Tensor Convolutional Dictionary Learning With CP Low-Rank Activations

Neural decoding is still a challenging and a hot 1 topic in neurocomputing science. Recently, many studies have 2 shown that brain network patterns containing rich spatiotem-3 poral structural information represent the brain's activation 4 information under external stimuli. In the traditional method, 5 brain network features are directly obtained using the standard 6 machine learning method and provide to a classifier, subsequently 7 decoding external stimuli. However, this method cannot effec-8 tively extract the multidimensional structural information hidden 9 in the brain network. Furthermore, studies on tensors have 10 show that the tensor decomposition model can fully mine unique 11 spatiotemporal structural characteristics of a spatiotemporal 12 structure in data with a multidimensional structure. This re-13 search proposed a stimulus-constrained Tensor Brain Network (s-14 TBN) model that involves the tensor decomposition and stimulus 15 category-constraint information. The model was verified on real 16 neuroimaging data obtained via magnetoencephalograph and 17 functional mangetic resonance imaging). Experimental results 18 show that the s-TBN model achieve accuracy matrices of greater 19 than 11.06% and 18.46% on the accuracy matrix compared with 20 other methods on two modal datasets. These results prove the 21 superiority of extracting discriminative characteristics using the 22 STN model, especially for decoding object stimuli with semantic 23 information. 24 Index Terms-Neural decoding, Brain network, Tensor decom-25 position, Tensor brain network 26 I. INTRODUCTION 27 N EURAL decoding is achieved by analyzing the neural 28 signals pattern collected using the noninvasive device 29 to evaluate the brain's activation patterns in response to 30 specific visual stimuli. Subsequently, these patterns are used to 31 deduce the external stimulus categories inversely [1]. One vital 32 step in this process is establishing the mapping relationships 33 among the neural activation patterns, neural signal patterns, 34 and external stimuli. Previous research focused on the brain 35 individual-area model to establish this mapping relationship 36 (such as a single brain area or channel signals) [2]. However, 37 the study of this mapping from the network model on the level 38 of the entire brain or the local system has become a research 39 hotspot [3]. 40 Brain network structure data typically involve various di-41 mensions, such as time and space, and indirectly represent 42

show abstract

Tensor Convolutional Dictionary Learning With CP Low-Rank Activations

Cited by 2 publications

References 55 publications

Multi-Dimensional Signal Recovery Using Low-Rank Deconvolution

Multi-Dimensional Signal Recovery Using Low-Rank Deconvolution

s-TBN: A New Neural Decoding Model to Identify Stimulus Categories From Brain Activity Patterns

Contact Info

Product

Resources

About