Hacking Deep Learning:

Beyond the Hype

Lecturer:

help from:

Calendar

First Day of Class: June 1 2015
Last Day of Class: August 31 2015

Cancelled Classes (tentative):
July 13 2015 (Invited Talk, IPAM, UCLA.),
Note that some of these classes may be covered by Guest Lectures.

Office Hours: spamhaus (Tuesday) Office Phone: 212.998.3464 Email Address: mishra@nyu.edu

Day, Time and Place:

Mondays, 6:00-7:30pm EST, 715 Broadway, 10th floor. (Further Discussions over Dinner around 8:00pm).

Credits for Course:

Prerequisites:

Mathematical Maturity, Programming and Algorithms

Grading Policy:

One or more publications on "Scholarly and Critical Review of Deep Learning."

Home Work/DataSet [ Cancer Data Set || HeroX Prize ] Talks [ Nvidia Class |:|.. ] Reading Assignment [ Lecture #1: Chapters 1 & 2: (Linear Algebra, Probability and Information Theory, Numerical Computation, Machine Learning) & (Deep Networks, Regularization, Numerical Optimization) & (Convolutional Networks, Recurrent Nets) || Lecture #2: Chapter 5: (Machine Learning): Lecturer: D. Kasofsky. || Lecture #2: Chapter 6: (Machine Learning): Lecturer: J. Lima. || Lecture #3: Chapter 7 & 8: (Machine Learning): Lecturer: P. Gupta and V.T. Rajan. || ] Notes [ Note #1 || Note #2 || Note #3 || Note #4 || ]

Syllabus:

1a) We will read the following book.
1b) Also see the review article .
1c) Also see interesting projects from Stanford.
2) We will read a few related papers. Understand how Deep Learning may be related to other topics in Data Sciences... Particular Emphasis: AdTech (DMP and Attribution), FinTech (compare to Kensho, Palantir, Sentient), Verifier-Recommenders (Cyber Security, Privacy, AI to obfuscate AI), Caner, Linguistics, ...
3) Each participant will implement a deep learning application with open source code (e.g., caffe or alexnet). In preparation for this we will read the following paper.
4) Jointly with MIT, Cornell and UW, we have submitted a grant proposal to NSF to understand interconnections among Manifold Learning, Deep Learning and PH-Learning (under review).
5) For cancer applications, we have been funded by NCI to start a center (focus on causality & topology). Also we are organizing a summer school.

Fequently Asked Questions (FAQs)

Q1. Can I attend this class?
A1. Yes, but... only if you have been invited.

Q2. Can I get a grade/credit?
A2. No, this is an informal class. But we will write paper(s), which will show how DEEP-ly we understand these topics. We can create some programming projects, which can be evaluated and displayed on a leader-board. We can also take up some problems in Kaggle and use their evaluations to demonstrate depth of our understanding.
Any other ideas?

Required Text(s):

Bud Mishra
September 1 2003