Συντάχθηκε 27-07-2018 10:02
Ημερομηνία: Τρίτη 31/7/2018
Ώρα: 12:00
Αίθουσα: Αμφιθέατρο του Κτ. Επιστημών
Title: Overcoming the Challenges of Distributed Learning
Speaker: Prof. Dimitris Papailiopoulos
University of Wisconsin-Madison
http://papail.io/
Abstract:
In this talk, I will highlight a few key challenges that limit our
capacity to effectively deploy machine learning solutions in real
distributed systems. I will first focus on communication bottlenecks
during model training, and discuss how they lead to poor speedup gains
when scaling out to hundreds of compute nodes. I will present theoretical
insights which suggest that we can only overcome these challenges either
by building new classes of neural networks, or by designing novel training
algorithms that require far less communication. We will then focus on
issues of robustness and discuss how model training is susceptible to
hardware failures and adversarial attacks. I will explain how simple
algebraic ideas, borrowed from coding theory, can be used to enable robust
distributed training. I will conclude with several open problems that lie
in the intersection of machine learning, optimization, and distributed
systems.
Short CV:
Dimitris Papailiopoulos is an Assistant Professor of Electrical and
Computer Engineering and Computer Sciences (by courtesy) at the University
of Wisconsin-Madison, a faculty fellow of the Grainger Institute for
Engineering, and a faculty affiliate at the Wisconsin Institute for
Discovery. Between 2014 and 2016, Dimitris was a postdoctoral researcher
at UC Berkeley and a member of the AMPLab. His research interests span
machine learning, information theory, and distributed systems, with a
current focus on communication-avoiding training algorithms and
coding-theoretic techniques for robust large-scale machine learning.
Dimitris earned his Ph.D. in ECE from UT Austin in 2014, under the
supervision of Alex Dimakis. In 2007, he received his ECE Diploma and in
2009 his M.Sc. degree from the Technical University of Crete, in Greece.
In 2015, Dimitris received the IEEE Signal Processing Society, Young
Author Best Paper Award. In 2018, he co-founded and was Program co-Chair
for SysML, a new conference that targets research at the intersection of
machine learning and systems.