INFORMS Boston Chapter Talk - April 25 2018 at 6:30 PM
RSVP by Monday, April 23, 2018 at 6 pm
Our next INFORMS Boston meeting will be Wednesday, April 25th, 2018 and feature a talk by Serdar Kadıoğlu regarding a new method for instance-specific algorithm configuration (ISAC). MITRE is asking that if you plan to attend please RSVP by Monday, April 16, 2018 at 6 pm by sending an email to lservi@mitre.org indicating your 1) name, 2) email, (3) company or university, (4) whether you are a US citizen and, if not, country of citizenship.
We are looking forward to seeing you!
Les Servi,
President, INFORMS Boston Chapter
Date/Time:
Wednesday, April 25th, 2018
6:30 pm Reception
7:00 pm - 8:00 pm Talk
Location:
The MITRE Corporation
M Building, 202 Burlington Road
Bedford, MA
Title: ISAC – Instance-Specific Algorithm Configuration
Speaker: Serdar Kadıoğlu
Abstract:
We present a new method for instance-specific algorithm configuration (ISAC). It is based on the integration of the algorithm configuration system GGA and the recently proposed stochastic offline programming paradigm. ISAC is provided a solver with categorical, ordinal, and/or continuous parameters, a training benchmark set of input instances for that solver, and an algorithm that computes a feature vector that characterizes any given instance. ISAC then provides high quality parameter settings for any new input instance. Experiments on a variety of different constrained optimization and constraint satisfaction solvers show that automatic algorithm configuration vastly outperforms manual tuning. Moreover, we show that instance-specific tuning frequently leads to significant speed-ups over instance-oblivious configurations.
Bio:
Serdar Kadıoğlu is leading an AI/ML team at Fidelity Investments and an adjunct faculty member in the Dept. of Computer Science at Brown University. Previously, he was leading the Advanced Constraint Technology R&D team at Oracle. His algorithmic research is focused on Artificial Intelligence, Machine Learning, and Optimization, while my practical interests are in building robust, scalable, data-driven products that address business-critical problems.
Dr. Kadıoğlu received his Ph.D. at Brown University under the supervision of Meinolf Sellmann. His thesis work explored deterministic inference methods for constraint satisfaction, search protocols for constrained optimization, and stochastic local search. He also worked at Adobe developing an AI-based test case generator for the ActionScript virtual machine and was a researcher at Astra Group on Combinatorial Optimization hosted by Prof. Pierre Flener.
------------------------------
Daniel Rice
Senior Vice President and Director of R&D
Technology Solutions Experts, Inc.
Natick MA
------------------------------