Fwd: Talk tomorrow Thu 11AM in E2 599 by Daniel Lowd, University of Oregon

[Lise Getoor]

Late breaking information about talk today that may be of interest to some folks on this list!

---------- Forwarded message ----------
From: Lise Getoor <get...@soe.ucsc.edu>
Date: Wed, Nov 19, 2014 at 8:06 AM
Subject: Talk tomorrow Thu 11AM in E2 599 by Daniel Lowd, University of Oregon
To: "ml-resea...@ucsc.edu" <ml-resea...@ucsc.edu>

hi all,

Daniel Lowd will be visiting tomorrow (Thu) and giving a talk on structure learning for sum product networks.   Details here:
https://www.soe.ucsc.edu/events/event/3688

Please join in if you can!

Daniel Lowd, University of Oregon
Thursday, November 20, 11:00 AM to 12:15 PM
Location: Engineering 2 room 599
Hosted By Professor Lise Getoor

ABSTRACT:
Probabilistic graphical models have been applied to many domains, including
computer vision, natural language processing, and bioinformatics.  However,
their effectiveness is limited by the complexity of inference, which is
generally intractable.  An appealing alternative is to work with tractable
probabilistic models, in which exact inference is efficient.  Sum-product
networks (SPNs) are a deep, tractable probabilistic representation that
generalize many other tractable model classes.  SPNs have achieved
state-of-the-art results on computer vision and density estimation
problems, but selecting a good structure for an SPN is challenging.

In this talk, I will provide a brief introduction to SPNs and then discuss
several recent approaches to learning SPN structures from data.  The first
approach is to adapt standard graphical model structure learning
algorithms, resulting in SPNs that represent tractable graphical models.
The second approach is to recursively cluster instances and variables,
resulting in SPNs that represent hierarchical mixture models.  These two
approaches capture different types of patterns in the data.  The ID-SPN
algorithm uses a combined approach, leading to much better structures on a
variety of benchmark domains.  In many cases, ID-SPN learns SPNs that are
more accurate than intractable Bayesian networks, demonstrating that SPNs
can maintain tractability without sacrificing accuracy.

BIO:
Daniel Lowd is an Assistant Professor in the Department of Computer and
Information Science at the University of Oregon.  His research interests
include learning and inference with probabilistic graphical models,
adversarial machine learning, and statistical relational machine learning.
He received his Ph.D. in 2010 from the University of Washington.  He
maintains Libra, an open-source toolkit for Learning and Inference in
Bayesian networks, Random fields, and Arithmetic circuits, and recently
received a Google Faculty Award for related work on learning tractable
models.

[Ram Akella]

Hi Lisa,

Is there a way to stream and record these talks on a regular basis? Ccing Pat Mantey for his assistance and inputs.

They are very interesting, but adapting schedules, especially late breaking talks, os challenging.

Thank you,

Ram

Sent from my iPhone

--
You received this message because you are subscribed to the Google Groups "Data Science at UCSC" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dssc+uns...@soe.ucsc.edu.
To post to this group, send email to ds...@soe.ucsc.edu.