Showing 1–2 of 2 results for author: Theunissen, E

Distributed Bayesian: a continuous Distributed Constraint Optimization Problem solver

Authors: Jeroen Fransman, Joris Sijs, Henry Dol, Erik Theunissen, Bart De Schutter

Abstract: In this work, the novel Distributed Bayesian (D-Bay) algorithm is presented for solving multi-agent problems within the continuous Distributed Constraint Optimization Problem (DCOP) framework. This framework extends the classical DCOP framework towards utility functions with continuous domains. Traditional DCOP solvers discretize the continuous domains, which increases the problem size exponential… ▽ More In this work, the novel Distributed Bayesian (D-Bay) algorithm is presented for solving multi-agent problems within the continuous Distributed Constraint Optimization Problem (DCOP) framework. This framework extends the classical DCOP framework towards utility functions with continuous domains. Traditional DCOP solvers discretize the continuous domains, which increases the problem size exponentially. D-Bay overcomes this problem by utilizing Bayesian optimization for the adaptive sampling of variables to avoid discretization entirely. We theoretically show that D-Bay converges to the global optimum of the DCOP for Lipschitz continuous utility functions. The performance of the algorithm is evaluated empirically based on the sample efficiency. The proposed algorithm is compared to a centralized approach with equidistant discretization of the continuous domains for the sensor coordination problem. We find that our algorithm generates better solutions while requiring less samples. △ Less

Submitted 8 February, 2020; originally announced February 2020.

Comments: 31 pages, 7 figures

PrAGMATiC: a Probabilistic and Generative Model of Areas Tiling the Cortex

Authors: Alexander G. Huth, Thomas L. Griffiths, Frederic E. Theunissen, Jack L. Gallant

Abstract: Much of the human cortex seems to be organized into topographic cortical maps. Yet few quantitative methods exist for characterizing these maps. To address this issue we developed a modeling framework that can reveal group-level cortical maps based on neuroimaging data. PrAGMATiC, a probabilistic and generative model of areas tiling the cortex, is a hierarchical Bayesian generative model of cortic… ▽ More Much of the human cortex seems to be organized into topographic cortical maps. Yet few quantitative methods exist for characterizing these maps. To address this issue we developed a modeling framework that can reveal group-level cortical maps based on neuroimaging data. PrAGMATiC, a probabilistic and generative model of areas tiling the cortex, is a hierarchical Bayesian generative model of cortical maps. This model assumes that the cortical map in each individual subject is a sample from a single underlying probability distribution. Learning the parameters of this distribution reveals the properties of a cortical map that are common across a group of subjects while avoiding the potentially lossy step of co-registering each subject into a group anatomical space. In this report we give a mathematical description of PrAGMATiC, describe approximations that make it practical to use, show preliminary results from its application to a real dataset, and describe a number of possible future extensions. △ Less

Submitted 14 April, 2015; originally announced April 2015.