slot that works for you, to rice_dl_course@googlegroups.com. One of the TAs will get back to you as soon as we can. Depending on the question, we will answer it on Piazza or schedule a meeting with you.

Prerequisites

  • Level: This class is for GRADUATE students, ready to do research! (I will grant exceptions to highly advanced undergrads who are ready to do research. Please see me after class)
  • Programming: Assignments will use Python, using Numpy, TensorFlow. In addition, students can use PyTorch for assignments and the final project. Willingness to learn the DL software ecosystem (Linux, packages, git, etc.)
  • Calculus: Differentiation, chain rule
  • Linear Algebra: Vectors, matrices, eigenvalues/vector, Singular Value Decomposition.
  • Probability and Statistics: random variables, multivariate Gaussians (mean covariance), Bayes Rule, Law of Total Probability, Conditional probabilities/expectations
  • Machine Learning: Principal Component Analysis, Factor Analysis, Gaussian Mixture Models, train/test splitting, cross-validation, preprocessing and working with datasets.
  • Optimization: Cost functions, taking gradients, penalty terms, implementing in code

Grading

  • Assignments: 3 for total of 70%
    • Late Policy: 2 late days to allocate, after that 25% penalty per day
  • Final Project Report + Poster Presentation: 30%
  • Extra Credits will be awarded to those who ask or answer valueable questions on Piazza
  • Graders: TBD
  • Note: initial scores on assignments will be scaled based on the distribution of student performance.

University Disability Accommodation Policy

Any student with a documented disability needing academic adjustments or accommodations is requested to speak with the course instructor during the first two weeks of class. All discussions will remain confidential. Students with disabilities should also contact Disabled Student Services in Allen Center.

Any student with a disability requiring accommodations in this course is encouraged to contact the instructor after class or during office hours. Additionally, students should contact Disability Support Services in Allen Center.

If you have a documented disability that will impact your work in this class, please contact the course’s instructor to discuss your needs. Additionally, you will need to register with the Disability Support Services Office in Allen Center.