Workshop on Machine Learning for Network Data
January 29, 2019
Registration has closed.
Synopsis:
The goal of this workshop is to discuss our incipient understanding of how we can take advantage of the underlying irregular structure of a graph signal to produce transformations that take advantage of the signal's internal symmetries. In particular, we will cover advances on graph signal processing, graph neural networks, and graph scattering transforms:
The State of the Art in Machine Learning. Machine learning is at a crucial juncture where the success achieved in some problems has opened up the possibility of developing a new industry. To make this possibility a reality, what are the challenges we need to overcome in the next decade? The processing of signals supported in irregular domains is one of these fundamental challenges but in this workshop we want to discuss what other advances are needed to push machine learning forward.
Graph Signal Processing. Graph signal processing is an emerging field whose goal is to extend classical signal processing tools to signals supported on graphs. Central to this effort are generalizations of Fourier transforms, wavelets, and convolutions to irregular domains. Graph Signal Processing has had remarkable success in classical processing tasks such as filtering, systems identification, and sampling. How does the field need to evolve to address the higher cognitive questions that are typical of machine learning?
Graph Neural Networks. Graph neural networks are layered architectures akin to Convolutional Neural Networks that use graph convolutions in lieu of standard convolutions. There are several architectures that have been proposed in the literature that build on different possible ways in which convolutions can be generalized. Naturally, graph neural networks lag behind convolutional neural networks in the richness of their architectures and their application domains. How can we expand the reach of graph neural networks into new application domains? How can we expand the range of architectures to process signals supported on graphs?
Graph Scattering Transforms. Why are convolutional neural networks useful information processing architectures? Scattering transforms provide an answer to this question in the form of the notion of Lipschitz stability with respect to deformations. Graph scattering transforms have been introduced as an attempt to answer the analogous questions for graph neural networks. While preliminary stability results have been established, our understanding of stability properties of graph neural networks remains limited. How can we further our understanding of stability properties of graph neural networks? What other properties of graph neural networks are important to understand their behavior?
Program:
9:00am - 9:10am: Arrival
9:10am - 10:00am: Yann LeCun, New York University, Facebook
10:00am - 10:50am: Brian Sadler, Army Research Laboratory
10:50am - 12:10pm: Session 1- Fundamental properties
Joan Bruna, New York University
Risi Kondor, University of Chicago
Matthew Hirn, Michigan State University
Gilad Lerman, University of Minnesota
12:10pm - 1:00pm: Lunch and Breakout Discussions
1:00pm - 2:20pm: Session 2- Architectures for information processing on graphs and manifolds
Alejandro Ribeiro, University of Pennsylvania
Le Song, Georgia Tech
Kostas Daniilidis, University of Pennsylvania
Anima Anandkumar, Caltech
2:20pm - 3:10pm: Coffee and Breakout Discussions
3:10pm - 5:10pm: Session 3- Network data
Rajesh Ranganath, New York University
Jure Leskovec, Stanford
Aylin Yener, Penn State University
He He, Amazon Web Services
Arthur Szlam, Facebook Research
Vikram Krishnamurthy, Cornell Tech
5:10pm - 6:00pm: Breakout Discussions
6:00pm - 7:30pm: Dinner and Poster Session
Organizers:
Joan Bruna - New York University
Alejandro Ribeiro - University of Pennsylvania
Location:
NYU Center for Data Science (CDS)
60 5th Avenue
New York NY 10011
Please proceed to the 7th floor and follow directions to the CDS Open Space.
Hotels:
Washington Square Hotel (closest hotel to the workshop location)
103 Waverly Pl, New York, NY 10011
Reservations Manager: Beverley Harris, 212-777-9515
124 E 28th St, New York, NY 10016
Courtyard by Marriott New York Manhattan/Chelsea
135 W 30th St, New York, NY 10001
Club Quarters Hotel, Times Square - Midtown
40 W 45th St, New York, NY 10036
SpringHill Suites by Marriott New York Midtown Manhattan/Fifth Avenue
25 W 37th St, New York, NY 10018
Residence Inn New York Manhattan/Central Park
1717 Broadway, New York, NY 10019
Participants:
Rick Blum, Lehigh
Addison Bohannon, Army Research Laboratory
Ceyhun Eksin, Texas A&M
Yonina Eldar, Technion-Israel Institute of Technology
Nageen Himayat, Intel
Purush Iyer, Army Research Office
Soummya Kar, Carnegie Mellon University
Alec Koppel, Army Research Laboratory
Jelena Kovacevic, Carnegie Mellon University
Mauro Magionni, Johns Hopkins University
Gonzalo Mateos, Rochester
Navid Naderializadeh, Intel
Kumar Vijay Mishra, Army Research Laboratory
Mike Rabbat, Facebook Research
Gesualdo Scutari, Purdue
Santiago Segarra, Rice
Eric Vanden-Eijnden, New York University
Dongmian Zou, University of Minnesota