A review on transfer learning in EEG signal analysis

Cross-Domain Transfer of EEG to EEG or ECG Learning for CNN Classification Models

Electroencephalography (EEG) is often used to evaluate several types of neurological brain disorders because of its noninvasive and high temporal resolution. In contrast to electrocardiography (ECG), EEG can be uncomfortable and inconvenient for patients. Moreover, deep-learning techniques require a large dataset and a long time for training from scratch. Therefore, in this study, EEG-EEG or EEG-ECG transfer learning strategies were applied to explore their effectiveness for the training of simple cross-domain convolutional neural networks (CNNs) used in seizure prediction and sleep staging systems, respectively. The seizure model detected interictal and preictal periods, whereas the sleep staging model classified signals into five stages. The patient-specific seizure prediction model with six frozen layers achieved 100% accuracy for seven out of nine patients and required only 40 s of training time for personalization. Moreover, the cross-signal transfer learning EEG-ECG model for sleep staging achieved an accuracy approximately 2.5% higher than that of the ECG model; additionally, the training time was reduced by >50%. In summary, transfer learning from an EEG model to produce personalized models for a more convenient signal can both reduce the training time and increase the accuracy; moreover, challenges such as data insufficiency, variability, and inefficiency can be effectively overcome.

A Greedy Optimized Intelligent Framework for Early Detection of Alzheimer's Disease Using EEG Signal

Recent researchers have been drawn to the analysis of electroencephalogram (EEG) signals in order to confirm the disease and severity range by viewing the EEG signal which has complicated the dataset. The conventional models such as machine learning, classifiers, and other mathematical models achieved the lowest classification score. The current study proposes to implement a novel deep feature with the best solution for EEG signal analysis and severity specification. A greedy sandpiper-based recurrent neural system (SbRNS) model for predicting Alzheimer's disease (AD) severity has been proposed. The filtered data are used as input for the feature analysis and the severity range is divided into three classes: low, medium, and high. The designed approach was then implemented in the matrix laboratory (MATLAB) system, and the effectiveness score was calculated using key metrics such as precision, recall, specificity, accuracy, and misclassification score. The validation results show that the proposed scheme achieved the best classification outcome.

A Scalable Framework for Closed-Loop Neuromodulation with Deep Learning

Closed-loop neuromodulation measures dynamic neural or physiological activity to optimize interventions for clinical and nonclinical behavioral, cognitive, wellness, attentional, or general task performance enhancement. Conventional closed-loop stimulation approaches can contain biased biomarker detection (decoders and error-based triggering) and stimulation-type application. We present and verify a novel deep learning framework for designing and deploying flexible, data-driven, automated closed-loop neuromodulation that is scalable using diverse datasets, agnostic to stimulation technology (supporting multi-modal stimulation: tACS, tDCS, tFUS, TMS), and without the need for personalized ground-truth performance data. Our approach is based on identified periods of responsiveness - detected states that result in a change in performance when stimulation is applied compared to no stimulation. To demonstrate our framework, we acquire, analyze, and apply a data-driven approach to our open sourced GX dataset, which includes concurrent physiological (ECG, EOG) and neuronal (EEG) measures, paired with continuous vigilance/attention-fatigue tracking, and High-Definition transcranial electrical stimulation (HD-tES). Our framework's decision process for intervention application identified 88.26% of trials as correct applications, showed potential improvement with varying stimulation types, or missed opportunities to stimulate, whereas 11.25% of trials were predicted to stimulate at inopportune times. With emerging datasets and stimulation technologies, our unifying and integrative framework; leveraging deep learning (Convolutional Neural Networks - CNNs); demonstrates the adaptability and feasibility of automated multimodal neuromodulation for both clinical and nonclinical applications.

Classification of motor imagery electroencephalogram signals by using adaptive cross-subject transfer learning

Introduction

Electroencephalogram (EEG)-based motor imagery (MI) classification is an important aspect in brain-computer interfaces (BCIs), which bridges between neural system and computer devices decoding brain signals into recognizable machine commands. However, due to the small number of training samples of MI electroencephalogram (MI-EEG) for a single subject and the great individual differences of MI-EEG among different subjects, the generalization and accuracy of the model on the specific MI task may be poor.

Methods

To solve these problems, an adaptive cross-subject transfer learning algorithm is proposed, which is based on kernel mean matching (KMM) and transfer learning adaptive boosting (TrAdaBoost) method. First, the common spatial pattern (CSP) is used to extract the spatial features. Then, in order to make the feature distribution more similar among different subjects, the KMM algorithm is used to compute a sample weight matrix for aligning the mean between source and target domains and reducing distribution differences among different subjects. Finally, the sample weight matrix from KMM is used as the initialization weight of TrAdaBoost, and then TrAdaBoost is used to adaptively select source domain samples that are closer to the target task distribution to assist in building a classification model.

Results

In order to verify the effectiveness and feasibility of the proposed method, the algorithm is applied to BCI Competition IV datasets and in-house datasets. The results show that the average classification accuracy of the proposed method on the public datasets is 89.1%, and the average classification accuracy on the in-house datasets is 80.4%.

Discussion

Compared with the existing methods, the proposed method effectively improves the classification accuracy of MI-EEG signals. At the same time, this paper also applies the proposed algorithm to the in-house dataset, the results verify the effectiveness of the algorithm again, and the results of this study have certain clinical guiding significance for brain rehabilitation.

Emerging Materials, Wearables, and Diagnostic Advancements in Therapeutic Treatment of Brain Diseases

Among the most critical health issues, brain illnesses, such as neurodegenerative conditions and tumors, lower quality of life and have a significant economic impact. Implantable technology and nano-drug carriers have enormous promise for cerebral brain activity sensing and regulated therapeutic application in the treatment and detection of brain illnesses. Flexible materials are chosen for implantable devices because they help reduce biomechanical mismatch between the implanted device and brain tissue. Additionally, implanted biodegradable devices might lessen any autoimmune negative effects. The onerous subsequent operation for removing the implanted device is further lessened with biodegradability. This review expands on current developments in diagnostic technologies such as magnetic resonance imaging, computed tomography, mass spectroscopy, infrared spectroscopy, angiography, and electroencephalogram while providing an overview of prevalent brain diseases. As far as we are aware, there hasn't been a single review article that addresses all the prevalent brain illnesses. The reviewer also looks into the prospects for the future and offers suggestions for the direction of future developments in the treatment of brain diseases.

Supervised Machine Learning and Deep Learning Techniques for Epileptic Seizure Recognition Using EEG Signals-A Systematic Literature Review

Electroencephalography (EEG) is a complicated, non-stationary signal that requires extensive preprocessing and feature extraction approaches to be accurately analyzed. In recent times, Deep learning (DL) has shown great promise in exploiting the characteristics of EEG signals as it can learn relevant features from raw data autonomously. Although studies involving DL have become more common in the last two years, the topic of whether DL truly delivers advantages over conventional Machine learning (ML) methodologies remains unsettled. This study aims to present a detailed overview of the main challenges in the field of seizure detection, prediction, and classification utilizing EEG data, and the approaches taken to solve them using ML and DL methods. A systematic review was conducted surveying peer-reviewed publications published between 2017 and 16 July 2022 using two scientific databases (Web of Science and Scopus) totaling 6822 references after discarding duplicate publications. Whereas 2262 articles were screened based on the title, abstract, and keywords, only 214 were eligible for full-text assessment. A total of 91 papers have been included in this survey after meeting the eligible inclusion and exclusion criteria. The most significant findings from the review are summarized, and several important concepts involving ML and DL for seizure detection, prediction, and classification are discussed in further depth. This review aims to learn more about the different approaches for identifying different types and stages of epileptic seizures, which may then be employed to enhance the lives of epileptic patients in the future, as well as aid experts in the field.

A novel heterogeneous transfer learning method based on data stitching for the sequential coding brain computer interface

OBJECTIVE

For the brain computer interface (BCI), it is necessary to collect enough electroencephalography (EEG) signals to train the classification model. When the operation dimension of BCI is large, it will bring great burden to data acquisition. Fortunately, this problem can be solved by our proposed transfer learning method.

METHOD

For the sequential coding experimental paradigm, the multi-band data stitching with label alignment and tangent space mapping (MDSLATSM) algorithm is proposed as a novel heterogeneous transfer learning method. After filtering by multi-band filtering, the artificial signals can be obtained by data stitching from the source domain, which build a bridge between the source domain and target domain. To make the distribution of two domains closer, their covariance matrices are aligned by label alignment. After mapping to the tangent space, the features are extracted from the Riemannian manifold. Finally, the classification results are obtained with feature selection and classification.

RESULTS

Our data set includes the EEG signals from 16 subjects. For the heterogeneous transfer learning of cross-label, the average classification accuracy is 78.28%. MDSLATSM is also tested for cross-subject, and the average classification accuracy is 64.01%, which is better than existing methods.

SIGNIFICANCE

Combining multi-band filtering, data stitching, label alignment and tangent space mapping, a novel heterogeneous transfer learning method can be achieved with superior performance, which promotes the practical application of the BCI systems.

Segment alignment based cross-subject motor imagery classification under fading data

Motor imagery (MI) aims to use brain imagination without actual body activities to support motor learning, and machine learning algorithms such as common spatial patterns (CSP) are proven effective in the analysis of MI signals. In the conventional machine learning-based approaches, there are two main difficulties in feature extraction and recognition of MI signals: high personalization and data fading. The high personalization problem is due to the multi-subject nature when collecting MI signals, and the data fading problem as a recurring issue in MI signal quality is first raised by us but is not widely discussed at present. Aiming to solve the above two mentioned problems, a cross-subject fading data classification approach with segment alignment is proposed to classify the fading data of one single target with the model trained with the normal data of multiple sources in this paper. he effectiveness of proposed method is verified via two experiments: a dataset-based experiment with the dataset from BCI Competition and a lab-based experiment designed and conducted by us. The experimental results obtained from both experiments show that the proposed method can obtain optimal classification performance effectively under different fading levels with data from different subjects.

Few-Electrode EEG from the Wearable Devices Using Domain Adaptation for Depression Detection

Nowadays, major depressive disorder (MDD) has become a crucial mental disease that endangers human health. Good results have been achieved by electroencephalogram (EEG) signals in the detection of depression. However, EEG signals are time-varying, and the distributions of the different subjects' data are non-uniform, which poses a bad influence on depression detection. In this paper, the deep learning method with domain adaptation is applied to detect depression based on EEG signals. Firstly, the EEG signals are preprocessed and then transformed into pictures by two methods: the first one is to present the three channels of EEG separately in the same image, and the second one is the RGB synthesis of the three channels of EEG. Finally, the training and prediction are performed in the domain adaptation model. The results indicate that the domain adaptation model can effectively extract EEG features and obtain an average accuracy of 77.0 ± 9.7%. This paper proves that the domain adaptation method can effectively weaken the inherent differences of EEG signals, making the diagnosis of different users more accurate.

Subject adaptation convolutional neural network for EEG-based motor imagery classification

Objective.Deep transfer learning has been widely used to address the nonstationarity of electroencephalogram (EEG) data during motor imagery (MI) classification. However, previous deep learning approaches suffer from limited classification accuracy because the temporal and spatial features cannot be effectively extracted.Approach.Here, we propose a novel end-to-end deep subject adaptation convolutional neural network (SACNN) to handle the problem of EEG-based MI classification. Our proposed model jointly optimizes three modules, i.e. a feature extractor, a classifier, and a subject adapter. Specifically, the feature extractor simultaneously extracts the temporal and spatial features from the raw EEG data using a parallel multiscale convolution network. In addition, we design a subject adapter to reduce the feature distribution shift between the source and target subjects by using the maximum mean discrepancy. By minimizing the classification loss and the distribution discrepancy, the model is able to extract the temporal-spatial features to the prediction of a new subject.Main results.Extensive experiments are carried out on three EEG-based MI datasets, i.e. brain-computer interface (BCI) competition IV dataset IIb, BCI competition III dataset IVa, and BCI competition IV dataset I, and the average accuracy reaches to 86.42%, 81.71% and 79.35% on the three datasets respectively. Furthermore, the statistical analysis also indicates the significant performance improvement of SACNN.Significance.This paper reveals the importance of the temporal-spatial features on EEG-based MI classification task. Our proposed SACNN model can make fully use of the temporal-spatial information to achieve the purpose.

Emotion recognition based on multi-modal physiological signals and transfer learning

In emotion recognition based on physiological signals, collecting enough labeled data of a single subject for training is time-consuming and expensive. The physiological signals’ individual differences and the inherent noise will significantly affect emotion recognition accuracy. To overcome the difference in subject physiological signals, we propose a joint probability domain adaptation with the bi-projection matrix algorithm (JPDA-BPM). The bi-projection matrix method fully considers the source and target domain’s different feature distributions. It can better project the source and target domains into the feature space, thereby increasing the algorithm’s performance. We propose a substructure-based joint probability domain adaptation algorithm (SSJPDA) to overcome physiological signals’ noise effect. This method can avoid the shortcomings that the domain level matching is too rough and the sample level matching is susceptible to noise. In order to verify the effectiveness of the proposed transfer learning algorithm in emotion recognition based on physiological signals, we verified it on the database for emotion analysis using physiological signals (DEAP dataset). The experimental results show that the average recognition accuracy of the proposed SSJPDA-BPM algorithm in the multimodal fusion physiological data from the DEAP dataset is 63.6 and 64.4% in valence and arousal, respectively. Compared with joint probability domain adaptation (JPDA), the performance of valence and arousal recognition accuracy increased by 17.6 and 13.4%, respectively.

Decoding working memory-related information from repeated psychophysiological EEG experiments using convolutional and contrastive neural networks

OBJECTIVE

Extracting reliable information from EEG signals is difficult because the low signal-to-noise ratio and significant intersubject variability seriously hinder statistical analyses. However, recent advances in explainable machine learning open a new strategy to address this problem.

APPROACH

The current study evaluates this approach using results from the classification and decoding of electrical brain activity associated with information retention. We designed four neural network models differing in architecture, training strategies, and input representation to classify single experimental trials of a working memory task.

MAIN RESULTS

Our best models achieved an accuracy of 65.29$±0.76 and Matthews correlation coefficient of 0.288±0.018, outperforming the reference model trained on the same data. The highest correlation between classification score and behavioral performance was 0.36 (p=0.0007). Using analysis of input perturbation, we estimated the importance of EEG channels and frequency bands in the task at hand. The set of essential features identified for each network varies. We identified a subset of features common to all models that identified brain regions and frequency bands consistent with current neurophysiological knowledge of the processes critical to attention and working memory. Finally, we proposed sanity checks to examine further the robustness of each model's set of features.

SIGNIFICANCE

Our results indicate that explainable deep learning is a powerful tool for decoding information from EEG signals. It is crucial to train and analyze a range of models to identify stable and reliable features. Our results highlight the need for explainable modeling as the model with the highest accuracy appeared to use residual artifactual activity.

Single-Source to Single-Target Cross-Subject Motor Imagery Classification Based on Multisubdomain Adaptation Network

In the electroencephalography (EEG) based cross-subject motor imagery (MI) classification task, the device and subject problems can cause the time-related data distribution shift problem. In a single-source to single-target (STS) MI classification task, such a shift problem will certainly provoke an increase in the overall data distribution difference between the source and target domains, giving rise to poor classification accuracy. In this paper, a novel multi-subdomain adaptation method (MSDAN) is proposed to solve the shift problem and improve the classification accuracy of the traditional approaches. In the proposed MSDAN, the adaptation losses in both class-related and time-related subdomains (that are divided by different data labels and session labels) are obtained by measuring the distribution differences between the source and target subdomains. Then, the adaptation and classification losses in the loss function of MSDAN are minimized concurrently. To illustrate the application value of the proposed method, our method is applied to solve the STS MI classification task about data analysis with respect to the brain-computer interface (BCI) competition III-IVa dataset. The resultant experiment results demonstrate that compared with other well-known domain adaptation and deep learning methods, the proposed method is capable of solving the time-related data distribution problem at higher classification accuracy.

Flexible coding scheme for robotic arm control driven by motor imagery decoding

OBJECTIVE

Brain computer interface (BCI) technology is a new way of information exchange, which can effectively convert physiological signals into control instructions of machines. Due to its spontaneity and device-independence, motor imagery (MI) electroencephalography (EEG) signal is used as a common BCI signal source to achieve direct control of external devices. However, the generalization ability of current classification model of MI tasks is still limited. Moreover, the real time prototype is far from established in practice.

APPROACH

In order to solve these problems, this paper proposed an optimized neural network architecture based on our previous work. Firstly, the artifact components in MI-EEG signal are removed by using the threshold and threshold function related to the artifact removal evaluation index, and then the data is augmented by the empirical mode decomposition (EMD) algorithm. Furthermore, ensemble learning (EL) method and fine-tuning strategy in transfer learning (TL) are used to optimize the classification model. Finally, combined with the flexible binary encoding strategy, the EEG signal recognition results are mapped to the control commands of the robotic arm, which realizes the multiple degree of freedom control of the robotic arm.

MAIN RESULTS

The results show that EMD has an obvious data amount enhancement effect on small dataset, and the EL and TL can improve intra-subject and inter-subject model evaluation performance, respectively. The use of binary coding method realizes the expansion of control instructions, i.e., four kinds of MI-EEG signals are used to complete the control of 7 degrees of freedom of the robotic arm.

SIGNIFICANCE

Our work not only improves the classification accuracy of subject and the generality of classification model, but also extends the BCI control instruction set.

Black hole algorithm with convolutional neural networks for the creation of brain-computer interface based in visual perception and visual imagery

Non-invasive brain-computer interfaces can be implemented through different paradigms, the most used one being motor imagery and evoked potentials, although recently there has been an interest in paradigms based on perception and visual imagery. Following this approach, this work demonstrates the classification of visual imagery, visual perception and also the possibility of knowledge transfer between these two domains from EEG signals using convolutional neural networks. Also, we propose an adequate framework for such classification, which uses convolutional neural networks and the black hole heuristic algorithm for the search for optimal neural network structures.

Multiclass Classification of Imagined Speech Vowels and Words of Electroencephalography Signals Using Deep Learning

The paper’s emphasis is on the imagined speech decoding of electroencephalography (EEG) neural signals of individuals in accordance with the expansion of the brain-computer interface to encompass individuals with speech problems encountering communication challenges. Decoding an individual’s imagined speech from nonstationary and nonlinear EEG neural signals is a complex task. Related research work in the field of imagined speech has revealed that imagined speech decoding performance and accuracy require attention to further improve. The evolution of deep learning technology increases the likelihood of decoding imagined speech from EEG signals with enhanced performance. We proposed a novel supervised deep learning model that combined the temporal convolutional networks and the convolutional neural networks with the intent of retrieving information from the EEG signals. The experiment was carried out using an open-access dataset of fifteen subjects’ imagined speech multichannel signals of vowels and words. The raw multichannel EEG signals of multiple subjects were processed using discrete wavelet transformation technique. The model was trained and evaluated using the preprocessed signals, and the model hyperparameters were adjusted to achieve higher accuracy in the classification of imagined speech. The experiment results demonstrated that the multiclass imagined speech classification of the proposed model exhibited a higher overall accuracy of 0.9649 and a classification error rate of 0.0350. The results of the study indicate that individuals with speech difficulties might well be able to leverage a noninvasive EEG-based imagined speech brain-computer interface system as one of the long-term alternative artificial verbal communication mediums.

A data-driven machine learning approach for brain-computer interfaces targeting lower limb neuroprosthetics

Prosthetic devices that replace a lost limb have become increasingly performant in recent years. Recent advances in both software and hardware allow for the decoding of electroencephalogram (EEG) signals to improve the control of active prostheses with brain-computer interfaces (BCI). Most BCI research is focused on the upper body. Although BCI research for the lower extremities has increased in recent years, there are still gaps in our knowledge of the neural patterns associated with lower limb movement. Therefore, the main objective of this study is to show the feasibility of decoding lower limb movements from EEG data recordings. The second aim is to investigate whether well-known neuroplastic adaptations in individuals with an amputation have an influence on decoding performance. To address this, we collected data from multiple individuals with lower limb amputation and a matched able-bodied control group. Using these data, we trained and evaluated common BCI methods that have already been proven effective for upper limb BCI. With an average test decoding accuracy of 84% for both groups, our results show that it is possible to discriminate different lower extremity movements using EEG data with good accuracy. There are no significant differences (p = 0.99) in the decoding performance of these movements between healthy subjects and subjects with lower extremity amputation. These results show the feasibility of using BCI for lower limb prosthesis control and indicate that decoding performance is not influenced by neuroplasticity-induced differences between the two groups.

Exploiting Multiple EEG Data Domains with Adversarial Learning

Electroencephalography (EEG) is shown to be a valuable data source for evaluating subjects' mental states. However, the interpretation of multi-modal EEG signals is challenging, as they suffer from poor signal-to-noise-ratio, are highly subject-dependent, and are bound to the equipment and experimental setup used, (i.e. domain). This leads to machine learning models often suffer from poor generalization ability, where they perform significantly worse on real-world data than on the exploited training data. Recent research heavily focuses on cross-subject and cross-session transfer learning frameworks to reduce domain calibration efforts for EEG signals. We argue that multi-source learning via learning domain-invariant representations from multiple data-sources is a viable alternative, as the available data from different EEG data-source domains (e.g., subjects, sessions, experimental setups) grow massively. We propose an adversarial inference approach to learn data-source invariant representations in this context, enabling multi-source learning for EEG-based brain- computer interfaces. We unify EEG recordings from different source domains (i.e., emotion recognition datasets SEED, SEED-IV, DEAP, DREAMER), and demonstrate the feasibility of our invariant representation learning approach in suppressing data- source-relevant information leakage by 35% while still achieving stable EEG-based emotion classification performance.

Semi-Supervised Transfer Learning Method for Bearing Fault Diagnosis with Imbalanced Data

Fault diagnosis is essential for assuring the safety and dependability of rotating machinery systems. Several emerging techniques, especially artificial intelligence-based technologies, are used to overcome the difficulties in this field. In most engineering scenarios, machines perform in normal conditions, which implies that fault data may be hard to acquire and limited. Therefore, the data imbalance and the deficiency of labels are practical challenges in the fault diagnosis of machinery bearings. Among the mainstream methods, transfer learning-based fault diagnosis is highly effective, as it transfers the results of previous studies and integrates existing resources. The knowledge from the source domain is transferred via Domain Adversarial Training of Neural Networks (DANN) while the dataset of the target domain is partially labeled. A semi-supervised framework based on uncertainty-aware pseudo-label selection (UPS) is adopted in parallel to improve the model performance by utilizing abundant unlabeled data. Through experiments on two bearing datasets, the accuracy of bearing fault classification surpassed the independent approaches.

Comparison of cross-subject EEG emotion recognition algorithms in the BCI Controlled Robot Contest in World Robot Contest 2021

Electroencephalogram (EEG) data depict various emotional states and reflect brain activity. There has been increasing interest in EEG emotion recognition in brain–computer interface systems (BCIs). In the World Robot Contest (WRC), the BCI Controlled Robot Contest successfully staged an emotion recognition technology competition. Three types of emotions (happy, sad, and neutral) are modeled using EEG signals. In this study, 5 methods employed by different teams are compared. The results reveal that classical machine learning approaches and deep learning methods perform similarly in offline recognition, whereas deep learning methods perform better in online cross-subject decoding.

Transfer Discriminative Dictionary Pair Learning Approach for Across-Subject EEG Emotion Classification

Electroencephalogram (EEG) signals are not easily camouflaged, portable, and noninvasive. It is widely used in emotion recognition. However, due to the existence of individual differences, there will be certain differences in the data distribution of EEG signals in the same emotional state of different subjects. To obtain a model that performs well in classifying new subjects, traditional emotion recognition approaches need to collect a large number of labeled data of new subjects, which is often unrealistic. In this study, a transfer discriminative dictionary pair learning (TDDPL) approach is proposed for across-subject EEG emotion classification. The TDDPL approach projects data from different subjects into the domain-invariant subspace, and builds a transfer dictionary pair learning based on the maximum mean discrepancy (MMD) strategy. In the subspace, TDDPL learns shared synthesis and analysis dictionaries to build a bridge of discriminative knowledge from source domain (SD) to target domain (TD). By minimizing the reconstruction error and the inter-class separation term for each sub-dictionary, the learned synthesis dictionary is discriminative and the learned low-rank coding is sparse. Finally, a discriminative classifier in the TD is constructed on the classifier parameter, analysis dictionary and projection matrix, without the calculation of coding coefficients. The effectiveness of the TDDPL approach is verified on SEED and SEED IV datasets.

Multi-Source Selection Transfer Learning with Privacy-Preserving

Transfer learning has ability to create learning task of weakly labeled or unlabeled target domain by using knowledge of source domain to help, which can effectively improve the performance of target learning task. At present, the increased awareness of privacy protection restricts access to data sources and poses new challenges to the development of transfer learning. However, the research on privacy protection in transfer learning is very rare. The existing work mainly uses differential privacy technology and does not consider the distribution difference between data sources, or does not consider the conditional probability distribution of data, which causes negative transfer to harm the effect of algorithm. Therefore, this paper proposes multi-source selection transfer learning algorithm with privacy-preserving MultiSTLP, which is used in scenarios where target domain contains unlabeled data sets with only a small amount of group probability information and multiple source domains with a large number of labeled data sets. Group probability means that the class label of each sample in target data set is unknown, but the probability of each class in a given data group is available, and multiple source domains indicate that there are more than two source domains. The number of data set contains more than two data sets of source domain and one data set of target domain. The algorithm adapts to the marginal probability distribution and conditional probability distribution differences between domains, and can protect the privacy of target data and improve classification accuracy by fusing the idea of multi-source transfer learning and group probability into support vector machine. At the same time, it can select the representative dataset in source domains to improve efficiency relied on speeding up the training process of algorithm. Experimental results on several real datasets show the effectiveness of MultiSTLP, and it also has some advantages compared with the state-of-the-art transfer learning algorithm.

An insight into diagnosis of depression using machine learning techniques: a systematic review

BACKGROUND

In this modern era, depression is one of the most prevalent mental disorders from which millions of individuals are affected today. The symptoms of depression are heterogeneous and often coincide with other disorders such as bipolar disorder, Parkinson's, schizophrenia, etc. It is a serious mental illness that may lead to other health problems if left untreated. Currently, identifying individuals with depression is totally based on the expertise of the clinician's experience. In order to assist clinicians in identifying the characteristics and classifying depressed people, different types of data modalities and machine learning techniques have been incorporated by researchers in this field. This study aims to find the answers to some important questions related to the trend of publications, data modality, machine learning models, dataset usage, pre-processing techniques and feature extraction and selection techniques that are prevalent and guide the direction of future research on depression diagnosis.

METHODS

This systematic review was conducted using a broad range of articles from two major databases: IEEE Xplore and PubMed. Studies ranging from the years 2011 to April 2021 were retrieved from the databases resulting in a total of 590 articles (53 articles from the IEEE Xplore database and 537 articles from the PubMed database). Out of those, the articles which satisfied the defined inclusion criteria were investigated for further analysis.

RESULTS

A total of 135 articles were identified and analysed for this review. High growth in the number of publications has been observed in recent years. Furthermore, significant diversity in the use of data modalities and machine learning classifiers has also been noted in this study. fMRI data with an SVM classifier was found to be the most popular choice among researchers. In most of the studies, data scarcity and small sample size, particularly for neuroimaging data are major concerns. The use of identical data pre-processing tools for similar data modalities can be seen. This study also provides statistical analysis of the current framework with respect to the modality, machine learning classifier, sample size and accuracy by applying one-way ANOVA and the Tukey - Kramer test.

CONCLUSION

The results indicate that an effective fusion of machine learning techniques with a potential data modality has a promising future for assisting clinicians in automatic depression diagnosis.

Key Feature Extraction Method of Electroencephalogram Signal by Independent Component Analysis for Athlete Selection and Training

Emotion is an important expression generated by human beings to external stimuli in the process of interaction with the external environment. It affects all aspects of our lives all the time. Accurate identification of human emotional states and further application in artificial intelligence can better improve and assist human life. Therefore, the research on emotion recognition has attracted the attention of many scholars in the field of artificial intelligence in recent years. Brain electrical signal conversion becomes critical, and it needs a brain electrical signal processing method to extract the effective signal to realize the human-computer interaction However, nonstationary nonlinear characteristics of EEG signals bring great challenge in characteristic signal extraction. At present, although there are many feature extraction methods, none of them can reflect the global feature of the signal. The following solutions are used to solve the above problems: (1) this paper proposed an ICA and sample entropy algorithm-based framework for feature extraction of EEG signals, which has not been applied for EEG and (2) simulation signals were used to verify the feasibility of this method, and experiments were carried out on two real-world data sets, to show the advantages of the new algorithm in feature extraction of EEG signals.

Analytic beamformer transformation for transfer learning in motion-onset visual evoked potential decoding

OBJECTIVE

While decoders of EEG-based event-related potentials (ERPs) are routinely tailored to the individual user to maximize performance, developing them on populations for individual usage has proven much more challenging. We propose the analytic beamformer transformation (ABT) to extract phase and/or magnitude information from spatiotemporal ERPs in response to motion-onset stimulation.

APPROACH

We have tested ABT on 52 motion-onset visual evoked potential (mVEP) datasets from 26 healthy subjects and compared the classification accuracy of support vector machine (SVM), spatiotemporal beamformer (stBF) and stepwise linear discriminant analysis (SWLDA) when trained on individual subjects and on a population thereof.

MAIN RESULTS

When using phase- and combined phase/magnitude information extracted by ABT, we show significant improvements in accuracy of population-trained classifiers applied to individual users (p<0.001). We also show that 450 epochs are needed for a correct functioning of ABT, which corresponds to 2 minutes of paradigm stimulation.

SIGNIFICANCE

We have shown that ABT can be used to create population-trained mVEP classifiers using a limited number of epochs. We expect this to pertain to other ERPs or synchronous stimulation paradigms, allowing for a more effective, population-based training of visual BCIs. Finally, as ABT renders recordings across subjects more structurally invariant, it could be used for transfer learning purposes in view of plug-and-play BCI applications.

An Unsupervised Deep-Transfer-Learning-Based Motor Imagery EEG Classification Scheme for Brain-Computer Interface

Brain–computer interface (BCI) research has attracted worldwide attention and has been rapidly developed. As one well-known non-invasive BCI technique, electroencephalography (EEG) records the brain’s electrical signals from the scalp surface area. However, due to the non-stationary nature of the EEG signal, the distribution of the data collected at different times or from different subjects may be different. These problems affect the performance of the BCI system and limit the scope of its practical application. In this study, an unsupervised deep-transfer-learning-based method was proposed to deal with the current limitations of BCI systems by applying the idea of transfer learning to the classification of motor imagery EEG signals. The Euclidean space data alignment (EA) approach was adopted to align the covariance matrix of source and target domain EEG data in Euclidean space. Then, the common spatial pattern (CSP) was used to extract features from the aligned data matrix, and the deep convolutional neural network (CNN) was applied for EEG classification. The effectiveness of the proposed method has been verified through the experiment results based on public EEG datasets by comparing with the other four methods.

Explainable Artificial Intelligence and Wearable Sensor-Based Gait Analysis to Identify Patients with Osteopenia and Sarcopenia in Daily Life

Osteopenia and sarcopenia can cause various senile diseases and are key factors related to the quality of life in old age. There is need for portable tools and methods that can analyze osteopenia and sarcopenia risks during daily life, rather than requiring a specialized hospital setting. Gait is a suitable indicator of musculoskeletal diseases; therefore, we analyzed the gait signal obtained from an inertial-sensor-based wearable gait device as a tool to manage bone loss and muscle loss in daily life. To analyze the inertial-sensor-based gait, the inertial signal was classified into seven gait phases, and descriptive statistical parameters were obtained for each gait phase. Subsequently, explainable artificial intelligence was utilized to analyze the contribution and importance of descriptive statistical parameters on osteopenia and sarcopenia. It was found that XGBoost yielded a high accuracy of 88.69% for osteopenia, whereas the random forest approach showed a high accuracy of 93.75% for sarcopenia. Transfer learning with a ResNet backbone exhibited appropriate performance but showed lower accuracy than the descriptive statistical parameter-based identification result. The proposed gait analysis method confirmed high classification accuracy and the statistical significance of gait factors that can be used for osteopenia and sarcopenia management.

Explainable health prediction from facial features with transfer learning

In the recent years, Artificial Intelligence (AI) has been widely deployed in the healthcare industry. The new AI technology enables efficient and personalized healthcare systems for the public. In this paper, transfer learning with pre-trained VGGFace model is applied to identify sick symptoms based on the facial features of a person. As the deep learning model’s operation is unknown for making a decision, this paper investigates the use of Explainable AI (XAI) techniques for soliciting explanations for the predictions made by the model. Various XAI techniques including Integrated Gradient, Explainable region-based AI (XRAI) and Local Interpretable Model-Agnostic Explanations (LIME) are studied. XAI is crucial to increase the model’s transparency and reliability for practical deployment. Experimental results demonstrate that the attribution method can give proper explanations for the decisions made by highlighting important attributes in the images. The facial features that account for positive and negative classes predictions are highlighted appropriately for effective visualization. XAI can help to increase accountability and trustworthiness of the healthcare system as it provides insights for understanding how a conclusion is derived from the AI model.

Investigating EEG-based functional connectivity patterns for multimodal emotion recognition

Objective. Previous studies on emotion recognition from electroencephalography (EEG) mainly rely on single-channel-based feature extraction methods, which ignore the functional connectivity between brain regions. Hence, in this paper, we propose a novel emotion-relevant critical subnetwork selection algorithm and investigate three EEG functional connectivity network features: strength, clustering coefficient, and eigenvector centrality. Approach. After constructing the brain networks by the correlations between pairs of EEG signals, we calculated critical subnetworks through the average of brain network matrices with the same emotion label to eliminate the weak associations. Then, three network features were conveyed to a multimodal emotion recognition model using deep canonical correlation analysis along with eye movement features. The discrimination ability of the EEG connectivity features in emotion recognition is evaluated on three public datasets: SEED, SEED-V, and DEAP. Main results. The experimental results reveal that the strength feature outperforms the state-of-the-art features based on single-channel analysis. The classification accuracies of multimodal emotion recognition are 95.08 ± 6.42 % on the SEED dataset, 84.51 ± 5.11 % on the SEED-V dataset, and 85.34 ± 2.90 % and 86.61 ± 3.76 % for arousal and valence on the DEAP dataset, respectively, which all achieved the best performance. In addition, the brain networks constructed with 18 channels achieve comparable performance with that of the 62-channel network and enable easier setups in real scenarios. Significance. The EEG functional connectivity networks combined with emotion-relevant critical subnetworks selection algorithm we proposed is a successful exploration to excavate the information between channels.

Tree based Ensemble for Enhanced Prediction (TEEP) of epileptic seizures

Accurate and timely prediction of seizures can improve the quality of life of epileptic patients to a huge extent. This work presents a seizure prediction model that performs data extraction and feature engineering to enable effective demarcation of preictal signals from interictal signals. The proposed Tree based Ensemble for Enhanced Prediction (TEEP) model is composed of three major phases; the feature extraction phase, feature selection phase and the prediction phase. The data is preprocessed, and features are extracted based on the nature of the data. This enables the prediction algorithm to perform time-based predictions. Further, statistical features are also extracted, followed by the process of feature aggregation. The resultant data is passed to the feature selection module to identify the attributes that exhibit highest correlation with the prediction variable. Incorporation of these two modules enhances the generalization capability of the TEEP model. The resultant features are passed to the boosted ensemble model for training and prediction. The TEEP model is analyzed using the Epileptic Seizure Recognition Data from University Hospital of Bonn and the NIH Seizure Prediction data from Melbourne University, Australia. Results from both the datasets indicate effective performances. Comparisons with the existing state-of-the-art models in literature exhibits the enhanced prediction levels of the TEEP model.

Simultaneous Classification of Both Mental Workload and Stress Level Suitable for an Online Passive Brain-Computer Interface

Research studies on EEG-based mental workload detection for a passive BCI generally focus on classifying cognitive states associated with the performance of tasks at different levels of difficulty, with no other aspects of the user's mental state considered. However, in real-life situations, different aspects of the user's state such as their cognitive (e.g., level of mental workload) and affective (e.g., level of stress/anxiety) states will often change simultaneously, and performance of a BCI system designed considering just one state may be unreliable. Moreover, multiple mental states may be relevant to the purposes of the BCI-for example both mental workload and stress level might be related to an aircraft pilot's risk of error-and the simultaneous prediction of states may be critical in maximizing the practical effectiveness of real-life online BCI systems. In this study we investigated the feasibility of performing simultaneous classification of mental workload and stress level in an online passive BCI. We investigated both subject-specific and cross-subject classification approaches, the latter with and without the application of a transfer learning technique to align the distributions of data from the training and test subjects. Using cross-subject classification with transfer learning in a simulated online analysis, we obtained accuracies of 77.5 ± 6.9% and 84.1 ± 5.9%, across 18 participants for mental workload and stress level detection, respectively.

Benchmarking cEEGrid and Solid Gel-Based Electrodes to Classify Inattentional Deafness in a Flight Simulator

Transfer from experiments in the laboratory to real-life tasks is challenging due notably to the inability to reproduce the complexity of multitasking dynamic everyday life situations in a standardized lab condition and to the bulkiness and invasiveness of recording systems preventing participants from moving freely and disturbing the environment. In this study, we used a motion flight simulator to induce inattentional deafness to auditory alarms, a cognitive difficulty arising in complex environments. In addition, we assessed the possibility of two low-density EEG systems a solid gel-based electrode Enobio (Neuroelectrics, Barcelona, Spain) and a gel-based cEEGrid (TMSi, Oldenzaal, Netherlands) to record and classify brain activity associated with inattentional deafness (misses vs. hits to odd sounds) with a small pool of expert participants. In addition to inducing inattentional deafness (missing auditory alarms) at much higher rates than with usual lab tasks (34.7% compared to the usual 5%), we observed typical inattentional deafness-related activity in the time domain but also in the frequency and time-frequency domains with both systems. Finally, a classifier based on Riemannian Geometry principles allowed us to obtain more than 70% of single-trial classification accuracy for both mobile EEG, and up to 71.5% for the cEEGrid (TMSi, Oldenzaal, Netherlands). These results open promising avenues toward detecting cognitive failures in real-life situations, such as real flight.

Cross-task Cognitive Workload Recognition Based on EEG and Domain Adaptation

Cognitive workload recognition is pivotal to maintain the operator's health and prevent accidents in the human-robot interaction condition. So far, the focus of workload research is mostly restricted to a single task, yet cross-task cognitive workload recognition has remained a challenge. Furthermore, when extending to a new workload condition, the discrepancy of electroencephalogram (EEG) signals across various cognitive tasks limits the generalization of the existed model. To tackle this problem, we propose to construct the EEG-based cross-task cognitive workload recognition models using domain adaptation methods in a leave-one-task-out cross-validation setting, where we view any task of each subject as a domain. Specifically, we first design a fine-grained workload paradigm including working memory and mathematic addition tasks. Then, we explore four domain adaptation methods to bridge the discrepancy between the two different tasks. Finally, based on the supporting vector machine classifier, we conduct experiments to classify the low and high workload levels on a private EEG dataset. Experimental results demonstrate that our proposed task transfer framework outperforms the non-transfer classifier with improvements of 3% to 8% in terms of mean accuracy, and the transfer joint matching (TJM) consistently achieves the best performance.

A 1D CNN for high accuracy classification and transfer learning in motor imagery EEG-based brain-computer interface

OBJECTIVE

Brain-computer interface (BCI) aims to establish communication paths between the brain processes and external devices. Different methods have been used to extract human intentions from electroencephalography (EEG) recordings. Those based on motor imagery (MI) seem to have a great potential for future applications. These approaches rely on the extraction of EEG distinctive patterns during imagined movements. Techniques able to extract patterns from raw signals represent an important target for BCI as they do not need labor-intensive data pre-processing.

APPROACH

We propose a new approach based on a 10-layer one-dimensional convolution neural network (1D-CNN) to classify five brain states (four MI classes plus a 'baseline' class) using a data augmentation algorithm and a limited number of EEG channels. In addition, we present a transfer learning method used to extract critical features from the EEG group dataset and then to customize the model to the single individual by training its outer layers with only 12-minute individual-related data.

MAIN RESULTS

The model tested with the 'EEG Motor Movement/Imagery Dataset' outperforms the current state-of-the-art models by achieving a 99.38% accuracy at the group level. In addition, the transfer learning approach we present achieves an average accuracy of 99.46%.

SIGNIFICANCE

The proposed methods could foster future BCI applications relying on few-channel portable recording devices and individual-based training.

Combining Statistical Analysis and Machine Learning for EEG Scalp Topograms Classification

Incorporating brain-computer interfaces (BCIs) into daily life requires reducing the reliance of decoding algorithms on the calibration or enabling calibration with the minimal burden on the user. A potential solution could be a pre-trained decoder demonstrating a reasonable accuracy on the naive operators. Addressing this issue, we considered ambiguous stimuli classification tasks and trained an artificial neural network to classify brain responses to the stimuli of low and high ambiguity. We built a pre-trained classifier utilizing time-frequency features corresponding to the fundamental neurophysiological processes shared between subjects. To extract these features, we statistically contrasted electroencephalographic (EEG) spectral power between the classes in the representative group of subjects. As a result, the pre-trained classifier achieved 74% accuracy on the data of newly recruited subjects. Analysis of the literature suggested that a pre-trained classifier could help naive users to start using BCI bypassing training and further increased accuracy during the feedback session. Thus, our results contribute to using BCI during paralysis or limb amputation when there is no explicit user-generated kinematic output to properly train a decoder. In machine learning, our approach may facilitate the development of transfer learning (TL) methods for addressing the cross-subject problem. It allows extracting the interpretable feature subspace from the source data (the representative group of subjects) related to the target data (a naive user), preventing the negative transfer in the cross-subject tasks.

Rolling Bearing Incipient Fault Diagnosis Method Based on Improved Transfer Learning with Hybrid Feature Extraction

Data-driven based rolling bearing fault diagnosis has been widely investigated in recent years. However, in real-world industry scenarios, the collected labeled samples are normally in a different data distribution. Moreover, the features of bearing fault in the early stages are extremely inconspicuous. Due to the above mentioned problems, it is difficult to diagnose the incipient fault under different scenarios by adopting the conventional data-driven methods. Therefore, in this paper a new unsupervised rolling bearing incipient fault diagnosis approach based on transfer learning is proposed, with a novel feature extraction method based on a statistical algorithm, wavelet scattering network, and a stacked auto-encoder network. Then, the geodesic flow kernel algorithm is adopted to align the feature vectors on the Grassmann manifold, and the k-nearest neighbor classifier is used for fault classification. The experiment is conducted based on two bearing datasets, the bearing fault dataset of Case Western Reserve University and the bearing fault dataset of Xi’an Jiaotong University. The experiment results illustrate the effectiveness of the proposed approach on solving the different data distribution and incipient bearing fault diagnosis issues.