Relational Neural Expectation Maximization
ثبت نشده
چکیده
Common-sense physical reasoning is an essential ingredient for any intelligent agent operating in the real-world. For example, it can be used to simulate the environment, or to infer the state of parts of the world that are currently unobserved. In order to match real-world conditions this causal knowledge must be learned without access to supervised data. To solve this problem, we present a novel method that incorporates prior knowledge about the compositional nature of human perception to factor interactions between object-pairs and to learn them efficiently. It learns to discover objects and to model physical interactions between them from raw visual images in a purely unsupervised fashion. On videos of bouncing balls we show the superior modelling capabilities of our method compared to other unsupervised neural approaches, that do not incorporate such prior knowledge. We show its ability to handle occlusion and that it can extrapolate learned knowledge to environments with different numbers of objects.
منابع مشابه
Efficient query-driven biclustering of gene expression data using Probabilistic Relational Models
Biclustering is an increasingly popular technique to identify gene regulatory modules that are linked to biological processes. We describe a novel method, called ProBic, that was developed within the framework of Probabilistic Relational Models (PRMs). ProBic is an efficient biclustering algorithm that simultaneously identifies a set of potentially overlapping biclusters in a gene expression da...
متن کاملPiecewise Surface Regression Modeling in Intelligent Decision Guidance System
An intelligent decision guidance system which is composed of data collection, learning, optimization, and prediction is proposed in the paper. Built on the traditional relational database management system, the regression learning ability is incorporated. The Expectation Maximization Multi-Step Piecewise Surface Regression Learning (EMMPSR) algorithm is proposed to solve piecewise surface regre...
متن کاملComparison of the Expectation Maximization Algorithm and an Artificial Neural Network for Automated Bone Segmentation
Comparison of the Expectation Maximization Algorithm and an Artificial Neural Network for Automated Bone Segmentation Austin Ramme4, Vincent Magnotta1,2, Nicole M. Grosland2,3 1Radiology, University of Iowa, Iowa City, IA; 2Biomedical Engineering, University of Iowa, Iowa City, IA; 3Center for Computer Aided Design, The University of Iowa, Iowa City, IA; 4Carver College of Medicine, The Univers...
متن کاملUsing Expectation-Maximization for Reinforcement Learning
We discuss Hinton’s (1989) relative payoff procedure (RPP), a static reinforcement learning algorithm whose foundation is not stochastic gradient ascent. We show circumstances under which applying the RPP is guaranteed to increase the mean return, even though it can make large changes in the values of the parameters. The proof is based on a mapping between the RPP and a form of the expectation-...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2017