Subject-Independent Brain–Computer Interfaces Based on Deep Convolutional Neural Networks

Abstract

For a brain-computer interface (BCI) system, a calibration procedure is required for each individual user before he/she can use the BCI. This procedure requires approximately 20-30 min to collect enough data to build a reliable decoder. It is, therefore, an interesting topic to build a calibration-free, or subject-independent, BCI. In this article, we construct a large motor imagery (MI)-based electroencephalography (EEG) database and propose a subject-independent framework based on deep convolutional neural networks (CNNs). The database is composed of 54 subjects performing the left-and right-hand MI on two different days, resulting in 21 600 trials for the MI task. In our framework, we formulated the discriminative feature representation as a combination of the spectral-spatial input embedding the diversity of the EEG signals, as well as a feature representation learned from the CNN through a fusion technique that integrates a variety of discriminative brain signal patterns. To generate spectral-spatial inputs, we first consider the discriminative frequency bands in an information-theoretic observation model that measures the power of the features in two classes. From discriminative frequency bands, spectral-spatial inputs that include the unique characteristics of brain signal patterns are generated and then transformed into a covariance matrix as the input to the CNN. In the process of feature representations, spectral-spatial inputs are individually trained through the CNN and then combined by a concatenation fusion technique. In this article, we demonstrate that the classification accuracy of our subject-independent (or calibration-free) model outperforms that of subject-dependent models using various methods [common spatial pattern (CSP), common spatiospectral pattern (CSSP), filter bank CSP (FBCSP), and Bayesian spatiospectral filter optimization (BSSFO)].

A Simplified CNN Classification Method for MI-EEG via the Electrode Pairs Signals

“…Motor imagery electroencephalography (MI-EEG) is a self-regulated EEG without an external stimulus, which can be detected by electrodes. It was suggested in a literature survey that MI is consistent with changes caused by actual exercise in the motor cortex region (Jenson et al, 2019;Kwon et al, 2019).…”

Deep Learning Based Inter-subject Continuous Decoding of Motor Imagery for Practical Brain-Computer Interfaces

“…The training time for Architecture-1 (intra-subject) is 794 s which is less than (Tabar and Halici, 2016) where the training time is 1,157 s. The training time for Architecture-2 (inter-subject) is 1934 s which is also less than other inter-subject architecture such as Kwon et al (2019) where the training time is 12 min. The single-trial decoding time in Tabar and Halici (2016) was 400 ms and in Kwon et al (2019) it was 150 ms, whereas in the current study the single-trial decoding time is 102.52 ms which is much smaller than others. Thus, it shows that the computational complexity of the proposed CNN architectures is less or comparable to other competitive architectures given in previous studies.…”

Prior context influences motor brain areas in an auditory oddball task and prefrontal cortex multitasking modelling

“…Finally, bearing in mind discussion of multitask experiment [3] discussed in use of person identi cation with reliable decoders [2] and re-identi cation using different visual views [1] in systems with different interfaces. These interfaces may involve not only EEG but also precise electrodes position inferred or combined with fMRI or fNIRS as occipital images, as the present work suggest.…”

“…iRecent publications in stimulus-driven works on neural network and learning systems are awakening the interest of multimodal attention systems, such as the visual works in Brain Computer Interface (BCI) discussed to auditory modality [1]- [2], bearing in mind several task conditions [3]. In the present work, the interaction of the auditory and motor systems is studied and modeled when a visual stimulus is xedly observed in an auditory-motor task.…”