Supervised structure learning

This paper concerns structure learning or discovery of discrete generative models. It focuses on Bayesian model selection and the assimilation of training data or content, with a special emphasis on the order in which data are ingested. A key move-in the ensuing schemes-is to place priors on the selection of models, based upon expected free energy. In this setting, expected free energy reduces to a constrained mutual information, where the constraints inherit from priors over outcomes (i.e., preferred outcomes). The resulting scheme is first used to perform image classification on the MNIST dataset to illustrate the basic idea, and then tested on a more challenging problem of discovering models with dynamics, using a simple sprite-based visual disentanglement paradigm and the Tower of Hanoi (cf., blocks world) problem. In these examples, generative models are constructed autodidactically to recover (i.e., disentangle) the factorial structure of latent states-and their characteristic paths or dynamics.

Authors

Institutions

Publication Details

Journal
Biological Psychology
Published
2024-10-19
DOI
https://doi.org/10.1016/j.biopsycho.2024.108891
Citations
18
Primary Topic
Neural Networks and Applications
Type
article
Field-Weighted Citation Impact
6.25

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Supervised structure learning

Victorita Neacsu, Alex Kiefer, Thomas Parr, Christopher L. Buckley et al.
18 citations
Biological Psychology
Neural Networks and Applications
6.25
article

Supervised structure learning

Victorita Neacsu, Alex Kiefer, Thomas Parr, Christopher L. Buckley, Conor Heins, Alexander Tschantz, Tommaso Salvatori, Magnus Koudahl, Tim Verbelen, Noor Sajid, Dimitrije Marković, Karl Friston, Lancelot Da Costa
article en
18 citations

Abstract

This paper concerns structure learning or discovery of discrete generative models. It focuses on Bayesian model selection and the assimilation of training data or content, with a special emphasis on the order in which data are ingested. A key move-in the ensuing schemes-is to place priors on the selection of models, based upon expected free energy. In this setting, expected free energy reduces to a constrained mutual information, where the constraints inherit from priors over outcomes (i.e., preferred outcomes). The resulting scheme is first used to perform image classification on the MNIST dataset to illustrate the basic idea, and then tested on a more challenging problem of discovering models with dynamics, using a simple sprite-based visual disentanglement paradigm and the Tower of Hanoi (cf., blocks world) problem. In these examples, generative models are constructed autodidactically to recover (i.e., disentangle) the factorial structure of latent states-and their characteristic paths or dynamics.

Biological PsychologyVol. 193
University of Sussex (GB), Wellcome Centre for Human Neuroimaging (GB), University of Oxford (GB), VERSES (United States), University College London (GB), Technische Universität Dresden (DE)
Wellcome Trust
Affordable and clean energy
Openalex Percentile: Top 3%
Neural Networks and Applications
6.25
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.