[Colloquium] REMINDER: 8/16 Thesis Defense: Shubhendu Trivedi, TTIC

[Mary Marre via Colloquium]

 *When:*      Thursday, August 16th at 2:30 pm

*Where:    * TTIC, 6045 S. Kenwood Avenue, 5th Floor, Room 526

*Who:        *Shubhendu Trivedi, TTIC


*Title:        *Discriminative Learning of Similarity and Group Equivariant
Representations.

*Abstract:*
One of the most fundamental problems in machine learning is to compare
examples: Given a pair of objects we want to return a value which indicates
degree of (dis)similarity. Similarity is often task specific, and
pre-defined distances can perform poorly, leading to work in metric
learning. However, being able to learn a similarity-sensitive distance
function also presupposes access to a rich, discriminative representation
for the objects at hand.  In this talk we present contributions towards
both ends.

In the first part of the thesis talk, assuming good representations for the
data, we present a formulation for metric learning that makes a more direct
attempt to optimize for the k-NN accuracy as compared to prior work. Our
approach considers the choice of k neighbors as a discrete valued latent
variable, and casts the metric learning problem as a large margin
structured prediction problem. We present experiments comparing to a suite
of popular metric learning methods. We also present extensions of this
formulation to metric learning for kNN regression, and discriminative
learning of Hamming distance.

In the second part, we consider a situation where we are on a limited
computational budget i.e. optimizing over a space of possible metrics would
be infeasible, but access to a label aware distance metric is still
desirable. We present a simple, and computationally inexpensive approach
for estimating a well motivated metric that relies only on gradient
estimates, we also discuss theoretical as well as experimental results of
using this approach in regression and multiclass settings.

In the final part, we address representational issues, considering group
equivariant neural networks (GCNNs). Equivariance to symmetry
transformations is explicitly encoded in GCNNs; a classical CNN being the
simplest example. Following recent work by Kondor et. al., we present a
SO(3)-equivariant neural network architecture for spherical data, that
operates entirely in Fourier space, while using tensor products and the
Clebsch-Gordan decomposition as the only source of non-linearity. We report
state of the art results, and emphasize the wider applicability of our
approach, in that it also provides a formalism for the design of fully
Fourier neural networks that are equivariant to the action of any
continuous compact group.


*Thesis advisor:* Gregory Shakhnarovich





Mary C. Marre
Administrative Assistant
*Toyota Technological Institute*
*6045 S. Kenwood Avenue*
*Room 523*
*Chicago, IL  60637*
*p:(773) 834-1757*
*f: (773) 357-6970*
*mmarre at ttic.edu <mmarre at ttic.edu>*

On Tue, Aug 14, 2018 at 8:44 AM, Mary Marre <mmarre at ttic.edu> wrote:

> *When:*      Thursday, August 16th at 2:30 pm
>
> *Where:    * TTIC, 6045 S. Kenwood Avenue, 5th Floor, Room 526
>
> *Who:        *Shubhendu Trivedi, TTIC
>
>
> *Title:        *Discriminative Learning of Similarity and Group
> Equivariant Representations.
>
> *Abstract:*
> One of the most fundamental problems in machine learning is to compare
> examples: Given a pair of objects we want to return a value which indicates
> degree of (dis)similarity. Similarity is often task specific, and
> pre-defined distances can perform poorly, leading to work in metric
> learning. However, being able to learn a similarity-sensitive distance
> function also presupposes access to a rich, discriminative representation
> for the objects at hand.  In this talk we present contributions towards
> both ends.
>
> In the first part of the thesis talk, assuming good representations for
> the data, we present a formulation for metric learning that makes a more
> direct attempt to optimize for the k-NN accuracy as compared to prior work.
> Our approach considers the choice of k neighbors as a discrete valued
> latent variable, and casts the metric learning problem as a large margin
> structured prediction problem. We present experiments comparing to a suite
> of popular metric learning methods. We also present extensions of this
> formulation to metric learning for kNN regression, and discriminative
> learning of Hamming distance.
>
> In the second part, we consider a situation where we are on a limited
> computational budget i.e. optimizing over a space of possible metrics would
> be infeasible, but access to a label aware distance metric is still
> desirable. We present a simple, and computationally inexpensive approach
> for estimating a well motivated metric that relies only on gradient
> estimates, we also discuss theoretical as well as experimental results of
> using this approach in regression and multiclass settings.
>
> In the final part, we address representational issues, considering group
> equivariant neural networks (GCNNs). Equivariance to symmetry
> transformations is explicitly encoded in GCNNs; a classical CNN being the
> simplest example. Following recent work by Kondor et. al., we present a
> SO(3)-equivariant neural network architecture for spherical data, that
> operates entirely in Fourier space, while using tensor products and the
> Clebsch-Gordan decomposition as the only source of non-linearity. We report
> state of the art results, and emphasize the wider applicability of our
> approach, in that it also provides a formalism for the design of fully
> Fourier neural networks that are equivariant to the action of any
> continuous compact group.
>
>
> *Thesis advisor:* Gregory Shakhnarovich
>
>
>
>
>
> Mary C. Marre
> Administrative Assistant
> *Toyota Technological Institute*
> *6045 S. Kenwood Avenue*
> *Room 523*
> *Chicago, IL  60637*
> *p:(773) 834-1757*
> *f: (773) 357-6970*
> *mmarre at ttic.edu <mmarre at ttic.edu>*
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20180816/7d92366a/attachment.html>