Deep Learning for Epileptic Seizure Detection Using a Causal-Spatio-Temporal Model Based on Transfer Entropy.

Drug-resistant epilepsy is frequent, persistent, and brings a heavy economic burden to patients and their families. Traditional epilepsy detection methods ignore the causal relationship of seizures and focus on a single time or spatial dimension, and the effect varies greatly in different patients. Therefore, it is necessary to research accurate automatic detection technology of epilepsy in different patients. We propose a causal-spatio-temporal graph attention network (CSTGAT), which uses transfer entropy (TE) to construct a causal graph between multiple channels, combining graph attention network (GAT) and bi-directional long short-term memory (BiLSTM) to capture temporal dynamic correlation and spatial topological structure information. The accuracy, specificity, and sensitivity of the SWEZ dataset were 97.24%, 97.92%, and 98.11%. The accuracy of the private dataset reached 98.55%. The effectiveness of each module was proven through ablation experiments and the impact of different network construction methods was compared. The experimental results indicate that the causal relationship network constructed by TE could accurately capture the information flow of epileptic seizures, and GAT and BiLSTM could capture spatiotemporal dynamic correlations. This model accurately captures causal relationships and spatiotemporal correlations on two datasets, and it overcomes the variability of epileptic seizures in different patients, which may contribute to clinical surgical planning.

MFAM-AD: an anomaly detection model for multivariate time series using attention mechanism to fuse multi-scale features.

Multivariate time series anomaly detection has garnered significant attention in fields such as IT operations, finance, medicine, and industry. However, a key challenge lies in the fact that anomaly patterns often exhibit multi-scale temporal variations, which existing detection models often fail to capture effectively. This limitation significantly impacts detection accuracy. To address this issue, we propose the MFAM-AD model, which combines the strengths of convolutional neural networks (CNNs) and bi-directional long short-term memory (Bi-LSTM). The MFAM-AD model is designed to enhance anomaly detection accuracy by seamlessly integrating temporal dependencies and multi-scale spatial features. Specifically, it utilizes parallel convolutional layers to extract features across different scales, employing an attention mechanism for optimal feature fusion. Additionally, Bi-LSTM is leveraged to capture time-dependent information, reconstruct the time series and enable accurate anomaly detection based on reconstruction errors. In contrast to existing algorithms that struggle with inadequate feature fusion or are confined to single-scale feature analysis, MFAM-AD effectively addresses the unique challenges of multivariate time series anomaly detection. Experimental results on five publicly available datasets demonstrate the superiority of the proposed model. Specifically, on the datasets SMAP, MSL, and SMD1-1, our MFAM-AD model has the second-highest F1 score after the current state-of-the-art DCdetector model. On the datasets NIPS-TS-SWAN and NIPS-TS-GECCO, the F1 scores of MAFM-AD are 0.046 (6.2%) and 0.09 (21.3%) higher than those of DCdetector, respectively(the value ranges from 0 to 1). These findings validate the MFAMAD model's efficacy in multivariate time series anomaly detection, highlighting its potential in various real-world applications.

Air quality forecasting using a spatiotemporal hybrid deep learning model based on VMD-GAT-BiLSTM.

The precise forecasting of air quality is of great significance as an integral component of early warning systems. This remains a formidable challenge owing to the limited information of emission source and the considerable uncertainties inherent in dynamic processes. To improve the accuracy of air quality forecasting, this work proposes a new spatiotemporal hybrid deep learning model based on variational mode decomposition (VMD), graph attention networks (GAT) and bi-directional long short-term memory (BiLSTM), referred to as VMD-GAT-BiLSTM, for air quality forecasting. The proposed model initially employ a VMD to decompose original PM2.5 data into a series of relatively stable sub-sequences, thus reducing the influence of unknown factors on model prediction capabilities. For each sub-sequence, a GAT is then designed to explore deep spatial relationships among different monitoring stations. Next, a BiLSTM is utilized to learn the temporal features of each decomposed sub-sequence. Finally, the forecasting results of each decomposed sub-sequence are aggregated and summed as the final air quality prediction results. Experiment results on the collected Beijing air quality dataset show that the proposed model presents superior performance to other used methods on both short-term and long-term air quality forecasting tasks.

A novel optimization-assisted multi-scale and dilated adaptive hybrid deep learning network with feature fusion for event detection from social media.

Social media networks become an active communication medium for connecting people and delivering new messages. Social media can perform as the primary channel, where the globalized events or instances can be explored. Earlier models are facing the pitfall of noticing the temporal and spatial resolution for enhancing the efficacy. Therefore, in this proposed model, a new event detection approach from social media data is presented. Firstly, the essential data is collected and undergone for pre-processing stage. Further, the Bidirectional Encoder Representations from Transformers (BERT) and Term Frequency Inverse Document Frequency (TF-IDF) are employed for extracting features. Subsequently, the two resultant features are given to the multi-scale and dilated layer present in the detection network of GRU and Res-Bi-LSTM, named as Multi-scale and Dilated Adaptive Hybrid Deep Learning (MDA-HDL) for event detection. Moreover, the MDA-HDL network's parameters are tuned by Improved Gannet Optimization Algorithm (IGOA) to enhance the performance. Finally, the execution of the system is done over the Python platform, where the system is validated and compared with baseline methodologies. The accuracy findings of model acquire as 94.96 for dataset 1 and 96.42 for dataset 2. Hence, the recommended model outperforms with the superior results while detecting the social events.

Short-term wind speed prediction of wind farm based on TSO-VMD-BiLSTM.

Aiming at the random and intermittent characteristics of wind speed, a short-term wind speed prediction (SWSP) method based on TSO-VMD-BiLSTM is proposed in this article. Firstly, open-source historical data from a certain region in 2022, including wind speed, direction, pressure, and temperature is analyzed. The data is processed through variational mode decomposition (VMD) to fully extract feature data from historical wind speed records. Secondly, taking historical wind speed, direction, pressure, and temperature as inputs and wind speed as output, a SWSP model based on long short-term memory (LSTM) network is constructed. Thirdly, the tuna swarm optimization (TSO) algorithm is utilized for parameters optimization, and a bi-directional long short-term memory (BiLSTM) network is incorporated to enhance prediction accuracy for micrometeorological parameters. The proposed TSO-VMD-BiLSTM model is validated through comparison with other models, demonstrating its higher accuracy with the maximum absolute error of only 2.52 m/s, the maximum root mean square error of 0.81, the maximum mean absolute error of only 0.54, and the maximum mean absolute percentage error of 6.89%.

STPDA: Leveraging spatial-temporal patterns for downstream analysis in spatial transcriptomic data.

Spatial transcriptomics, a groundbreaking field in cellular biology, faces the challenge of effectively deciphering complex spatial-temporal gene expression patterns. Traditional data analysis methods often fail to capture the intricate nuances of this data, limiting the depth of understanding in spatial distribution and gene interactions. In response, we present Spatial-Temporal Patterns for Downstream Analysis (STPDA), a sophisticated computational framework tailored for spatial transcriptomic data analysis. STPDA leverages high-resolution mapping to bridge the gap between genomics and histopathology, offering a comprehensive perspective on the spatial dynamics of gene expression within tissues. This approach enables a view of cellular function and organization, marking a paradigm shift in our comprehension of biological systems. By employing Autoregressive Moving Average (ARMA) and Long Short-Term Memory (LSTM) models, STPDA effectively deciphers both global and local spatio-temporal dynamics in cellular environments. This integration of spatial-temporal patterns for downstream analysis offers a transformative approach to spatial transcriptomics data analysis. STPDA excels in various single-cell analytical tasks, including the identification of ligand-receptor interactions and cell type classification. Its ability to harness spatial-temporal patterns not only matches but frequently surpasses the performance of existing state-of-the-art methods. To ensure widespread usability and impact, we have encapsulated STPDA in a scalable and accessible Python package, addressing single-cell tasks through advanced spatial-temporal pattern analysis. This development promises to enhance our understanding of cellular biology, offering novel insights and therapeutic strategies, and represents a substantial advancement in the field of spatial transcriptomics.

Rolling Bearing Remaining Useful Life Prediction Based on CNN-VAE-MBiLSTM.

Ensuring precise prediction of the remaining useful life (RUL) for bearings in rolling machinery is crucial for preventing sudden machine failures and optimizing equipment maintenance strategies. Since the significant interference encountered in real industrial environments and the high complexity of the machining process, accurate and robust RUL prediction of rolling bearings is of tremendous research importance. Hence, a novel RUL prediction model called CNN-VAE-MBiLSTM is proposed in this paper by integrating advantages of convolutional neural network (CNN), variational autoencoder (VAE), and multiple bi-directional long short-term memory (MBiLSTM). The proposed approach includes a CNN-VAE model and a MBiLSTM model. The CNN-VAE model performs well for automatically extracting low-dimensional features from time-frequency spectrum of multi-axis signals, which simplifies the construction of features and minimizes the subjective bias of designers. Based on these features, the MBiLSTM model achieves a commendable performance in the prediction of RUL for bearings, which independently captures sequential characteristics of features in each axis and further obtains differences among multi-axis features. The performance of the proposed approach is validated through an industrial case, and the result indicates that it exhibits a higher accuracy and a better anti-noise capacity in RUL predictions than comparable methods.

OOA-modified Bi-LSTM network: An effective intrusion detection framework for IoT systems.

Currently, the Internet of Things (IoT) generates a huge amount of traffic data in communication and information technology. The diversification and integration of IoT applications and terminals make IoT vulnerable to intrusion attacks. Therefore, it is necessary to develop an efficient Intrusion Detection System (IDS) that guarantees the reliability, integrity, and security of IoT systems. The detection of intrusion is considered a challenging task because of inappropriate features existing in the input data and the slow training process. In order to address these issues, an effective meta heuristic based feature selection and deep learning techniques are developed for enhancing the IDS. The Osprey Optimization Algorithm (OOA) based feature selection is proposed for selecting the highly informative features from the input which leads to an effective differentiation among the normal and attack traffic of network. Moreover, the traditional sigmoid and tangent activation functions are replaced with the Exponential Linear Unit (ELU) activation function to propose the modified Bi-directional Long Short Term Memory (Bi-LSTM). The modified Bi-LSTM is used for classifying the types of intrusion attacks. The ELU activation function makes gradients extremely large during back-propagation and leads to faster learning. This research is analysed in three different datasets such as N-BaIoT, Canadian Institute for Cybersecurity Intrusion Detection Dataset 2017 (CICIDS-2017), and ToN-IoT datasets. The empirical investigation states that the proposed framework obtains impressive detection accuracy of 99.98 %, 99.97 % and 99.88 % on the N-BaIoT, CICIDS-2017, and ToN-IoT datasets, respectively. Compared to peer frameworks, this framework obtains high detection accuracy with better interpretability and reduced processing time.

Fault Diagnosis of Hydraulic Components Based on Multi-Sensor Information Fusion Using Improved TSO-CNN-BiLSTM.

In order to realize the accurate and reliable fault diagnosis of hydraulic systems, a diagnostic model based on improved tuna swarm optimization (ITSO), optimized convolutional neural networks (CNNs), and bi-directional long short-term memory (BiLSTM) networks is proposed. Firstly, sensor selection is implemented using the random forest algorithm to select useful signals from six kinds of physical or virtual sensors including pressure, temperature, flow rate, vibration, motor power, and motor efficiency coefficient. After that, fused features are extracted by CNN, and then, BiLSTM is applied to learn the forward and backward information contained in the data. The ITSO algorithm is adopted to adaptively optimize the learning rate, regularization coefficient, and node number to obtain the optimal CNN-BiLSTM network. Improved Chebyshev chaotic mapping and the nonlinear reduction strategy are adopted to improve population initialization and individual position updating, further promoting the optimization effect of TSO. The experimental results show that the proposed method can automatically extract fusion features and effectively utilize multi-sensor information. The diagnostic accuracies of the plunger pump, cooler, throttle valve, and accumulator are 99.07%, 99.4%, 98.81%, and 98.51%, respectively. The diagnostic results of noisy data with 0 dB, 5 dB, and 10 dB signal-to-noise ratios (SNRs) show that the ITSO-CNN-BiLSTM model has good robustness to noise interference.

Swarm intelligence based deep learning model via improved whale optimization algorithm and Bi-directional long short-term memory for fault diagnosis of chemical processes.

The chemical production process typically possesses complexity and high risks. Effective fault diagnosis is a key technology for ensuring the reliability and safety of chemical production processes. In this study, a comprehensive fault diagnosis method based on time-varying filtering empirical mode decomposition (TVF-EMD), kernel principal component analysis (KPCA), and an improved whale optimization algorithm (WOA) to optimize bi-directional long short-term memory (BiLSTM) is proposed. This research utilizes TVF-EMD and KPCA to analyze and preprocess the raw data, eliminating noise and and reducing the dimensions of the fault data. Subsequently, BiLSTM is employed for fault data classification. To address the hyperparameters within BiLSTM, the enhanced WOA is used for optimization. Finally, the efficacy and superiority of this approach are validated through two fault diagnosis examples.

Estimation of potential wildfire behavior characteristics to assess wildfire danger in southwest China using deep learning schemes.

Accurate estimation of potential wildfire behavior characteristics (PWBC) can improve wildfire danger assessment. However, wildfire behavior has been estimated by most fire spread models with immeasurable uncertainties and difficulties in large-scale applications. In this study, a PWBC estimation model (named PWBC-QR-BiLSTM) was proposed by coupling the Bi-directional Long Short-Term Memory (BiLSTM) and quantile regression (QR) methods. Multi-source data, including fuel, weather, topography, infrastructure, and landscape variables, were input into the PWBC-QR-BiLSTM model to estimate the potential rate of spread (ROS) and fire radiative power (FRP) over western Sichuan of China, and then to estimate the probability density of ROS and FRP. Daily ROS and FRP were extracted from the Global Fire Atlas and the MOD14A1/MYD14A1 product. The optimal PWBC-QR-BiLSTM model was determined using the Non-dominated Sorting Genetic Algorithm Ⅱ (NAGA-Ⅱ). Results showed that the PWBC-QR-BiLSTM performed well in estimating potential ROS and FRP with high accuracy (ROS: R2 > 0.7 and MAPE<30%, FRP: R2 > 0.8 and MAPE<25%). The modal PWBC values extracted from the estimated probability density were closer to the observed values, which can be regarded as a good indicator for wildfire danger assessment. The variable importance analysis also verified that fuel and infrastructure variables played an important role in driving wildfire behavior. This study suggests the potential of utilizing artificial intelligence to estimate PWBC and its probability density to improve the guidance on wildfire management.

EBi-LSTM: an enhanced bi-directional LSTM for time-series data classification by heuristic development of optimal feature integration in brain computer interface.

Generally, time series data is referred to as the sequential representation of data that observes from different applications. Therefore, such expertise can use Electroencephalography (EEG) signals to fetch data regarding brain neural activities in brain-computer interface (BCI) systems. Due to massive and myriads data, the signals are appealed in a non-stationary format that ends with a poor quality resolution. To overcome this existing issue, a new framework of enhanced deep learning methods is proposed. The source signals are collected and undergo feature extraction in four ways. Hence, the features are concatenated to enhance the performance. Subsequently, the concatenated features are given to probability ratio-based Reptile Search Algorithm (PR-RSA) to select the optimal features. Finally, the classification is conducted using Enhanced Bi-directional Long Short-Term Memory (EBi-LSTM), where the hyperparameters are optimized by PR-RSA. Throughout the result analysis, it is confirmed that the offered model obtains elevated classification accuracy, and thus tends to increase the performance.

Deep learning precipitation prediction models combined with feature analysis.

Precise rainfall forecasting modeling assumes a pivotal role in water resource planning and management. Conducting a comprehensive analysis of the rainfall time series and making appropriate adjustments to the forecast model settings based on the characterization results of the rainfall series significantly enhance the accuracy of the forecast model. This paper employed the Mann-Kendall and sliding T mutation tests to identify the mutational components in rainfall between 1961 and 2013 at four meteorological stations located in Henan Province. Wavelet analysis was utilized to determine the periodicity of the precipitation series. The model parameters were adjusted based on the mutation and periodicity findings, and appropriate training and test sets were selected accordingly. Rainfall simulation in Henan Province, China, was conducted using a combination of complementary ensemble empirical mode decomposition (CEEMD) and bi-directional long short-term memory (BiLSTM) networks. The integrated approach aimed at predicting rainfall in the region. The findings of this study demonstrate that the CEEMD-BiLSTM model, coupled with feature analysis, yielded favorable results in terms of prediction accuracy. The model achieved a mean MAE (mean absolute error) of 131.210, a mean MRE (mean relative error) of 0.637, a mean RMSE (root mean square error) of 187.776, and an NSE (Nash-Sutcliffe efficiency) above 0.910. The data processed according to the feature analysis results and then predicted; Zhengzhou, Anyang, Lushi, and Xinyang stations improved by 39.548%, 14.478%, 11.548%, and 19.037% respectively compared with the original prediction model.

SeriesSleepNet: an EEG time series model with partial data augmentation for automatic sleep stage scoring.

Introduction: We propose an automatic sleep stage scoring model, referred to as SeriesSleepNet, based on convolutional neural network (CNN) and bidirectional long short-term memory (bi-LSTM) with partial data augmentation. We used single-channel raw electroencephalography signals for automatic sleep stage scoring. Methods: Our framework was focused on time series information, so we applied partial data augmentation to learn the connected time information in small series. In specific, the CNN module learns the time information of one epoch (intra-epoch) whereas the bi-LSTM trains the sequential information between the adjacent epochs (inter-epoch). Note that the input of the bi-LSTM is the augmented CNN output. Moreover, the proposed loss function was used to fine-tune the model by providing additional weights. To validate the proposed framework, we conducted two experiments using the Sleep-EDF and SHHS datasets. Results and Discussion: The results achieved an overall accuracy of 0.87 and 0.84 and overall F1-score of 0.80 and 0.78 and kappa value of 0.81 and 0.78 for five-class classification, respectively. We showed that the SeriesSleepNet was superior to the baselines based on each component in the proposed framework. Our architecture also outperformed the state-of-the-art methods with overall F1-score, accuracy, and kappa value. Our framework could provide information on sleep disorders or quality of sleep to automatically classify sleep stages with high performance.

CDBA: a novel multi-branch feature fusion model for EEG-based emotion recognition.

EEG-based emotion recognition through artificial intelligence is one of the major areas of biomedical and machine learning, which plays a key role in understanding brain activity and developing decision-making systems. However, the traditional EEG-based emotion recognition is a single feature input mode, which cannot obtain multiple feature information, and cannot meet the requirements of intelligent and high real-time brain computer interface. And because the EEG signal is nonlinear, the traditional methods of time domain or frequency domain are not suitable. In this paper, a CNN-DSC-Bi-LSTM-Attention (CDBA) model based on EEG signals for automatic emotion recognition is presented, which contains three feature-extracted channels. The normalized EEG signals are used as an input, the feature of which is extracted by multi-branching and then concatenated, and each channel feature weight is assigned through the attention mechanism layer. Finally, Softmax was used to classify EEG signals. To evaluate the performance of the proposed CDBA model, experiments were performed on SEED and DREAMER datasets, separately. The validation experimental results show that the proposed CDBA model is effective in classifying EEG emotions. For triple-category (positive, neutral and negative) and four-category (happiness, sadness, fear and neutrality), the classification accuracies were respectively 99.44% and 99.99% on SEED datasets. For five classification (Valence 1-Valence 5) on DREAMER datasets, the accuracy is 84.49%. To further verify and evaluate the model accuracy and credibility, the multi-classification experiments based on ten-fold cross-validation were conducted, the elevation indexes of which are all higher than other models. The results show that the multi-branch feature fusion deep learning model based on attention mechanism has strong fitting and generalization ability and can solve nonlinear modeling problems, so it is an effective emotion recognition method. Therefore, it is helpful to the diagnosis and treatment of nervous system diseases, and it is expected to be applied to emotion-based brain computer interface systems.

Intracranial hemorrhage detection in 3D computed tomography images using a bi-directional long short-term memory network-based modified genetic algorithm.

Introduction: Intracranial hemorrhage detection in 3D Computed Tomography (CT) brain images has gained more attention in the research community. The major issue to deal with the 3D CT brain images is scarce and hard to obtain the labelled data with better recognition results. Methods: To overcome the aforementioned problem, a new model has been implemented in this research manuscript. After acquiring the images from the Radiological Society of North America (RSNA) 2019 database, the region of interest (RoI) was segmented by employing Otsu's thresholding method. Then, feature extraction was performed utilizing Tamura features: directionality, contrast, coarseness, and Gradient Local Ternary Pattern (GLTP) descriptors to extract vectors from the segmented RoI regions. The extracted vectors were dimensionally reduced by proposing a modified genetic algorithm, where the infinite feature selection technique was incorporated with the conventional genetic algorithm to further reduce the redundancy within the regularized vectors. The selected optimal vectors were finally fed to the Bi-directional Long Short Term Memory (Bi-LSTM) network to classify intracranial hemorrhage sub-types, such as subdural, intraparenchymal, subarachnoid, epidural, and intraventricular. Results: The experimental investigation demonstrated that the Bi-LSTM based modified genetic algorithm obtained 99.40% sensitivity, 99.80% accuracy, and 99.48% specificity, which are higher compared to the existing machine learning models: Naïve Bayes, Random Forest, Support Vector Machine (SVM), Recurrent Neural Network (RNN), and Long Short-Term Memory (LSTM) network.

[Automatic sleep staging model based on single channel electroencephalogram signal].

Sleep staging is the basis for solving sleep problems. There's an upper limit for the classification accuracy of sleep staging models based on single-channel electroencephalogram (EEG) data and features. To address this problem, this paper proposed an automatic sleep staging model that mixes deep convolutional neural network (DCNN) and bi-directional long short-term memory network (BiLSTM). The model used DCNN to automatically learn the time-frequency domain features of EEG signals, and used BiLSTM to extract the temporal features between the data, fully exploiting the feature information contained in the data to improve the accuracy of automatic sleep staging. At the same time, noise reduction techniques and adaptive synthetic sampling were used to reduce the impact of signal noise and unbalanced data sets on model performance. In this paper, experiments were conducted using the Sleep-European Data Format Database Expanded and the Shanghai Mental Health Center Sleep Database, and achieved an overall accuracy rate of 86.9% and 88.9% respectively. When compared with the basic network model, all the experimental results outperformed the basic network, further demonstrating the validity of this paper's model, which can provide a reference for the construction of a home sleep monitoring system based on single-channel EEG signals.

Change Point Detection for Fine-Grained MFR Work Modes with Multi-Head Attention-Based Bi-LSTM Network.

Detection of the changes in Multi-Functional Radar (MFR) work modes is a critical situation assessment task for Electronic Support Measure (ESM) systems. There are two major challenges that must be addressed: (i) The received radar pulse stream may contain multiple work mode segments of unknown number and duration, which makes the Change Point Detection (CPD) difficult. (ii) Modern MFRs can produce a variety of parameter-level (fine-grained) work modes with complex and flexible patterns, which are challenging to detect through traditional statistical methods and basic learning models. To address the challenges, a deep learning framework is proposed for fine-grained work mode CPD in this paper. First, the fine-grained MFR work mode model is established. Then, a multi-head attention-based bi-directional long short-term memory network is introduced to abstract high-order relationships between successive pulses. Finally, temporal features are adopted to predict the probability of each pulse being a change point. The framework further improves the label configuration and the loss function of training to mitigate the label sparsity problem effectively. The simulation results showed that compared with existing methods, the proposed framework effectively improves the CPD performance at parameter-level. Moreover, the F1-score was increased by 4.15% under hybrid non-ideal conditions.

[Construction of an epileptic seizure prediction model using a semi-supervised method of generative adversarial and long short term memory network combined with Stockwell transform].

OBJECTIVE: To propose a semi-supervised epileptic seizure prediction model (ST-WGAN-GP-Bi-LSTM) to enhance the prediction performance by improving time-frequency analysis of electroencephalogram (EEG) signals, enhancing the stability of the unsupervised feature learning model and improving the design of back-end classifier. METHODS: Stockwell transform (ST) of the epileptic EEG signals was performed to locate the time-frequency information by adaptive adjustment of the resolution and retaining the absolute phase to obtain the time-frequency inputs. When there was no overlap between the generated data distribution and the real EEG data distribution, to avoid failure of feature learning due to a constant JS divergence, Wasserstein GAN was used as a feature learning model, and the cost function based on EM distance and gradient penalty strategy was adopted to constrain the unsupervised training process to allow the generation of a high-order feature extractor. A temporal prediction model was finally constructed based on a bi-directional long short term memory network (Bi-LSTM), and the classification performance was improved by obtaining the temporal correlation between high-order time-frequency features. The CHB-MIT scalp EEG dataset was used to validate the proposed patient-specific seizure prediction method. RESULTS: The AUC, sensitivity, and specificity of the proposed method reached 90.40%, 83.62%, and 86.69%, respectively. Compared with the existing semi-supervised methods, the propose method improved the original performance by 17.77%, 15.41%, and 53.66%. The performance of this method was comparable to that of a supervised prediction model based on CNN. CONCLUSION: The utilization of ST, WGAN-GP, and Bi-LSTM effectively improves the prediction performance of the semi-supervised deep learning model, which can be used for optimization of unsupervised feature extraction in epileptic seizure prediction.

[Research on migraine time-series features classification based on small-sample functional magnetic resonance imaging data].

The extraction of neuroimaging features of migraine patients and the design of identification models are of great significance for the auxiliary diagnosis of related diseases. Compared with the commonly used image features, this study directly uses time-series signals to characterize the functional state of the brain in migraine patients and healthy controls, which can effectively utilize the temporal information and reduce the computational effort of classification model training. Firstly, Group Independent Component Analysis and Dictionary Learning were used to segment different brain areas for small-sample groups and then the regional average time-series signals were extracted. Next, the extracted time series were divided equally into multiple subseries to expand the model input sample. Finally, the time series were modeled using a bi-directional long-short term memory network to learn the pre-and-post temporal information within each time series to characterize the periodic brain state changes to improve the diagnostic accuracy of migraine. The results showed that the classification accuracy of migraine patients and healthy controls was 96.94%, the area under the curve was 0.98, and the computation time was relatively shorter. The experiments indicate that the method in this paper has strong applicability, and the combination of time-series feature extraction and bi-directional long-short term memory network model can be better used for the classification and diagnosis of migraine. This work provides a new idea for the lightweight diagnostic model based on small-sample neuroimaging data, and contributes to the exploration of the neural discrimination mechanism of related diseases.