<abstract xml:lang="en" xmlns="http://www.w3.org/1999/xhtml"><p>In the paper we describe basic functions of the Hierarchical Temporal Memory (HTM) network based on a novel biologically inspired model of the large-scale structure of the mammalian neocortex. The focus of this paper is in a systematic exploration of possibilities how to optimize important controlling parameters of the HTM model applied to the classification of hand-written digits from the USPS database. The statistical properties of this database are analyzed using the permutation test which employs a randomization distribution of the training and testing data. Based on a notion of the homogeneous usage of input image pixels, a methodology of the HTM parameter optimization is proposed. In order to study effects of two substantial parameters of the architecture: the <italic>patch size</italic> and the <italic>overlap</italic> in more details, we have restricted ourselves to the single-level HTM networks. A novel method for construction of the training sequences by ordering series of the static images is developed. A novel method for estimation of the parameter <italic>maxDist</italic> based on the box counting method is proposed. The parameter <italic>sigma</italic> of the inference Gaussian is optimized on the basis of the maximization of the belief distribution entropy. Both optimization algorithms can be equally applied to the multi-level HTM networks as well. The influences of the parameters <italic>transitionMemory</italic> and <italic>requestedGroupCount</italic> on the HTM network performance have been explored. Altogether, we have investigated 2736 different HTM network configurations. The obtained classification accuracy results have been benchmarked with the published results of several conventional classifiers.</p></abstract>

In the paper we describe basic functions of the Hierarchical Temporal Memory (HTM) network based on a novel biologically inspired model of the large-scale structure of the mammalian neocortex. The focus of this paper is in a systematic exploration of possibilities how to optimize important controlling parameters of the HTM model applied to the classification of hand-written digits from the USPS database. The statistical properties of this database are analyzed using the permutation test which employs a randomization distribution of the training and testing data. Based on a notion of the homogeneous usage of input image pixels, a methodology of the HTM parameter optimization is proposed. In order to study effects of two substantial parameters of the architecture: the patch size and the overlap in more details, we have restricted ourselves to the single-level HTM networks. A novel method for construction of the training sequences by ordering series of the static images is developed. A novel method for estimation of the parameter maxDist based on the box counting method is proposed. The parameter sigma of the inference Gaussian is optimized on the basis of the maximization of the belief distribution entropy. Both optimization algorithms can be equally applied to the multi-level HTM networks as well. The influences of the parameters transitionMemory and requestedGroupCount on the HTM network performance have been explored. Altogether, we have investigated 2736 different HTM network configurations. The obtained classification accuracy results have been benchmarked with the published results of several conventional classifiers.

On the Optimum Architecture of the Biologically Inspired Hierarchical Temporal Memory Model Applied to the Hand-Written Digit Recognition

Safety & Security Department, Austrian Institute of Technology, Seibersdorf, Austria

Institute of Measurement Science, Slovak Academy of Sciences, Bratislava, Slovakia

Measurement Science Review

In the paper we describe basic functions of the Hierarchical Temporal Memory (HTM) network based on a novel biologically inspired model of the large-scale...

On the Optimum Architecture of the Biologically Inspired Hierarchical Temporal Memory Model Applied to the Hand-Written Digit Recognition

Publicado en línea: 14 may 2010

Páginas: 28 - 49

DOI: https://doi.org/10.2478/v10048-010-0008-4

Palabras clave
hierarchical temporal memory (HTM), optimum network architecture, visual pattern recognition, USPS hand-written digits

This content is open access.

On the Optimum Architecture of the Biologically Inspired Hierarchical Temporal Memory Model Applied to the Hand-Written Digit Recognition

Publicado en línea: 14 may 2010

Páginas: 28 - 49

DOI: https://doi.org/10.2478/v10048-010-0008-4

Palabras clavehierarchical temporal memory (HTM), optimum network architecture, visual pattern recognition, USPS hand-written digits

This content is open access.

Palabras clave
hierarchical temporal memory (HTM), optimum network architecture, visual pattern recognition, USPS hand-written digits