A Deep Neural Network Architecture Using Dimensionality Reduction with Sparse Matrices

Matsumoto, Wataru; Hagiwara, Manabu; Boufounos, Petros T.; Fukushima, Kunihiko; Mariyama, Toshisada; Zhao, Xiongxin

doi:10.1007/978-3-319-46681-1_48

Cited by 4 publications

(1 citation statement)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One challenge regards how to implement such algorithms in a fast online form. In this respect, we note that the mathematical functions we used here are suitable to be implemented in programmable hardware and can be even approximated by using artificial neural networks (Hinton and Salakhutdinov, 2006 ; Matsumoto et al, 2016 ), suggesting that they could be in principle implementable in adaptive and low-power consumption devices such as neuromorphic circuits (Chicca et al, 2014 ; Qiao et al, 2015 ; Boi et al, 2016 ). Another technological challenge regards how to track down online the temporal dynamics of pre-stimulus activity, which we showed to be useful to increase state-dependent information.…”

Section: Discussionmentioning

confidence: 99%

State-Dependent Decoding Algorithms Improve the Performance of a Bidirectional BMI in Anesthetized Rats

et al. 2017

View full text Add to dashboard Cite

Brain-machine interfaces (BMIs) promise to improve the quality of life of patients suffering from sensory and motor disabilities by creating a direct communication channel between the brain and the external world. Yet, their performance is currently limited by the relatively small amount of information that can be decoded from neural activity recorded form the brain. We have recently proposed that such decoding performance may be improved when using state-dependent decoding algorithms that predict and discount the large component of the trial-to-trial variability of neural activity which is due to the dependence of neural responses on the network's current internal state. Here we tested this idea by using a bidirectional BMI to investigate the gain in performance arising from using a state-dependent decoding algorithm. This BMI, implemented in anesthetized rats, controlled the movement of a dynamical system using neural activity decoded from motor cortex and fed back to the brain the dynamical system's position by electrically microstimulating somatosensory cortex. We found that using state-dependent algorithms that tracked the dynamics of ongoing activity led to an increase in the amount of information extracted form neural activity by 22%, with a consequently increase in all of the indices measuring the BMI's performance in controlling the dynamical system. This suggests that state-dependent decoding algorithms may be used to enhance BMIs at moderate computational cost.

show abstract