Videos: Workshop on Learning Data Representation: Hierarchies and Invariance

November 22, 2013 - 9:00 am to November 24, 2013 - 6:00 pm

The workshop “Learning Data Representation: Hierarchies and Invariance” was held at the McGovern Institute for Brain Research at MIT (Bldg. 46), from November 22-24, 2013.

The goal of the meeting is to investigate advances and challenges in learning “good representations” from data, in particular representations that can reduce the complexity of later supervised learning stages. The meeting will gather experts in the field to discuss current and future challenges for the theory and applications of learning representations.

This  workshop was organized by Tomaso Poggio and Lorenzo Rosasco and sponsored by the Laboratory for Computational and Statistical Learning (LCSL), joint between the Istituto Italiano di Tecnologia and Massachusetts Institute of Technology and the Center for Brains, Minds and Machines (CBMM).

Read PDF version of workshop’s agenda>

Visit event website for list of participants and detailed agenda>


Speakers have kindly permitted us to share their presentation slides along with videos of most of the talks. Please note a few talks are unavailable due to technical difficulties experienced with the AV equipment during the event.

 

Friday November 22, 2013

Introduction to the workshop

Tomaso Poggio, Director of the Center for Brains, Minds and Machines (CBMM), presented an introduction to CBMM and introduced workshop organizer Dr. Lorenzo Rosasco.

Click here to see slides>

Lorenzo Rosasco (LCSL, MIT) welcomes guests to the Workshop on Learning Data Representation: Hierarchies and Invariance.

 

Session 1: Early Features in Vision

Chair: Carlo Tomasi (Duke U.)

Filters and other potions for early vision and recognition, Pietro Perona (Caltech)
Click here to see slides>

Transformations in early vision from a symmetry argument, Charles Stevens (Salk Institute)
Click here to see slides.

Hierarchical models of the visual cortex, Thomas Serre (Brown U.)
* This video is not available due to technical issues with the recording
Click here to see slides>

Panel discussion: Early Features in Vision
Moderator: Carlo Tomasi (Duke U.)
Panel: William Freeman (MIT CSAIL), Charles Stevens (Salk Institute), Thomas Serre (Brown University), Giulio Sandini (IIT), and Pietro Perona (Caltech.)

Early Features in Vision, Carlo Tomasi (Duke U.)
Click here to see slides>

Eulerian Video Magnification: Engineering Applications of a V1-like Image Representation, Bill Freeman (MIT CSAIL)
 Click here to see slides>

Early Visions and Actions, Giulio Sandini (IIT)
Click here to see slides>

Saturday November 23, 2013

Session 2: Learning Features and Representations

Chair: Alessandro Verri (Università degli Studi di Genova)

Latent structure beyond sparse codes, Benjamin Recht (UC Berkeley)
Click here to see slides>

Dictionary learning: 3+ snippets on learning features, Guillermo Sapiro (Duke U.)
Click here to see slides>

Panel discussion
Moderator: Alessandro Verri (Università degli Studi di Genova)
Panel: Benjamin Recht (UC Berkeley), Lorenzo Rosasco (LCSL, MIT), Misha Belkin (Ohio State), Joachim Buhmann (ETH Zurich), and Guillermo Sapiro (Duke).

Learning Features and representationsLorenzo Rosasco (LCSL, MIT)
Click here to see slides>

Features and Representations, Misha Belkin (Ohio State U.)
Click here to see slides>

Learning Features and Representations, Joachim Buhmann (ETH Zurich)
Click here to see slides>

Session 3: Learning Invariances and Hierarchies

Chair: James DiCarlo (MIT)

Learning invariant feature hierarchies, Yann LeCun (NYU)
Click here to see slides>

Scattering bricks to build invariants for perception (part 1), Stéphane Mallat (École Polytechnique)

Scattering bricks to build invariants for perception (part 2), Stéphane Mallat (École Polytechnique)
Click here to see slides>

M-theory, Tomaso Poggio (MIT, CBMM)
Click here to see slides>

Panel discussion
Moderator: James DiCarlo (MIT)
Panel: Yann LeCun (NYU), Stéphane Mallat (École Polytechnique), Tomaso Poggio (MIT), Pierre Baldi (U.C.Irvine), and Maximilian Riesenhuber (Georgetown).

Going After Object Recognition to Discover How the Ventral Stream Works, James DiCarlo (MIT)
Click here to see slides>

Learning Invariances and HierarchiesPierre Baldi (U.C.Irvine)
Click here to see slides>

Note: Maximilian Riesenhuber (Georgetown) participated in this panel. Unfortunately, the video and slides of his talk are not available.

Sunday November 24, 2013

Session 4: Beyond Feedforward Architectures

Chair: Josh Tenenbaum (MIT, CBMM)

The role of mobility and control in the inference of representations, Stefano Soatto (UCLA)
Click here to see slides>

Atoms of recognition, Shimon Ullman (MIT/Weizmann Inst.)
Click here to see slides>

Compositional models: complexity of representation and inference, Alan Yuille (UCLA)

Panel discussion (video)
Modertor: Josh Tenenbaum (MIT, CBMM)

Panel: Stefano Soatto (UCLA), Shimon Ulllman (MIT/Weizmann Inst.), Alan Yuille (UCLA), and Russ Salakhutdinoff (U. Toronto).

What Makes a Good Representation? From Invariance to Causality, Josh Tenenbaum (MIT, CBMM)

Representation Learning, Russ Salakhutdinov (U. Toronto)
Click here to see slides>

Final remarksTomaso Poggio (MIT, CBMM)
Click here to see slides>

Details

Date: 
November 22, 2013 to November 24, 2013
Time: 
9:00 am to 6:00 pm
Venue: 
McGovern Institute for Brain Research at MIT (Bldg. 46)