@inproceedings{6476fdc7a43e436982eada5c36049a8e,
title = "Comparing the inductive biases of simple neural networks and Bayesian models",
abstract = "Understanding the relationship between connectionist and probabilistic models is important for evaluating the compatibility of these approaches. We use mathematical analyses and computer simulations to show that a linear neural network can approximate the generalization performance of a probabilistic model of property induction, and that training this network by gradient descent with early stopping results in similar performance to Bayesian inference with a particular prior. However, this prior differs from distributions defined using discrete structure, suggesting that neural networks have inductive biases that can be differentiated from probabilistic models with structured representations.",
keywords = "Bayesian modeling, connectionism, inductive biases, property induction",
author = "Griffiths, {Thomas L.} and Austerweil, {Joseph L.} and Berthiaume, {Vincent G.}",
note = "Publisher Copyright: {\textcopyright} CogSci 2012.All rights reserved.; 34th Annual Meeting of the Cognitive Science Society: Building Bridges Across Cognitive Sciences Around the World, CogSci 2012 ; Conference date: 01-08-2012 Through 04-08-2012",
year = "2012",
language = "English (US)",
series = "Building Bridges Across Cognitive Sciences Around the World - Proceedings of the 34th Annual Meeting of the Cognitive Science Society, CogSci 2012",
publisher = "The Cognitive Science Society",
pages = "402--407",
editor = "Naomi Miyake and David Peebles and Cooper, {Richard P.}",
booktitle = "Building Bridges Across Cognitive Sciences Around the World - Proceedings of the 34th Annual Meeting of the Cognitive Science Society, CogSci 2012",
}