Fall 2020

September 28, 2020

Distributed Machine Learning over Networks

Speaker: Francis Bach (Institut National de Recherche en Sciences et Technologies du Numérique (INRIA))

Event Recording:
The success of machine learning models is in part due to their capacity to train on large amounts of data. Distributed systems are the common way to process more data than one computer can store, but they can also be used to increase the...

October 19, 2020

SOAP: New Breakthroughs in Scheduling Theory

Speaker: Mor Harchol-Balter (Carnegie Mellon University)

Event Recording:
Scheduling policies are at the heart of computer systems. The right scheduling policy can dramatically reduce response times, ensure fairness, provide class-based priority, etc., without requiring additional resources. While stochastic...

October 26, 2020

Hadamard Differential Calculus and Applications

Speaker: Michel Delfour (Université de Montréal)

Event Recording:
The Hadamard differential was introduced in 1923 by Hadamard and promoted in 1937 by Fréchet who extended it to vector spaces of functions. Infinite dimension is equivalent to the Fréchet differential introduced in 1911, but in function...

November 9, 2020

Thresholds for Reliable Computation with Noisy Gates, and Applications in Quantum Nonlocality

Speaker: Mary Wootters (Stanford)

Event Recording:
Suppose you are given a bunch of logic gates -- ANDs, XORs, and NOTs -- but they are noisy, and with some probability will return the incorrect answer.  At what noise levels is reliable computation possible using these gates?  We investigate...

November 23, 2020

Safe and Efficient Exploration in Reinforcement Learning

Speaker: Andreas Krause (ETH Zürich)

Event Recording:
At the heart of Reinforcement Learning lies the challenge of trading exploration -- collecting data for identifying better models -- and exploitation -- using the estimate to make decisions.  In simulated environments (e.g., games),...

November 30, 2020

Long Duration Autonomy With Applications to Persistent Environmental Monitoring

Speaker: Magnus Egerstedt (Georgia Institute of Technology)

Event Recording:
When robots are to be deployed over long time scales, optimality should take a backseat to “survivability”, i.e., it is more important that the robots do not break or completely deplete their energy sources than that they perform certain...