An end-to-end 3D convolutional neural network for decoding attentive mental state

Objective. Steady-state visual evoked potentials(SSVEPs) based braincomputer interface(BCI) has received great interests owing to the high information transfer rate(ITR) and available large number of targets. However, the performance of frequency recognition methods heavily depends on the amount of the calibration data for intra-subject classiﬁcation. Some research adopted the deep learning(DL) algorithm to conduct the inter-subject classiﬁcation, which could reduce the calculation procedure, but the performance still has large room to improve compared with the intra-subject classiﬁcation. Approach. To address these issues, we proposed an eﬃcient SSVEP DL NETwork (termed SSVEPNET) based on 1D convolution and long short-term memory (LSTM) module. To enhance the performance of SSVEPNT, we adopted the spectral normalization and label smoothing technologies during implementing the network architecture. We evaluated the SSVEPNET and compared it with other methods for the intra- and inter-subject classiﬁcation under diﬀerent conditions, i.e., two datasets, two time-window lengths (1 s and 0.5 s), three sizes of training data. Main results. Under all the experimental settings, the proposed SSVEPNET achieved the highest average accuracy for the intra- and inter-subject classiﬁcation on the two SSVEP datasets, when compared with other traditional and DL baseline methods. Signif icance. The extensive experimental results demonstrate that the proposed DL model holds promise to enhance frequency recognition performance in SSVEP-based BCIs. Besides, the mixed network structures with CNN and LSTM, and the spectral normalization and label smoothing could be useful optimization strategies to design eﬃcient models for EEG data.

show abstract

“…• EEGNet: EEGNet is a compact model, and has become a baseline model for EEG data analysis [20,[26][27][28]…”

Section: Dl-based Methodsmentioning

confidence: 99%

An efficient CNN-LSTM network with spectral normalization and label smoothing technologies for SSVEP frequency recognition

2022

Self Cite

View full text Add to dashboard Cite

show abstract

“…Then the data were downsampled to 200 Hz. Following [49], only the first half of each attention trial was utilized to balance the samples between attention and 1 http://doc.ml.tu-berlin.de/simultaneous_EEG_NIRS/ 2 https://figshare.com/articles/dataset/MultichannelEEGrecordingsduringasus tained-attentiondrivingtask/6427334…”

Section: B Preprocessingmentioning

confidence: 99%

LGGNet: Learning From Local-Global-Graph Representations for Brain–Computer Interface

Ding

Robinson

Tong

et al. 2024

IEEE Trans. Neural Netw. Learning Syst.

View full text Add to dashboard Cite

Neuropsychological studies suggest that cooperative activities among different brain functional areas drive high-level cognitive processes. To learn the brain activities within and among different functional areas of the brain, we propose local-global-graph network (LGGNet), a novel neurologically inspired graph neural network (GNN), to learn local-globalgraph (LGG) representations of electroencephalography (EEG) for brain-computer interface (BCI). The input layer of LGGNet comprises a series of temporal convolutions with multiscale 1-D convolutional kernels and kernel-level attentive fusion. It captures temporal dynamics of EEG which then serves as input to the proposed local-and global-graph-filtering layers. Using a defined neurophysiologically meaningful set of local and global graphs, LGGNet models the complex relations within and among functional areas of the brain. Under the robust nested crossvalidation settings, the proposed method is evaluated on three publicly available datasets for four types of cognitive classification tasks, namely the attention, fatigue, emotion, and preference classification tasks. LGGNet is compared with state-of-the-art (SOTA) methods, such as DeepConvNet, EEGNet, R2G-STNN, TSception, regularized graph neural network (RGNN), attentionbased multiscale convolutional neural network-dynamical graph convolutional network (AMCNN-DGCN), hierarchical recurrent neural network (HRNN), and GraphNet. The results show that LGGNet outperforms these methods, and the improvements are statistically significant ( p < 0.05) in most cases. The results show that bringing neuroscience prior knowledge into neural network design yields an improvement of classification performance. The source code can be found at https://github.com/yi-ding-cs/LGG.

show abstract

“…In pHMM, emission probability eS(c) is defined as the probability of output character c from state S defined as p(cjS), and transition probability aS;S0 is the probability of the state changing from S to S0 given as p(S0jS), Here, state transition is based on the prior state. The sequence probability p(x) [20] is given by Markov chain rule as indicated in equation ( 4):…”

Section: Figure-2 Overall Architecture Of Auto Encoder With Hidden Ma...mentioning

confidence: 99%

Cognitive Based Attention Deficit Hyperactivity Disorder Detection with Ability Assessment Using Auto Encoder Based Hidden Markov Model

Mahesh

Goswami

Sriramulu

et al. 2022

Int. j. commun. netw. inf. secur.

View full text Add to dashboard Cite

Attention deficit hyperactivity disorder (ADHD) is a frequent Neuro-generative mental disorder. It can persist in adulthood and be expressed as a cognitive complaint. Behavioural analysis of ADHD consumes more time. This is a multi-informant complex procedure due to the overlaps in symptomatology which is the cause for delay in diagnosis and treatment. Dur to these behavioural consequences and various causes, no single test is utilized till now for diagnosing this disorder. Hence, a diagnosing model of ADHD based on Continuous Ability Assessment Test (CAAT) can enhance and balance behavioural assessment. The objective behind this study is to use a deep learning based model with CAAT for predicting ADHD. The proposed Auto Encoder Based Hidden Markov Model (AE-HMM) produces low-dimensional features of brain structures, and a novel Pearson Correlation Coefficient (PCC) is employed for normalizing these features in order to minimize batch effects over populations and datasets. This goal is consistently achieved and thus the proposed model outperforms few standard approaches which are considered like CogniLearn and 3-D Convolutional Neural Networks (3DCNN). It is found that the proposed AE-HMM method achieves 93.68% of accuracy, 90.66% of sensitivity, 87.72% of specificity, 87.78% of F1-score and 74.22% of kappa score.

show abstract

An end-to-end 3D convolutional neural network for decoding attentive mental state

Cited by 29 publications

References 31 publications

An efficient CNN-LSTM network with spectral normalization and label smoothing technologies for SSVEP frequency recognition

An efficient CNN-LSTM network with spectral normalization and label smoothing technologies for SSVEP frequency recognition

LGGNet: Learning From Local-Global-Graph Representations for Brain–Computer Interface

Cognitive Based Attention Deficit Hyperactivity Disorder Detection with Ability Assessment Using Auto Encoder Based Hidden Markov Model

Contact Info

Product

Resources

About