Search

Neural Architecture Search (NAS) methods are widely employed to address the time-consuming and costly challenges associated with manual operation and design of deep convolutional neural networks (DCNNs). Nonetheless, prevailing methods still encounter several pressing obstacles, including limited network architecture design, excessively lengthy search periods, and insufficient utilization of the search space. In light of these concerns, this study proposes an optimization strategy for residual networks that leverages an enhanced Particle swarm optimization algorithm. Primarily, low-complexity residual architecture block is employed as the foundational unit for architecture exploration, facilitating a more diverse investigation into network architectures while minimizing parameters. Additionally, we employ a depth initialization strategy to confine the search space within a reasonable range, thereby mitigating unnecessary particle exploration. Lastly, we present a novel approach for computing particle differences and updating velocity mechanisms to enhance the exploration of updated trajectories. This method significantly contributes to the improved utilization of the search space and the augmentation of particle diversity. Moreover, we constructed a crime-dataset comprising 13 classes to assess the effectiveness of the proposed algorithm. Experimental results demonstrate that our algorithm can design lightweight networks with superior classification performance on both benchmark datasets and the crime-dataset., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024. Published by Elsevier Ltd.)

Decoding brain activity from non-invasive electroencephalography (EEG) is crucial for brain-computer interfaces (BCIs) and the study of brain disorders. Notably, end-to-end EEG decoding has gained widespread popularity in recent years owing to the remarkable advances in deep learning research. However, many EEG studies suffer from limited sample sizes, making it difficult for existing deep learning models to effectively generalize to highly noisy EEG data. To address this fundamental limitation, this paper proposes a novel end-to-end EEG decoding algorithm that utilizes a low-rank weight matrix to encode both spatio-temporal filters and the classifier, all optimized under a principled sparse Bayesian learning (SBL) framework. Importantly, this SBL framework also enables us to learn hyperparameters that optimally penalize the model in a Bayesian fashion. The proposed decoding algorithm is systematically benchmarked on five motor imagery BCI EEG datasets ( N=192) and an emotion recognition EEG dataset ( N=45), in comparison with several contemporary algorithms, including end-to-end deep-learning-based EEG decoding algorithms. The classification results demonstrate that our algorithm significantly outperforms the competing algorithms while yielding neurophysiologically meaningful spatio-temporal patterns. Our algorithm therefore advances the state-of-the-art by providing a novel EEG-tailored machine learning tool for decoding brain activity.

Detecting event-related potential (ERP) is a challenging problem because of its low signal-to-noise ratio and complex spatial-temporal features. Conventional detection methods usually rely on the ensemble averaging technique, which may eliminate subtle but important information in ERP signals and lead to poor detection performance. Inspired by the good performance of discriminative restricted Boltzmann machine (DRBM) in feature extraction and classification, we propose a spatial-temporal DRBM (ST-DRBM) to extract spatial and temporal features for ERP detection. The experimental results and statistical analyses demonstrate that the proposed method is able to achieve state-of-the-art ERP detection performance. The ST-DRBM is not only an effective ERP detector, but also a practical tool for ERP analysis. Based on the proposed model, similar scalp distribution and temporal variations were found in the ERP signals of different sessions, which indicated the feasibility of cross-session ERP detection. Given its state-of-the-art performance and effective analytical technique, ST-DRBM is promising for ERP-based brain-computer interfaces and neuroscience research.

In off-line training of motor imagery-based brain-computer interfaces (BCIs), to enhance the generalization performance of the learned classifier, the local information contained in test data could be used to improve the performance of motor imagery as well. Further considering that the covariance matrices of electroencephalogram (EEG) signal lie on Riemannian manifold, in this paper, we construct a Riemannian graph to incorporate the information of training and test data into processing. The adjacency and weight in Riemannian graph are determined by the geodesic distance of Riemannian manifold. Then, a new graph embedding algorithm, called bilinear regularized locality preserving (BRLP), is derived upon the Riemannian graph for addressing the problems of high dimensionality frequently arising in BCIs. With a proposed regularization term encoding prior information of EEG channels, the BRLP could obtain more robust performance. Finally, an efficient classification algorithm based on extreme learning machine is proposed to perform on the tangent space of learned embedding. Experimental evaluations on the BCI competition and in-house data sets reveal that the proposed algorithms could obtain significantly higher performance than many competition algorithms after using same filter process.

Learning optimal spatio-temporal filters is a key to feature extraction for single-trial electroencephalogram (EEG) classification. The challenges are controlling the complexity of the learning algorithm so as to alleviate the curse of dimensionality and attaining computational efficiency to facilitate online applications, e.g., brain-computer interfaces (BCIs). To tackle these barriers, this paper presents a novel algorithm, termed regularized spatio-temporal filtering and classification (RSTFC), for single-trial EEG classification. RSTFC consists of two modules. In the feature extraction module, an l2 -regularized algorithm is developed for supervised spatio-temporal filtering of the EEG signals. Unlike the existing supervised spatio-temporal filter optimization algorithms, the developed algorithm can simultaneously optimize spatial and high-order temporal filters in an eigenvalue decomposition framework and thus be implemented highly efficiently. In the classification module, a convex optimization algorithm for sparse Fisher linear discriminant analysis is proposed for simultaneous feature selection and classification of the typically high-dimensional spatio-temporally filtered signals. The effectiveness of RSTFC is demonstrated by comparing it with several state-of-the-arts methods on three brain-computer interface (BCI) competition data sets collected from 17 subjects. Results indicate that RSTFC yields significantly higher classification accuracies than the competing methods. This paper also discusses the advantage of optimizing channel-specific temporal filters over optimizing a temporal filter common to all channels.

For M/EEG-based distributed source imaging, it has been established that the L2-norm-based methods are effective in imaging spatially extended sources, whereas the L1-norm-based methods are more suited for estimating focal and sparse sources. However, when the spatial extents of the sources are unknown a priori, the rationale for using either type of methods is not adequately supported. Bayesian inference by exploiting the spatio-temporal information of the patch sources holds great promise as a tool for adaptive source imaging, but both computational and methodological limitations remain to be overcome. In this paper, based on state-space modeling of the M/EEG data, we propose a fully data-driven and scalable algorithm, termed STRAPS, for M/EEG patch source imaging on high-resolution cortices. Unlike the existing algorithms, the recursive penalized least squares (RPLS) procedure is employed to efficiently estimate the source activities as opposed to the computationally demanding Kalman filtering/smoothing. Furthermore, the coefficients of the multivariate autoregressive (MVAR) model characterizing the spatial-temporal dynamics of the patch sources are estimated in a principled manner via empirical Bayes. Extensive numerical experiments demonstrate STRAPS's excellent performance in the estimation of locations, spatial extents and amplitudes of the patch sources with varying spatial extents.

Considering the two-class classification problem in brain imaging data analysis, we propose a sparse representation-based multi-variate pattern analysis (MVPA) algorithm to localize brain activation patterns corresponding to different stimulus classes/brain states respectively. Feature selection can be modeled as a sparse representation (or sparse regression) problem. Such technique has been successfully applied to voxel selection in fMRI data analysis. However, single selection based on sparse representation or other methods is prone to obtain a subset of the most informative features rather than all. Herein, our proposed algorithm recursively eliminates informative features selected by a sparse regression method until the decoding accuracy based on the remaining features drops to a threshold close to chance level. In this way, the resultant feature set including all the identified features is expected to involve all the informative features for discrimination. According to the signs of the sparse regression weights, these selected features are separated into two sets corresponding to two stimulus classes/brain states. Next, in order to remove irrelevant/noisy features in the two selected feature sets, we perform a nonparametric permutation test at the individual subject level or the group level. In data analysis, we verified our algorithm with a toy data set and an intrinsic signal optical imaging data set. The results show that our algorithm has accurately localized two class-related patterns. As an application example, we used our algorithm on a functional magnetic resonance imaging (fMRI) data set. Two sets of informative voxels, corresponding to two semantic categories (i.e., "old people" and "young people"), respectively, are obtained in the human brain.

In sparse representation, two important sparse solutions, the 0-norm and 1-norm solutions, have been receiving much of attention. The 0-norm solution is the sparsest, however it is not easy to obtain. Although the 1-norm solution may not be the sparsest, it can be easily obtained by the linear programming method. In many cases, the 0-norm solution can be obtained through finding the 1-norm solution. Many discussions exist on the equivalence of the two sparse solutions. This paper analyzes two conditions for the equivalence of the two sparse solutions. The first condition is necessary and sufficient, however, difficult to verify. Although the second is necessary but is not sufficient, it is easy to verify. In this paper, we analyze the second condition within the stochastic framework and propose a variant. We then prove that the equivalence of the two sparse solutions holds with high probability under the variant of the second condition. Furthermore, in the limit case where the 0-norm solution is extremely sparse, the second condition is also a sufficient condition with probability 1.

This paper discusses the estimation and numerical calculation of the probability that the 0-norm and 1-norm solutions of underdetermined linear equations are equivalent in the case of sparse representation. First, we define the sparsity degree of a signal. Two equivalence probability estimates are obtained when the entries of the 0-norm solution have different sparsity degrees. One is for the case in which the basis matrix is given or estimated, and the other is for the case in which the basis matrix is random. However, the computational burden to calculate these probabilities increases exponentially as the number of columns of the basis matrix increases. This computational complexity problem can be avoided through a sampling method. Next, we analyze the sparsity degree of mixtures and establish the relationship between the equivalence probability and the sparsity degree of the mixtures. This relationship can be used to analyze the performance of blind source separation (BSS). Furthermore, we extend the equivalence probability estimates to the small noise case. Finally, we illustrate how to use these theoretical results to guarantee a satisfactory performance in underdetermined BSS.

For many electroencephalogram (EEG)-based brain-computer interfaces (BCIs), a tedious and time-consuming training process is needed to set parameters. In BCI Competition 2005, reducing the training process was explicitly proposed as a task. Furthermore, an effective BCI system needs to be adaptive to dynamic variations of brain signals; that is, its parameters need to be adjusted online. In this article, we introduce an extended expectation maximization (EM) algorithm, where the extraction and classification of common spatial pattern (CSP) features are performed jointly and iteratively. In each iteration, the training data set is updated using all or part of the test data and the labels predicted in the previous iteration. Based on the updated training data set, the CSP features are reextracted and classified using a standard EM algorithm. Since the training data set is updated frequently, the initial training data set can be small (semi-supervised case) or null (unsupervised case). During the above iterations, the parameters of the Bayes classifier and the CSP transformation matrix are also updated concurrently. In online situations, we can still run the training process to adjust the system parameters using unlabeled data while a subject is using the BCI system. The effectiveness of the algorithm depends on the robustness of CSP feature to noise and iteration convergence, which are discussed in this article. Our proposed approach has been applied to data set IVa of BCI Competition 2005. The data analysis results show that we can obtain satisfying prediction accuracy using our algorithm in the semisupervised and unsupervised cases. The convergence of the algorithm and robustness of CSP feature are also demonstrated in our data analysis.

In this paper, we use a two-stage sparse factorization approach for blindly estimating the channel parameters and then estimating source components for electroencephalogram (EEG) signals. EEG signals are assumed to be linear mixtures of source components, artifacts, etc. Therefore, a raw EEG data matrix can be factored into the product of two matrices, one of which represents the mixing matrix and the other the source component matrix. Furthermore, the components are sparse in the time-frequency domain, i.e., the factorization is a sparse factorization in the time frequency domain. It is a challenging task to estimate the mixing matrix. Our extensive analysis and computational results, which were based on many sets of EEG data, not only provide firm evidences supporting the above assumption, but also prompt us to propose a new algorithm for estimating the mixing matrix. After the mixing matrix is estimated, the source components are estimated in the time frequency domain using a linear programming method. In an example of the potential applications of our approach, we analyzed the EEG data that was obtained from a modified Sternberg memory experiment. Two almost uncorrelated components obtained by applying the sparse factorization method were selected for phase synchronization analysis. Several interesting findings were obtained, especially that memory-related synchronization and desynchronization appear in the alpha band, and that the strength of alpha band synchronization is related to memory performance.

Brain-computer interfaces (BCIs) aim to provide an alternative channel for paralyzed patients to communicate with external world. Reducing the time needed for the initial calibration is one important objective in P300 based BCI research. In this paper, the training time is reduced by a semi-supervised learning approach. A model is trained by small training set first. The on-line test data with predicted labels are then added to the initial training data to extend the training data. And the model is updated online using the extended training set. The method is tested by a data set of P300 based word speller. The experimental results show that 93.4% of the training time for this data set can be reduced by the proposed method while keeping satisfactory accuracy rate. This semi-supervised learning approach is applied on-line to obtain robust and adaptive model for P300 based speller with small training set, which is believed to be very essential to improve the feasibility of the P300 based BCI.

In machine learning based Brain Computer Interfaces (BCIs), it is a challenge to use only a small amount of labelled data to build a classifier for a specific subject. This challenge was specifically addressed in BCI Competition 2005. Moreover, an effective BCI system should be adaptive to tackle the dynamic variations in brain signal. One of the solutions is to have its parameters adjustable while the system is used online. In this paper we introduce a new semi-supervised support vector machine (SVM) learning algorithm. In this method, the feature extraction and classification are jointly performed in iterations. This method allows us to use a small training set to train the classifier while maintaining high performance. Therefore, the tedious initial calibration process is shortened. This algorithm can be used online to make the BCI system robust to possible signal changes. We analyze two important issues of the proposed algorithm, the robustness of the features to noise and the convergence of algorithm. We applied our method to data from BCI competition 2005, and the results demonstrated the validity of the proposed algorithm.

In recent years, deep neural networks have achieved great success in many fields, such as computer vision and natural language processing. Traditional image recommendation algorithms use text-based recommendation methods. The process of displaying images requires a lot of time and labor, and the time-consuming labor is inefficient. Therefore, this article mainly studies image recommendation algorithms based on deep neural networks in social networks. First, according to the time stamp information of the dataset, the interaction records of each user are sorted by the closest time. Then, some feature vectors are created via traditional feature algorithms like LBP, BGC3, RTU, or CNN extraction. For image recommendation, two LSTM neural networks are established, which accept these feature vectors as input, respectively. The compressed output of the two sub-ESTM neural networks is used as the input of another LSTM neural network. The multilayer regression algorithm is adopted to randomly sample some network nodes to obtain the cognitive information of the nodes sampled in the entire network, predict the relationship between all nodes in the network based on the cognitive information, and perform low sampling to achieve relationship prediction. The experiments show that proposed LSTM model together with CNN feature vectors can outperform other algorithms. [ABSTRACT FROM AUTHOR]

Low energy consumption is crucial for body area networks (BANs). In BAN-enabled ECG monitoring, the continuous monitoring entails the need of the sensor nodes to transmit a huge data to the sink node, which leads to excessive energy consumption. To reduce airtime over energy-hungry wireless links, this paper presents an energy-efficient compressed sensing (CS)-based approach for on-node ECG compression. At first, an algorithm called minimal mutual coherence pursuit is proposed to construct sparse binary measurement matrices, which can be used to encode the ECG signals with superior performance and extremely low complexity. Second, in order to minimize the data rate required for faithful reconstruction, a weighted \ell_1 minimization model is derived by exploring the multisource prior knowledge in wavelet domain. Experimental results on MIT-BIH arrhythmia database reveals that the proposed approach can obtain higher compression ratio than the state-of-the-art CS-based methods. Together with its low encoding complexity, our approach can achieve significant energy saving in both encoding process and wireless transmission. [ABSTRACT FROM PUBLISHER]

The recently proposed modified-compressive sensing (modified-CS), which utilizes the partially known support as prior knowledge, significantly improves the performance of recovering sparse signals. However, modified-CS depends heavily on the reliability of the known support. An important problem, which must be studied further, is the recoverability of modified-CS when the known support contains a number of errors. In this letter, we analyze the recoverability of modified-CS in a stochastic framework. A sufficient and necessary condition is established for exact recovery of a sparse signal. Utilizing this condition, the recovery probability that reflects the recoverability of modified-CS can be computed explicitly for a sparse signal with nonzero entries. Simulation experiments have been carried out to validate our theoretical results. [ABSTRACT FROM AUTHOR]

Abstract: This paper discusses blind source extraction in various ill-conditioned cases based on a simple extraction network model. Extractability is first analyzed for the following ill-conditioned cases: the mixing matrix is square but singular, the number of sensors is smaller than that of sources, the number of sensors is larger than that of sources but the column rank of mixing matrix is deficient, and the number of sources is unknown and the column rank of mixing matrix is deficient. A necessary and sufficient condition for extractability is obtained. A cost function and an unsupervised learning algorithm for the extraction network model are developed. Simulation results are also presented to show the validity of the theoretical results and the performance and characteristics of the learning algorithm. [Copyright &y& Elsevier]

This paper proposes a novel grouping decision approach for blind source estimation of FIR (finite impulse response) channels with binary sources. First, solvability is discussed for single-input systems and multi-input systems. Necessary and sufficient conditions for recoverability are derived. For single-input systems, a new deterministic algorithm based on grouping and decision is proposed to recover the source up to a delay. The algorithm is easy to implement and has several advantages. For instance, when the solvability conditions are satisfied, it can be applied to cases in which: (i) the channel has zeros on the unit circle or outside of the unit circle; (ii) there are fewer sensors than sources; (iii) the source is temporarily dependent. To improve noise tolerance and reduce computational cost, the algorithm is further elaborated for highly noisy channels and high-order FIR channels, respectively. For the channels with high unimodal noise, fewer peaks appear in the probability density function (pdf) of the outputs compared to the pdf of the outputs of channels with a higher SNR. After the peaks representing cluster centers are estimated using a maximum likelihood (ML) approach, the deterministic algorithm can be used. Similar to highly noisy channels, the algorithm is also effective for high-order, exponentially decaying channels after fewer cluster centers are estimated. Furthermore, blind source estimation for multi-input systems also can be carried out as with the case of single input systems. Two deflation algorithms are presented for temporarily dependent sources and i.i.d. sources. Based on the source estimation and deflation algorithms, the sources can be obtained one by one. Finally, the validity and performance of the algorithms are illustrated by several simulation examples. [Copyright &y& Elsevier]

Presents information on a study which introduced a technique for sequential blind extraction of singularly mixed source signals. Description of a neural-network model and an adaptive algorithm for single-source blind extraction; Stability of the algorithm ; Simulation results; Conclusions.

Abstract: We present a new linear discriminant analysis method based on information theory, where the mutual information between linearly transformed input data and the class labels is maximized. First, we introduce a kernel-based estimate of mutual information with a variable kernel size. Furthermore, we devise a learning algorithm that maximizes the mutual information w.r.t. the linear transformation. Two experiments are conducted: the first one uses a toy problem to visualize and compare the transformation vectors in the original input space; the second one evaluates the performance of the method for classification by employing cross-validation tests on four datasets from the UCI repository. Various classifiers are investigated. Our results show that this method can significantly boost class separability over conventional methods, especially for nonlinear classification. [Copyright &y& Elsevier]

Electromagnetic (E/MEG) brain source imaging involves challenging problems that make it particularly difficult to estimate both the locations and extents of extended sources. In this study, we propose a new method called V ariation S parse S ource I maging based on C onditional M ean of the posterior (VSSI-CM), which is built upon a Bayesian framework, to reconstruct extended E/MEG generators. Based on the proposed framework, VSSI-CM can employ various spatial priors (e.g., the Laplace prior) to explore sparseness of current sources in transform domains (e.g., the variation transform in this study). Considering the complexity of posterior density in the estimated sources, we propose using the posterior mean instead of the typical maximum a posterior (MAP) estimate as a more accurate inverse solution. The posterior mean is obtained by fitting an approximated Gaussian distribution to the intractable true posterior distribution. An efficient double-loop algorithm is also proposed using convex analysis skills. Validation using synthetic and human experimental data sets indicates that VSSI-CM outperforms the well-studied L 2 -norm methods (i.e., sLORETA and dSPM) and the sparse constrained methods that explore sparseness in the original source domain. The estimates from VSSI-CM are also more accurate than that from MAP. [ABSTRACT FROM AUTHOR]

Semi-supervised learning has been applied in brain–computer interfaces (BCIs) to reduce calibration time for user. For example, a sequential updated self-training least squares support vector machine (SUST-LSSVM) was devised for online semi-supervised P300 speller. Despite its good performance, the computational complexity becomes too high after several updates, which hinders its practical online application. In this paper, we present a self-training regularized weighted online sequential extreme learning machine (ST-RWOS-ELM) for P300 speller. It achieves much lower complexity compared to SUST-LSSVM without affecting the spelling accuracy performance. The experimental results validate its effectiveness in the P300 system. [ABSTRACT FROM AUTHOR]

Abstract: Extraction and separation of functionally different event-related potentials (ERPs) from electroencephalography (EEG) is a long-standing problem in cognitive neuroscience. In this paper, we propose a Bayesian spatio-temporal model for estimating ERP components from multichannel EEG recorded under multiple experimental conditions. The model isolates the spatially and temporally overlapping ERP components by utilizing their phase-locking structure and the inter-condition non-stationarity structure of their amplitudes and latencies. Critically, unlike in previous multilinear algorithms, the non-phase-locked background EEGs are modeled as spatially correlated and non-isotropic signals. A variational algorithm was developed for approximate Bayesian inference of the proposed model, with the effective number of ERP components automatically determined as a part of the algorithm. The utility of the algorithm is demonstrated with applications to synthetic data and the EEG data collected from 13 subjects during a face inversion experiment. The results show that our algorithm more accurately and reliably estimates the spatio-temporal patterns, amplitudes, and latencies of the underlying ERP components in comparison with several state-of-the-art algorithms. [Copyright &y& Elsevier]