CS294-129 Designing, Visualizing and Understanding Deep Neural Networks

Reporting Issues

Piazza

Piazza is up here: https://piazza.com/berkeley/spring2018/cs294129/home Links to an external site.

Discussion Times

Time	Location
Mon 3-4pm	310 Soda
Weds 12-1pm	405 Soda
Thurs 1-2pm	320 Soda
Thurs 2-3pm	310 Soda

Course Information

Less frequently used information has moved here.

Office hours

John Canny - M 2:00-3:00, in 637 Soda
GSI (Carlos Florensa or Erin Grant)
- Tuesday 14:00 - 15:00, in 283H Soda
- Wednesday 15:00 - 16:00 in 283H Soda

Lectures Online

The class will be screen-captured and audio-recorded. Recordings will be available from bCourses and CalCentral. You should be able to get them from the "Course Captures" tab at the left.

Schedule

Date	Lecture Topic	Reading	Assignments/Section Notes
W 1/17	Introduction, Course Overview, Brief history of computer vision. pptx Download pptx and pdf Download pdf	Introduction Links to an external site. from Deep Learning Links to an external site.	Assignment 0: Discussion time poll.
M 1/22	Machine learning concepts: loss and risk, discriminative models, linear and logistic regression. pptx Download pptx and pdf Download pdf	Review: sections 1.1-1.3 and 6.6-6.9 from the CS189 book Links to an external site. (skip KL-div). Do Python/Numpy tutorial Links to an external site. if needed.	Week 2 Section notes Download Week 2 Section notes
W 1/24	SVMs, multiclass classification, softmax. Cross-validation. pptx Download pptx and pdf Download pdf	sections 6.10-6.11, 1.6 from the CS189 book Links to an external site. (skip Tikhonov)	Assignment 0.1 Assignment 1 out
M 1/29	Optimization, Stochastic Gradient Descent. pptx Download pptx and pdf Download pdf	Chapter 8 Links to an external site. of Deep Learning Links to an external site. Optimization Notes Links to an external site.Backpropagation Notes Links to an external site.
W 1/31	Backpropagation, Convolutional Networks. pptx Download pptx and pdf Download pdf	Convnet notes Links to an external site.	Week 3 Section notes Download Week 3 Section notes
M 2/5	CNN examples, activation functions, initialization, batch normalization. pptx Download pptx and pdf Download pdf	Convnet notes Links to an external site. Training Neural Networks 1 Links to an external site. Training Neural Networks 2 Links to an external site.	Project Proposal out
W 2/7	Training: Batch normalization, dropout, ensembles, hyperparameter tuning. pptx Download pptx and pdf Download pdf	Training Neural Networks 3 Links to an external site.	Week 4 Section notes
M 2/12	CNN applications and challenge datasets. pptx Download pptx and pdf Download pdf		Assignment 1 due 11pm
W 2/14	Recurrent Networks, LSTMs, applications. pptx Download pptx and pdf Download pdf	RNN chapter Links to an external site. from Deep Learning Links to an external site. Understanding LSTM Networks Links to an external site.	Project proposal due Assignment 2 out Week 5 Section notes
M 2/19	Academic Holiday
W 2/21	Visualizing Deep Networks. pptx Download pptx and pdf Download pdf	Quite a few visualizations will be covered. Browse this list Links to an external site.	Week 6 Section notes
M 2/26	Midterm 1.		Practice Midterm 1
W 2/28	Computer Vision Capstone lecture Alexei Efros Guest Lecture pdf		Week 7 Section notes
M 3/5	Semantic Models for Text. pptx Download pptx and pdf Download pdf	WordToVec Links to an external site. Skip-Thought Vectors Links to an external site.
W 3/7	Attention Networks pptx Download pptx and pdf Download pdf	Recurrent Models of Visual Attention Links to an external site.	Assignment 2 due 11pm
F 3/9			Assignment 3 out
M 3/12	Natural Language Translation. pptx Download pptx and pdf Download pdf	NMT by Jointly Learning to Align and Translate Links to an external site. Attention is All You Need Links to an external site.
W 3/14	Memory Networks, Text Question-Answering Systems. pptx Download pptx and pdf Download pdf	End-to-End Memory Networks Links to an external site. Learning End-to-End Goal Directed Dialog Links to an external site.	Week 8 Section notes Download Week 8 Section notes
M 3/19	Generative Adversarial Networks. Erin Guest Lecture. key Download key and pdf Download pdf	Generative Adversarial Networks Links to an external site.
W 3/21	Imitation Learning. pdf Download pdf	End-to-End Learning for Self-Driving Cars Links to an external site. DAGGER Links to an external site. Learning Transferable Policies Links to an external site. From Virtual Demonstration to Real-World Manipulation Links to an external site.	Week 9 Section notes Download Week 9 Section notes
M 3/26	Spring Break
W 3/28	Spring Break
M 4/2	Adversarial Networks Dawn Song and Bo Li Guest lecture pptx Download pptx and pdf Download pdf	Adversarial Examples Links to an external site. Transferability/blackbox attacks Links to an external site. Physical perturbation Links to an external site.	Assignment 3 due
W 4/4	Natural Language Capstone Michael Lewis Guest Lecture pdf Download pdf	Learning Cooperative Visual Dialog Agents Links to an external site. End-to-End Learning for Negotiation Dialogues Links to an external site. Hierarchical Text Generation and Planning for Strategic Dialogue Links to an external site. (optional)	Week 11 Section Notes Download Week 11 Section Notes
M 4/9	Midterm 2		Assignment 4 out
W 4/11	Reinforcement Learning: Policy Gradients. pdf Download pdf	Policy Gradient Methods, chapter 13 of "Reinforcement Learning" Links to an external site.
M 4/16	Reinforcement Learning: Value-based methods. Carlos Guest Lecture. pdf Download pdf	DQN paper (Nature) Links to an external site. Asynchronous Methods for Deep Reinforcement Learning Links to an external site.	Week 12 Section Notes (Policy Grad)
W 4/18	Imagination and Curiosity. pdf Download pdf	Unifying Count-Based Exploration and Intrinsic Motivation Links to an external site. Curiosity-driven Exploration by Self-supervised Prediction Links to an external site. Imagination-Augmented Agents for Deep Reinforcement Learning Links to an external site.
M 4/23	RL Capstone: Learning to Learn. Chelsea Finn Guest Lecture. pdf Download pdf	Learning to Learn blog post Matching Networks for One-Shot Learning Links to an external site. Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks Links to an external site.	Assignment 4 due 11pm (pdf to Gradescope Links to an external site. and zipfile to Bcourses)
W 4/25	Risks. Andrew Critch Guest Lecture. pdf Download pdf	AI and existential risk Links to an external site. Value alignment Links to an external site. Transparency Links to an external site.
Su 4/29			Final Project Presentation due 11pm
M 4/30	Final project presentations I	4-5:30pm in 306 Soda Hall	Final Project Presentation
W 5/2	Final project presentations II	12:30-2:30pm in 306 Soda Hall	Final Project Presentation Final Project Poster due
Th 5/3	Final project poster session	3-4:30pm in Soda 5th floor atrium	Project Poster
F 5/11			Final Project Report due

This course content is offered under a Public Domain Links to an external site. license. Content in this course can be considered under this license unless otherwise noted.