Back

Deep learning-based electroencephalography analysis: a systematic review

Yannick RoyHubert BanvilleIsabela AlbuquerqueAlexandre GramfortTiago H. FalkJocelyn Faubert

article2019Journal of Neural Engineering1,430 citations

Synthesizes 156 studies applying deep learning to electroencephalography, quantifying a median 5.4% accuracy improvement over traditional baselines while exposing critical reproducibility deficits and providing practical guidelines for future research.

Listen

Electroencephalography (EEG), which records electrical brain activity via scalp sensors, is critical for diagnosing neurological conditions such as epilepsy and sleep disorders, as well as for operating brain–computer interfaces. However, traditional EEG processing relies heavily on manual artifact removal and specialized, hand-crafted feature extraction. These conventional methods are labor-intensive, struggle to generalize across different individuals due to high subject variability, and often fail to scale efficiently. Advanced machine learning techniques, particularly deep learning, offer a promising alternative by learning representations directly from data, but their true performance advantage and best practices have remained unclear.

The article provides a comprehensive evaluation of deep learning applied to non-invasive EEG analysis. It aims to determine whether deep learning architectures outperform traditional processing approaches, identify methodological trends across application domains, and establish standards for experimental reproducibility.

To conduct this evaluation, the authors performed a systematic review of 154 studies published between January 2010 and July 2018 across scientific journals, conference proceedings, and electronic preprint repositories. The analysis covered five primary application areas: sleep staging, seizure detection, brain–computer interfaces, cognitive and affective monitoring, and processing tool improvements. For each study, the review extracted and analyzed roughly 70 distinct parameters spanning data characteristics, preprocessing pipelines, model architectures, training procedures, comparative performance metrics, and reproducibility factors.

The review revealed several key findings across the literature. First, deep learning models demonstrated a modest but consistent performance advantage, achieving a median accuracy gain of 5.4% over traditional baseline methods across evaluated tasks. Second, convolutional neural networks emerged as the dominant architecture, used in 40% of studies, followed by recurrent neural networks and autoencoders at 13% each. Most successful models were relatively shallow, using between 3 and 10 layers, which contrasts with the much deeper networks common in computer vision. Third, end-to-end learning proved viable: 49% of studies successfully trained models on raw or minimally filtered EEG time series rather than hand-engineered frequency features, and 47% bypassed explicit artifact removal without sacrificing performance. Finally, severe reproducibility deficiencies were identified across the field. While 53% of studies used publicly available data, only 13% shared their source code, leaving just 8% of the reviewed studies fully reproducible.

These findings indicate that deep learning can reduce the cost, timeline, and domain expertise required to build EEG pipelines by automating feature extraction and artifact handling. However, the modest 5.4% median performance improvement suggests that deep learning is not yet a complete replacement for established methods. The lack of standardized benchmarks and common baselines—such as state-of-the-art Riemannian geometry classifiers—means reported gains may be inflated by comparisons against weak baselines or compromised by publication bias toward positive results. Furthermore, because deep models can inadvertently learn non-brain artifacts (like muscle or eye movements) when trained on raw data, deploying uninspected models in clinical or critical settings introduces significant operational and compliance risks.

Organizations and researchers pursuing deep learning for EEG should prioritize rigorous, reproducible development workflows. Practitioners should adopt standardized reporting checklists, evaluate models on public benchmark datasets, and consistently compare new architectures against strong, open-source baselines rather than simplistic models. Where subject data is limited, teams should implement data augmentation techniques (such as overlapping windows) and explore hybrid transfer learning—pre-training models across large multi-subject pools before fine-tuning them on specific users. Before making substantial deployment decisions in clinical or high-stakes environments, organizations should conduct targeted pilot studies with transparent model inspection techniques to ensure decisions are driven by genuine neural signals rather than noise.

Confidence in the overarching architectural trends is high, but comparative performance conclusions must be interpreted with caution. The analyzed literature exhibits considerable heterogeneity in dataset sizes (ranging from under 10 minutes to thousands of hours), subject numbers (a median of only 13 subjects), validation schemes, and baseline selections. Readers should account for these data gaps and the prevailing lack of code transparency when evaluating reported performance gains.

INRIAInstitut national de la recherche scientifiqueInteraXon Inc.Université de Montréalbrain-computer interfacingconvolutional neural networksdeep learningdeep learning-based electroencephalography analysisEEG classificationEEG decoding
arXiv: 1901.05498
Cover for Deep learning-based electroencephalography analysis: a systematic review

Abstract

Electroencephalography (EEG) is a complex signal and can require several years of training to be correctly interpreted. Recently, deep learning (DL) has shown great promise in helping make sense of EEG signals due to its capacity to learn good feature representations from raw data. Whether DL truly presents advantages as compared to more traditional EEG processing approaches, however, remains an open question. In this work, we review 156 papers that apply DL to EEG, published between January 2010 and July 2018, and spanning different application domains such as epilepsy, sleep, brain-computer interfacing, and cognitive and affective monitoring. We extract trends and highlight interesting approaches in order to inform future research and formulate recommendations. Various data items were extracted for each study pertaining to 1) the data, 2) the preprocessing methodology, 3) the DL design choices, 4) the results, and 5) the reproducibility of the experiments. Our analysis reveals that the amount of EEG data used across studies varies from less than ten minutes to thousands of hours. As for the model, 40% of the studies used convolutional neural networks (CNNs), while 14% used recurrent neural networks (RNNs), most often with a total of 3 to 10 layers. Moreover, almost one-half of the studies trained their models on raw or preprocessed EEG time series. Finally, the median gain in accuracy of DL approaches over traditional baselines was 5.4% across all relevant studies. More importantly, however, we noticed studies often suffer from poor reproducibility: a majority of papers would be hard or impossible to reproduce given the unavailability of their data and code. To help the field progress, we provide a list of recommendations for future studies and we make our summary table of DL and EEG papers available and invite the community to contribute.

Topics

brain-computer interfacingconvolutional neural networksdeep learningelectroencephalography (eeg)recommendationrecurrent neural networksinterpretereeg recording

Keywords

deep learning-based electroencephalography analysisEEG classificationEEG decodingimagery BCIsingle-trial EEGEEG time seriesnon-invasive EEG

Table of Contents

  • 1 Introduction
  • 1.1 Measuring brain activity with EEG
  • 1.2 Current challenges in EEG processing
  • 1.3 Improving EEG processing with deep learning
  • 1.4 Terminology used in this review
  • 1.5 Objectives of the review
  • 1.6 Organization of the review
  • 2 Methods
  • 3 Results
  • 3.1 Origin of the selected studies
  • 3.2 Domains
  • 3.3 Data
  • 3.3.1 Quantity of data
  • 3.3.2 Subjects
  • 3.3.3 Recording parameters
  • 3.3.4 Data augmentation
  • 3.4 EEG processing
  • 3.4.1 Preprocessing
  • 3.4.2 Artifact handling
  • 3.4.3 Features
  • 3.5 Deep learning methodology
  • 3.5.1 Architecture
  • 3.5.2 Training
  • 3.6 Inspection of trained models
  • 3.7 Reporting of results
  • 3.7.1 Type of baseline
  • 3.7.2 Performance metrics
  • 3.7.3 Validation procedure
  • 3.7.4 Subject handling
  • 3.7.5 Statistical testing
  • 3.7.6 Comparison of results
  • 3.8 Reproducibility
  • 4 Discussion
  • 4.1 Rationale
  • 4.2 Data
  • 4.3 EEG processing
  • 4.4 Deep learning methodology
  • 4.4.1 Architecture
  • 4.4.2 Training and optimization
  • 4.4.3 Model inspection
  • 4.5 Reported results
  • 4.6 Reproducibility
  • 4.7 Recommendations
  • 4.7.1 Supplementary material
  • 4.8 Limitations
  • 5 Conclusion
  • References
  • A List of acronyms
  • B Checklist of items to include in a DL-EEG study

Knowls

  1. Knowl 1 — Empirical Performance Comparison of Deep Learning vs. Traditional Baselines in EEG Processing

    empirical result

    Analysis of 102 classification tasks across literature comparing proposed deep learning (DL) models to traditional baselines (pipelines utilizing hand-crafted feature extraction and classical classifiers such as Support Vector Machines, Naive Bayes, Linear Discriminant Analysis, or Random Forests) revealed that DL approaches achieved a median accuracy improvement of 5.4%5.4\% (interquartile range: 9.4%9.4\%). Out of 102 evaluated tasks, only four reported negative accuracy differences where DL underperformed traditional methods.

    Across peer-reviewed studies, the median accuracy improvement over baselines was 6.00%6.00\%, compared to 4.7%4.7\% in preprints (Mann–Whitney test p=0.072p = 0.072). A Kruskal–Wallis test revealed that the choice of neural network architecture significantly influenced the reported accuracy gain over traditional baselines (p=0.043p = 0.043), with Convolutional Neural Networks (CNNs) and Deep Belief Networks (DBNs) generally yielding higher accuracy gains compared to Autoencoder (AE)-based architectures.

    Only 19.5%19.5\% of reviewed studies performed formal statistical hypothesis testing (most frequently Wilcoxon signed-rank tests, followed by Analysis of Variance (ANOVA)) to assess whether DL performance improvements over baseline models were statistically significant.

  2. Knowl 2 — Distribution and Temporal Evolution of Deep Learning Architectures for EEG

    empirical result

    Across a corpus of 154 studies published between January 2010 and July 2018 applying deep learning to electroencephalography (EEG), the overall distribution of primary neural network architectures is:

    • Convolutional Neural Networks (CNNs): 40.3%40.3\% (6262 studies)
    • Recurrent Neural Networks (RNNs): 13.0%13.0\% (2020 studies)
    • Autoencoders (AEs): 13.0%13.0\% (2020 studies)
    • Deep Belief Networks (DBNs): 7.1%7.1\% (1111 studies)
    • Hybrid CNN + RNN models: 7.1%7.1\% (1111 studies)
    • Fully Connected (FC) networks: 6.5%6.5\% (1010 studies)
    • Restricted Boltzmann Machines (RBMs): 3.9%3.9\% (66 studies)
    • Generative Adversarial Networks (GANs): 2.6%2.6\% (44 studies)
    • Other architectures: 1.9%1.9\% (33 studies)
    • Architecture not specified: 4.5%4.5\% (77 studies)

    Prior to 2015, DBNs and FC networks constituted the majority of published DL-EEG architectures. From 2015 onward, CNNs became the dominant architecture due to their capacity for end-to-end spatial-temporal feature learning directly from raw signals. Concurrently, hybrid architectures combining convolutional and recurrent layers grew steadily to model spatial hierarchies alongside sequential temporal dynamics.

  3. Knowl 3 — Input Representation, Preprocessing, and Artifact Removal Practices in DL-EEG

    empirical result

    In deep learning applications to electroencephalography (EEG):

    • Input Feature Types: 49%49\% of studies trained models directly on raw or preprocessed EEG time series (most commonly using CNNs), 49%49\% utilized hand-engineered feature representations (of which 38%38\% were frequency-domain representations such as Short-Time Fourier Transform (STFT) or Power Spectral Density (PSD) extracted across delta (1–4 Hz1\text{--}4\text{ Hz}), theta (5–8 Hz5\text{--}8\text{ Hz}), alpha (9–13 Hz9\text{--}13\text{ Hz}), beta (14–30 Hz14\text{--}30\text{ Hz}), and gamma (31–40 Hz31\text{--}40\text{ Hz}) bands), and 2%2\% did not report the input representation.
    • Preprocessing: 72%72\% of studies applied at least one standard preprocessing step (e.g., bandpass filtering, downsampling to ≤250 Hz\le 250\text{ Hz}, average re-referencing, channel interpolation), 15%15\% explicitly applied no preprocessing, and 13%13\% did not mention preprocessing.
    • Artifact Handling: 47%47\% of studies applied no explicit artifact handling (such as Independent Component Analysis (ICA) or amplitude thresholding), leaving noise robustness to feature learning inside the network, 23%23\% applied explicit artifact removal or rejection, and 30%30\% did not state whether artifact removal was performed.

  4. Knowl 4 — Network Depth Distribution in Deep Learning Models for EEG

    empirical result

    Deep neural networks applied to EEG data operate with substantially shallower architectures compared to typical computer vision or natural language processing architectures. Across 154 reviewed studies:

    • 128128 studies utilized neural architectures with 1010 or fewer layers, with the majority containing fewer than 55 layers.
    • Only a small fraction implemented networks with ≥11\ge 11 layers, while 1616 studies did not report the layer count.

    Empirical evaluations across the reviewed literature indicate that shallow convolutional architectures frequently outperform deeper residual or fully convolutional architectures on non-invasive EEG tasks, as deeper networks suffer from overfitting given typical EEG sample sizes and low signal-to-noise ratio, unless specialized weight initialization and extensive hyperparameter tuning are employed.

  5. Knowl 5 — Dataset Characteristics, Data Quantity, and Data Augmentation in DL-EEG

    empirical result

    Across 154 analyzed DL-EEG studies, dataset size and sample generation parameters varied widely:

    • Subject Count: Mean of 223223 subjects, median of 1313 subjects (50%50\% of datasets contained ≤13\le 13 subjects; range: 11 to 16,00016,000).
    • Recording Duration: Mean of 62,602 minutes62,602\text{ minutes}, median of 360 minutes360\text{ minutes} (range: 22 to 4,800,000 minutes4,800,000\text{ minutes}).
    • Number of Examples: Mean of 251,532251,532 examples, median of 14,00014,000 examples (range: 6262 to 9,750,0009,750,000).
    • Electrode Channels: 50%50\% of studies used between 88 and 6262 channels (range: 11 to 256256).
    • Subject Handling Shift: 62%62\% of studies evaluated inter-subject models, 26%26\% intra-subject models, 8%8\% both, and 4%4\% unmentioned, reflecting a temporal shift from intra-subject to inter-subject generalization.
    • Data Augmentation: 3333 studies explicitly applied data augmentation, utilizing sliding overlapping windows (varying from small shifts up to 95%95\% overlap), additive Gaussian noise on raw signals or feature maps, Fourier transform (FT) phase surrogates for under-represented classes, conditional generative adversarial networks (cDCGANs), left-right hemisphere electrode swapping for symmetric tasks, and reusing temporal points discarded during downsampling by factor NN to yield an NN-fold sample increase.

  6. Knowl 6 — State of Reproducibility in Deep Learning for EEG Research

    empirical result

    Evaluation of reproducibility across 154 DL-EEG publications identified substantial barriers to scientific replication:

    • Data Accessibility: 53%53\% of studies used publicly available benchmark datasets (e.g., DEAP, SEED, BCI Competition, Bonn University, CHB-MIT, TUH EEG, MASS, Sleep-EDF), 42%42\% relied exclusively on private/proprietary datasets, and 4%4\% used a combination of public and private data.
    • Code Accessibility: Only 13%13\% (2020 papers) made their source code publicly available (predominantly hosted on GitHub).
    • Reproducibility Level: Only 8%8\% of studies (1212 papers) were categorized as easily reproducible (both dataset and code publicly accessible); 3%3\% were partially reproducible (open code, but partially private data); 50%50\% were hard to reproduce (public data available, but no source code); and 40%40\% (6161 papers) were impossible to reproduce (neither data nor code publicly accessible).

  7. Knowl 7 — Model Training, Optimization, and Hyperparameter Search Deficits in DL-EEG

    empirical result

    Methodological analysis of training protocols in DL-EEG literature reveals notable reporting deficits:

    • Training Regimes: 45%45\% of studies trained models using standard end-to-end supervised optimization, 25%25\% used a two-stage procedure (unsupervised pretraining via Autoencoders, Restricted Boltzmann Machines, or Deep Belief Networks followed by supervised fine-tuning), 4%4\% used other regimes (such as co-learning), and 25%25\% did not report the training procedure.
    • Optimizers: 47%47\% of studies omitted the optimizer used. Among those reporting optimizers, 30%30\% utilized Adam (increasing from 28.9%28.9\% in 2017 to 54.2%54.2\% in 2018), 17%17\% used Stochastic Gradient Descent (SGD / mini-batch SGD), and 6%6\% used other adaptive optimizers (RMSprop, Adagrad, Adadelta).
    • Regularization: 51%51\% of studies explicitly reported using regularization methods, commonly combining dropout, L1L_1/L2L_2 weight penalties, early stopping, or sparsity constraints.
    • Hyperparameter Optimization: 80%80\% of studies did not report using any hyperparameter search strategy. Among the 20%20\% that reported tuning, strategies were split between manual trial-and-error, grid search, and Bayesian optimization.

  8. Knowl 8 — Model Inspection and Interpretability Methodologies for EEG Deep Networks

    model/method

    While deep neural networks are often treated as black boxes, 27%27\% of the 154 reviewed DL-EEG studies implemented model inspection techniques to interpret learned representations and ensure decisions were not driven by non-neural artifacts. These methods comprise:

    • First-Layer Weight Inspection: Directly visualizing convolutional filter weights in the initial network layer to interpret spatial channel distributions and temporal frequency responses.
    • Activation Analysis: Monitoring hidden unit activations across input classes or stimulus conditions to identify feature selectivity.
    • Input-Perturbation Network-Prediction Correlation Maps: Applying systematic perturbations to input signals in the time- or frequency-domain (e.g., amplitude scaling or phase scrambling) and computing correlation maps between perturbations and output unit activations to establish causal input dependencies.
    • Occlusion Sensitivity: Systematically masking subsets of channels or time intervals to quantify degradation in prediction confidence.
    • Unit Activation Maximization: Using gradient ascent via backpropagation to synthesize artificial EEG input signals that maximize the activation of specific hidden neurons or output units.
    • Class Activation Mapping (CAM) and Saliency Maps: Computing spatially weighted combinations of feature maps from the final convolutional layer to generate scalp topographies of discriminative regions.

  9. Knowl 9 — Disambiguation of Overlapping Machine Learning and EEG Terminology

    definition

    To prevent ambiguity caused by discordant terminology across machine learning, statistics, and electrophysiology, terms are defined as follows:

    • Point or Sample: A single instantaneous scalar value of electrical potential measured by an EEG sensor at a discrete time step, typically expressed in microvolts (μV\mu\text{V}).
    • Example: A single input instance supplied to a machine learning model, denoted as Xi∈Rc×l\mathbf{X}_i \in \mathbb{R}^{c \times l} (where cc is the number of channels and ll is the number of time points in the window) or unrolled into a vector xi∈Rn\mathbf{x}_i \in \mathbb{R}^n where n=c×ln = c \times l.
    • Trial: A single continuous realization of an experimental task condition or stimulus event (e.g., the presentation of one stimulus image in an event-related potential experiment).
    • Window or Segment: A discrete temporal sequence of consecutive EEG samples extracted for analysis, typically lasting between 0.5 s0.5\text{ s} and 30 s30\text{ s}.
    • Epoch: In electrophysiology and EEG analysis, a time window of consecutive EEG data points extracted relative to a specific experimental marker or trigger event. In machine learning and deep learning, an epoch denotes one complete forward and backward training pass through the entire training dataset.

  10. Knowl 10 — Systematic Review Search Protocol and Paper Corpus Selection for DL-EEG

    experimental setup

    The systematic review searched PubMed, Google Scholar, and arXiv up to July 2, 2018, targeting English-language journal articles, conference papers, and electronic preprints published between January 2010 and July 2018. The database search string combined EEG identifiers with deep learning terminology:

    (EEG∨electroencephalogra*)∧(deep learning∨representation learning∨neural network*∨convolutional neural network*∨ConvNet∨CNN∨recurrent neural network*∨RNN∨long short-term memory∨LSTM∨generative adversarial network*∨GAN∨autoencoder∨restricted boltzmann machine*∨deep belief network*∨DBN)(\text{EEG} \lor \text{electroencephalogra*}) \land (\text{deep learning} \lor \text{representation learning} \lor \text{neural network*} \lor \text{convolutional neural network*} \lor \text{ConvNet} \lor \text{CNN} \lor \text{recurrent neural network*} \lor \text{RNN} \lor \text{long short-term memory} \lor \text{LSTM} \lor \text{generative adversarial network*} \lor \text{GAN} \lor \text{autoencoder} \lor \text{restricted boltzmann machine*} \lor \text{deep belief network*} \lor \text{DBN})

    The query yielded 553553 initial matches, plus 4949 papers found by reference tracking. Exclusion criteria eliminated studies focusing solely on invasive electrophysiology (electrocorticography (ECoG), intracortical EEG), magnetoencephalography (MEG), pure software toolboxes, and review papers. After excluding 448448 mismatched papers and 11 retracted article, the final analyzed corpus contained 154154 papers (5151 journal articles, 6161 conference papers, and 4242 preprints).

  11. Knowl 11 — Methodological and Reproducibility Guidelines for DL-EEG Studies

    model/method

    To improve scientific rigor and reproducibility across deep learning applications in electroencephalography, six core recommendations are established:

    1. Clearly Describe the Architecture: Provide a complete architectural diagram or tabular specification defining layer types, sequence, layer dimensions, kernel sizes, strides, activations, and parameter counts.
    2. Clearly Describe the Data: Specify exact subject counts and demographics, electrode montage and reference configurations, exact input array dimensions (channels ×\times samples), windowing and data augmentation schemes (including window duration and overlap step), and sample counts across training, validation, and test splits.
    3. Benchmark on Existing Public Datasets: Compare proposed models against established benchmarks on openly accessible public EEG datasets.
    4. Include State-of-the-Art Baselines: Benchmark proposed DL models against domain-standard traditional machine learning pipelines (such as Riemannian geometry classifiers or spectral feature extractors) under identical validation schemes.
    5. Share Internal Recordings: Make private experimental datasets publicly available whenever possible in standard formats (e.g., BIDS-EEG).
    6. Share Reproducible Code and Parameters: Release open-source code repositories containing preprocessing routines, exact hyperparameter configurations, random seeds, and trained model weights.

Coverage note — Specific manufacturer market shares of EEG acquisition hardware (e.g., Emotiv vs. BioSemi device counts) and external URLs for individual repositories were omitted in favor of comprehensive meta-analytic methodology, architecture, and reproducibility findings.

References

  1. 1.Aboalayon K A I, Faezipour M, Almuhammadi W S and Moslehpour S 2016 Sleep stage classification using EEG signal analysis: a comprehensive survey and new investigation Entropy 18 272
  2. 2.Acharya U R, Oh S L, Hagiwara Y, Tan J H and Adeli H 2017 Deep convolutional neural network for the automated detection and diagnosis of seizure using EEG signals Computers in Biology and Medicine pp 1–9
  3. 3.Acharya U R, Sree S V, Swapna G, Martis R J and Suri J S 2013 Automated EEG analysis of epilepsy: a review Knowl.-Based Syst. 45 147–65
  4. 4.Ahmedt-Aristizabal D, Fookes C, Nguyen K and Sridharan S 2018 Deep classification of epileptic signals (arXiv:1801.03610)
  5. 5.Al-Nafjan A, Hosny M, Al-Ohali Y and Al-Wabil A 2017 Review and classification of emotion recognition based on EEG brain–computer interface system research: a systematic review Appl. Sci. 7 1239
  6. 6.Alhagry S, Fahmy A A and El-Khoribi R A 2017 Emotion recognition based on EEG using LSTM recurrent neural network Int. J. Adv. Comput. Sci. Appl. 8 8–11
  7. 7.Almogbel M A, Dang A H and Kameyama W 2018 EEG-signals based cognitive workload detection of vehicle driver using deep learning 20th Int. Conf. on Advanced Communication Technology vol 7 pp 256–9
  8. 8.An J and Cho S 2016 Hand motion identification of grasp-and-lift task from electroencephalography recordings using recurrent neural networks Int. Conf. on Big Data and Smart Computing, BigComp 2016 pp 427–9
  9. 9.An X, Kuang D, Guo X, Zhao Y and He L 2014 A deep learning method for classification of EEG data based on motor imagery Lecture Notes Comput. Sci. 8590 LNBI 203–10 (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  10. 10.Andrzejak R G, Lehnertz K, Mormann F, Rieke C, David P and Elger C E 2001 Indications of nonlinear deterministic and finite-dimensional structures in time series of brain electrical activity: dependence on recording region and brain state Phys. Rev. E 64 061907
  11. 11.Arns M, Conners C K and Kraemer H C 2013 A decade of EEG theta/beta ratio research in ADHD: a meta-analysis J. Attention Disorders 17 374–83
  12. 12.Aznan N K N, Bonner S, Connolly J D, Moubayed N A and Breckon T P 2018 On the classification of SSVEP-based dry-EEG signals via convolutional neural networks (arXiv:1805.04157)
  13. 13.Bai S, Kolter J Z and Koltun V 2018 An empirical evaluation of generic convolutional and recurrent networks for sequence modeling (arXiv:1803.01271)
  14. 14.Baltatzis V, Bintsi K M, Apostolidis G K and Hadjileontiadis L J 2017 Bullying incidences identification within an immersive environment using HD EEG-based analysis: a swarm decomposition and deep learning approach Sci. Rep. 7 17292
  15. 15.Bashivan P, Rish I and Heisig S 2016 Mental state recognition via wearable EEG (arXiv:1602.00985)
  16. 16.Bashivan P, Rish I, Yeasin M and Codella N 2015 Learning representations from EEG with deep recurrent-convolutional neural networks (arXiv:1511.06448)
  17. 17.Behncke J, Schirrmeister R T, Burgard W and Ball T 2017 The signature of robot action success in EEG signals of a human observer: decoding and visualization using deep convolutional neural networks (arXiv:1711.06068)
  18. 18.Ben-David S, Blitzer J, Crammer K and Pereira F 2007 Analysis of representations for domain adaptation Advances in Neural Information Processing Systems pp 137–44
  19. 19.Ben Said A, Mohamed A, Elfouly T, Harras K and Wang Z J 2017 Multimodal deep learning approach for Joint EEG-EMG data compression and classification IEEE Wireless Communications and Networking Conf.
  20. 20.Bergstra J and Bengio Y 2012 Random search for hyper-parameter optimization J. Mach. Learn. Res. 13 281–305
  21. 21.Berka C, Levendowski D J, Lumicao M N, Yau A, Davis G, Zivkovic V T, Olmstead R E, Tremoulet P D and Craven P L 2007 {EEG} correlates of task engagement and mental workload in vigilance, learning, and memory tasks Aviat. Space Environ. Med. 78 B231–44
  22. 22.Biasiucci A, Franceschiello B and Murray M M 2019 Electroencephalography Curr. Biol. 29 R80–5
  23. 23.Bigdely-Shamlo N, Mullen T, Kothe C, Su K M and Robbins K A 2015 The PREP pipeline: standardized preprocessing for large-scale EEG analysis Frontiers Neuroinform. 9 16
  24. 24.Bishop C M 1995 Neural Networks for Pattern Recognition vol 92 (Oxford: Oxford University Press)
  25. 25.Biswal S, Kulas J, Sun H, Goparaju B, Westover M B, Bianchi M T and Sun J 2017 SLEEPNET: automated sleep staging system via deep learning 1–17 (arXiv:1707.08262)
  26. 26.Blankertz B et al 2004 The bci competition 2003 IEEE Trans. Biomed. Eng. 51 1044–51
  27. 27.Blankertz B, Muller K R, Krusienski D J, Schalk G, Wolpaw J R, Schlogl A, Pfurtscheller G, Millan J R, Schroder M and Birbaumer N 2006 The bci competition iii: validating alternative approaches to actual bci problems IEEE Trans. Neural Syst. Rehabil. Eng. 14 153–9
  28. 28.Brock A, Donahue J and Simonyan K 2018 Large scale gan training for high fidelity natural image synthesis (arXiv:1809.11096)
  29. 29.Bu N, Shima K and Tsuji T 2010 EEG discrimination using wavelet packet transform and a reduced-dimensional recurrent neural network Proc. of the 10th IEEE Int. Conf. on Information Technology and Applications in Biomedicine pp 1–4
  30. 30.Castellanos N P and Makarov V A 2006 Recovering EEG brain signals: artifact suppression with wavelet enhanced independent component analysis J. Neurosci. Methods 158 300–12
  31. 31.Cecotti H and Gräser A 2011 Convolutional neural networks for P300 detection with application to brain–computer interfaces IEEE Trans. Pattern Anal. Mach. Intell. 33 433–45
  32. 32.Cecotti H, Eckstein M P and Giesbrecht B 2014 Single-trial classification of event-related potentials in rapid serial visual presentation tasks using supervised spatial filtering IEEE Trans. Neural Netw. Learn. Syst. 25 2030–42
  33. 33.Chambon S, Galtier M N, Arnal P J, Wainrib G, Gramfort A, Paristech T and Nov M L 2017 A deep learning architecture for temporal sleep stage classification using multivariate and multimodal time series IEEE Trans. Neural Syst. Rehabil. Eng. 26 758–69
  34. 34.Chiarelli A M, Croce P, Merla A and Zappasodi F 2018 Deep learning for hybrid EEG-fNIRS brain–computer interface: application to motor imagery classification J. Neural Eng. 15 036028
  35. 35.Chu L, Qiu R, Liu H, Ling Z, Zhang T and Wang J 2017 Individual recognition in schizophrenia using deep learning methods with random forest and voting classifiers: insights from resting state EEG streams (arXiv:1707.03467)
  36. 36.Clerc M, Bougrain L and Lotte F 2016 Brain–Computer Interfaces 1: Foundations and Methods (New York: Wiley)
  37. 37.Cole S R and Voytek B 2018 Cycle-by-cycle analysis of neural oscillations (bioRxiv: 302000)
  38. 38.Comstock J R 1994 Mat-multi-attribute task battery for human operator workload and strategic behavior research
  39. 39.Congedo M, Barachant A and Bhatia R 2017 Riemannian geometry for EEG-based brain–computer interfaces; a primer and a review Brain–Comput. Interfaces 4 155–74
  40. 40.Corley I A and Huang Y 2018 Deep EEG super-resolution: upsampling EEG spatial resolution with generative adversarial networks IEEE EMBS Int. Conf. on Biomedical & Health Informatics pp 4–7
  41. 41.Deiss O, Biswal S, Jin J, Sun H, Westover M B and Sun J 2018 HAMLET: interpretable human and machine co-learning technique (arXiv:1803.09702)
  42. 42.Russakovsky O et al 2015 ImageNet large scale visual recognition challenge Int. J. Comput. Vis. 115 211–52
  43. 43.Dharamsi T, Das P, Pedapati T, Bramble G, Muthusamy V, Samulowitz H, Varshney K R, Rajamanickam Y, Thomas J and Dauwels J 2017 Neurology-as-a-service for the developing world (arXiv:1711.06195)
  44. 44.Ding S, Zhang N, Xu X, Guo L and Zhang J 2015 Deep extreme learning machine and its application in EEG classification Math. Problems Eng. 2015 129021
  45. 45.Dong H, Supratak A, Pan W, Wu C, Matthews P M and Guo Y 2018 Mixed neural network approach for temporal sleep stage classification IEEE Trans. Neural Syst. Rehabil. Eng. 26 324–33
  46. 46.Drouin-Picaro A and Falk T H 2016 Using deep neural networks for natural saccade classification from electroencephalograms Proc. IEEE EMBS Int. Student Conf.: Expanding the Boundaries of Biomedical Engineering and Healthcare (IEEE) pp 1–4
  47. 47.Duchi J, Hazan E and Singer Y 2011 Adaptive subgradient methods for online learning and stochastic optimization J. Mach. Learn. Res. 12 2121–59
  48. 48.Engemann D A et al 2018 Robust EEG-based cross-site and cross-protocol classification of states of consciousness Brain 141 3179–92
  49. 49.Frydenlund A and Rudzicz F 2015 Emotional affect estimation using video and EEG data in deep neural networks Lecture Notes Comput. Sci. 9091 273–80 (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  50. 50.Gao G, Shang L, Xiong K, Fang J, Zhang C and Gu X 2018 EEG classification based on sparse representation and deep learning NeuroQuantology 16 789–95
  51. 51.Gao Y, Lee H J and Mehmood R M 2015 Deep learninig of EEG signals for emotion recognition IEEE Int. Conf. on Multimedia and Expo Workshops (IEEE) pp 1–5
  52. 52.Ghassemi M M, Moody B E, Lehman L, Song C, Li Q, Sun H, Mark R G, Westover M B and Clifford G D 2018 You snooze, you win: the physionet/computing in cardiology challenge 2018 Comput. Cardiol. 45 (http://www.cinc.org/archives/2018/pdf/CinC2018-049.pdf)
  53. 53.Ghosh A, Dal Maso F, Roig M, Mitsis G D and Boudrias M H 2018 Deep semantic architecture with discriminative feature visualization for neuroimage analysis (arXiv:1805.11704)
  54. 54.Giacino J T, Fins J J, Laureys S and Schiff N D 2014 Disorders of consciousness after acquired brain injury: the state of the science Nat. Rev. Neurol. 10 99
  55. 55.Giri E P, Fanany M I and Arymurthy A M 2016 Combining generative and discriminative neural networks for sleep stages classification (arXiv:1610.01741)
  56. 56.Giri E P, Fanany M I and Arymurthy A M 2016 Ischemic stroke identification based on EEG and EOG using 1D convolutional neural network and batch normalization (arXiv:1610.01757)
  57. 57.Golmohammadi M, Torbati A H H N, de Diego S L, Obeid I and Picone J 2017 Automatic analysis of EEGs using big data and hybrid deep learning architectures (arXiv:1712.09771)
  58. 58.Golmohammadi M, Ziyabari S, Shah V, de Diego S L, Obeid I and Picone J 2017 Deep architectures for automated seizure detection in scalp EEGs (arXiv:1712.09776)
  59. 59.Golmohammadi M, Ziyabari S, Shah V, Von Weltin E, Campbell C, Obeid I and Picone J 2017 Gated recurrent networks for seizure detection IEEE Signal Processing in Medicine and Biology Symp. pp 1–5
  60. 60.Goodfellow I 2016 Nips 2016 tutorial: generative adversarial networks (arXiv:1701.00160)
  61. 61.Goodfellow I, Bengio Y and Courville A 2016 Deep Learning vol 1 (Cambridge, MA: MIT Press)
  62. 62.Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A and Bengio Y 2014 Generative adversarial nets Advances in Neural Information Processing Systems pp 2672–80
  63. 63.Gordienko Y, Stirenko S, Kochura Y, Alienin O, Novotarskiy M and Gordienko N 2017 Deep learning for fatigue estimation on the basis of multimodal human–machine interactions (arXiv:1801.06048)
  64. 64.Gorgolewski K J et al 2016 The brain imaging data structure, a format for organizing and describing outputs of neuroimaging experiments Sci. Data 3 160044
  65. 65.Gramfort A, Strohmeier D, Haueisen J, Hämäläinen M S and Kowalski M 2013 Time-frequency mixed-norm estimates: sparse M/EEG imaging with non-stationary source activations NeuroImage 70 410–22
  66. 66.Gulrajani I, Ahmed F, Arjovsky M, Dumoulin V and Courville A 2017 Improved training of wasserstein GANs Advances in Neural Information Processing Systems pp 5767–77
  67. 67.Alomari M H, Samaha A and AlKamha K 2013 Automated classification of L/R hand movement EEG signals using advanced feature extraction and machine learning Int. J. Adv. Comput. Sci. Appl. 4 6
  68. 68.Hagihira S 2015 Changes in the electroencephalogram during anaesthesia and their physiological basis Br. J. Anaesthesia 115 i27–31
  69. 69.Hajinoroozi M, Mao Z and Huang Y 2015 Prediction of driver's drowsy and alert states from EEG signals with deep learning IEEE 6th Int. Workshop on Computational Advances in Multi-Sensor Adaptive Processing pp 493–6
  70. 70.Hajinoroozi M, Mao Z and Lin Y P 2017 Deep transfer learning for cross-subject and cross-experiment prediction of image rapid serial visual presentation events from EEG data Int. Conf. on Augmented Cognition vol 10284 pp 45–55
  71. 71.Hajinoroozi M, Mao Z, Jung T P, Lin C T and Huang Y 2016 EEG-based prediction of driver's cognitive performance by deep convolutional neural network Signal Process.: Image Commun. 47 549–55
  72. 72.Hao Y, Khoo H M, von Ellenrieder N, Zazubovits N and Gotman J 2018 DeepIED: an epileptic discharge detector for EEG-fMRI based on deep learning NeuroImage 17 962–75
  73. 73.Harati A, López S, Obeid I and Picone J 2014 The TUH EEG CORPUS: a big data resource for automated EEG interpretation IEEE Signal Processing in Medicine and Biology Symp. (IEEE) pp 1–5
  74. 74.Hari R and Puce A 2017 MEG-EEG Primer (Oxford: Oxford University Press)
  75. 75.Hartmann K G, Schirrmeister R T and Ball T 2018 EEG-GAN: generative adversarial networks for electroencephalograhic (EEG) brain signals (arXiv:1806.01875)
  76. 76.Hartmann K G, Schirrmeister R T and Ball T 2018 Hierarchical internal representation of spectral features in deep convolutional networks trained for EEG decoding 6th Int. Conf. on Brain–Computer Interface (IEEE) pp 1–6
  77. 77.Hasib M M, Nayak T and Huang Y 2018 A hierarchical LSTM model with attention for modeling EEG non-stationarity for human decision prediction IEEE EMBS Int. Conf. on Biomedical and Health Informatics pp 104–7
  78. 78.He B, Sohrabpour A, Brown E and Liu Z 2018 Electrophysiological source imaging: a noninvasive window to brain dynamics Annu. Rev. Biomed. Eng. 20 171–96
  79. 79.He K, Zhang X, Ren S and Sun J 2015 Deep residual learning for image recognition Proc. IEEE Conf. on Computer Vision and Pattern Recognition pp 770–8
  80. 80.Hefron R, Borghetti B, Schubert Kabban C, Christensen J and Estepp J 2018 Cross-participant EEG-based assessment of cognitive workload using multi-path convolutional recurrent neural networks Sensors 18 1339
  81. 81.Hefron R G, Borghetti B J, Christensen J C and Kabban C M S 2017 Deep long short-term memory structures model temporal dependencies improving cognitive workload estimation Pattern Recognit. Lett. 94 96–104
  82. 82.Heilmeyer F A, Schirrmeister R T, Fiederer L D J, Völker M, Behncke J and Ball T 2018 A large-scale evaluation framework for EEG deep learning architectures (arXiv:1806.07741)
  83. 83.Heusel M, Ramsauer H, Unterthiner T, Nessler B and Hochreiter S 2017 GANs trained by a two time-scale update rule converge to a local nash equilibrium Advances in Neural Information Processing Systems pp 6626–37
  84. 84.Hohman F M, Kahng M, Pienta R and Chau D H 2018 Visual analytics in deep learning: an interrogative survey for the next frontiers IEEE Trans. Vis. Comput. Graph. pp 2674–93
  85. 85.Hussein R, Palangi H, Ward R and Wang Z J 2018 Epileptic seizure detection: a deep learning approach (arXiv:1803.09848)
  86. 86.Jas M, Engemann D A, Bekhti Y, Raimondo F and Gramfort A 2017 Autoreject: automated artifact rejection for MEG and EEG data NeuroImage 159 417–29
  87. 87.Jayaram V and Barachant A 2018 MOABB: trustworthy algorithm benchmarking for BCIs J. Neural Eng. 15 066011
  88. 88.Jirayucharoensak S, Pan-Ngum S and Israsena P 2014 EEG-based emotion recognition using deep learning network with principal component based covariate shift adaptation Sci. World J. 2014 627892
  89. 89.Kemp B, Zwinderman A H, Tuk B, Kamphuisen H A and Oberye J J 2000 Analysis of a sleep-dependent neuronal feedback loop: the slow-wave microcontinuity of the EEG IEEE Trans. Biomed. Eng. 47 1185–94
  90. 90.Kingma D P and Ba J 2014 Adam: a method for stochastic optimization (arXiv:1412.6980)
  91. 91.Koelstra S, Mühl C, Soleymani M, Lee J S, Yazdani A, Ebrahimi T, Pun T, Nijholt A and Patras I 2012 DEAP: a database for emotion analysis; using physiological signals IEEE Trans. Affective Comput. 3 18–31
  92. 92.Kuanar S, Athitsos V, Pradhan N, Mishra A and Rao K R 2018 Cognitive analysis of working memory load from EEG, by a deep recurrent neural network IEEE Signal Process. Soc. pp 2576–80
  93. 93.Kwak N S, Müller K R and Lee S W 2017 A convolutional neural network for steady state visual evoked potential classification under ambulatory environment PLoS One 12 1–20
  94. 94.Kwon Y, Nan Y and Kim S D 2017 Transformation of EEG signal for emotion analysis and dataset construction for DNN learning Lecture Notes in Electrical Engineering vol 474 (Berlin: Springer) pp 96–101
  95. 95.Längkvist M and Loutfi A 2018 A deep learning approach with an attention mechanism for automatic sleep stage classification (arXiv: 1805.05036)
  96. 96.Längkvist M, Karlsson L and Loutfi A 2012 Sleep stage classification using unsupervised feature learning Adv. Artif. Neural Syst. 2012 107046
  97. 97.Lawhern V J, Solon A J and Waytowich N R 2018 EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces J. Neural Eng. 15 056013
  98. 98.LeCun Y, Bengio Y and Hinton G 2015 Deep learning Nature 521 436
  99. 99.LeCun Y, Boser B, Denker J S, Henderson D, Howard R E, Hubbard W and Jackel L D 1989 Backpropagation applied to handwritten zip code recognition Neural Comput. 1 541–51
  100. 100.LeCun Y, Bottou L, Bengio Y and Haffner P 1998 Gradient-based learning applied to document recognition Proc. IEEE 86 2278–324
  101. 101.Lee W H, Ortiz J, Ko B and Lee R 2018 Time series segmentation through automatic feature learning (arXiv:1801.05394)
  102. 102.Lee Y and Huang Y 2018 Generating target/non-target images of an RSVP experiment from brain signals in by conditional generative adversarial network 2018 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI) pp 182–5
  103. 103.Li F, Zhang G, Wang W, Xu R, Schnell T, Wen J, McKenzie F and Li J 2017 Deep models for engagement assessment with scarce label information IEEE Trans. Hum.-Mach. Syst. 47 598–605
  104. 104.Li J and Cichocki A 2014 Deep learning of multifractal attributes from motor imagery induced EEG Neural Information Processing pp 503–10
  105. 105.Li J, Struzik Z, Zhang L and Cichocki A 2015 Feature learning from incomplete EEG with denoising autoencoder Neurocomputing 165 23–31
  106. 106.Li K, Li X, Zhang Y and Zhang A 2013 Affective state recognition from EEG with deep belief networks Proc.—2013 IEEE Int. Conf. on Bioinformatics and Biomedicine pp 305–10
  107. 107.Li R, Johansen J S, Ahmed H, Ilyevsky T V, Wilbur R B, Bharadwaj H M and Siskind J M 2018 Training on the test set? An analysis of Spampinato et al (arXiv:1609.00344)
  108. 108.Li X, Grandvalet Y and Davoine F 2018 Explicit inductive bias for transfer learning with convolutional networks (arXiv:1802.01483)
  109. 109.Li X, Zhang P, Song D, Yu G, Hou Y and Hu B 2015 EEG based emotion identification using unsupervised deep feature learning SIGIR2015 Workshop on Neuro-Physiological Methods in IR Research pp 2–4
  110. 110.Li Y, Schwing A, Wang K C and Zemel R 2017 Dualing GANs Advances in Neural Information Processing Systems pp 5606–16
  111. 111.Li Z, Tian X, Shu L, Xu X and Hu B 2018 Emotion recognition from EEG Using RASM and LSTM Internet Multimedia Computing and Service vol 819 (Berlin: Springer)
  112. 112.Liao C Y, Chen R C and Tai S K 2018 Emotion stress detection using EEG signal and deep learning technologies IEEE Int. Conf. on Applied System Invention pp 90–3
  113. 113.Lin W, Li C and Sun S 2017 Deep convolutional neural network for emotion recognition using EEG and peripheral physiological signal Int. Conf. on Image and Graphics pp 385–94
  114. 114.Liu W, Zheng W L and Lu B L 2016 Multimodal emotion recognition using multimodal deep learning
  115. 115.Loshchilov I and Hutter F 2016 SGDR: stochastic gradient descent with warm restarts (arXiv:1608.03983)
  116. 116.Lotte F, Bougrain L and Clerc M 2015 Electroencephalography (EEG)-Based Brain–Computer Interfaces (American Cancer Society) pp 1–20
  117. 117.Lotte F, Bougrain L, Cichocki A, Clerc M, Congedo M, Rakotomamonjy A and Yger F 2018 A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update J. Neural Eng. 15 031005
  118. 118.Major T C and Conrad J M 2017 The effects of pre-filtering and individualizing components for electroencephalography neural network classification Conf. Proc.—IEEE SOUTHEASTCON
  119. 119.Makeig S, Bell A J, Jung T P and Sejnowski T J 1996 Independent component analysis of electroencephalographic data Advances in Neural Information Processing Systems vol 8 pp 145–51
  120. 120.Manor R and Geva A B 2015 Convolutional neural network for multi-category rapid serial visual presentation BCI Frontiers Comput. Neurosci. 9 1–12
  121. 121.Manzano M, Guillén A, Rojas I and Herrera L J 2017 Deep learning using EEG data in time and frequency domains for sleep stage classification Lecture Notes Comput. Sci. 10305 LNCS 132–41 (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  122. 122.Munafò M R, Nosek B A, Bishop D V M, Button K S, Chambers C D, du Sert N P, Simonsohn U, Wagenmakers E-J, Ware J J and Ioannidis J P A 2017 A manifesto for reproducible science Nat. Hum. Behav. 1 0021
  123. 123.Mehmood R M, Du R and Lee H J 2017 Optimal feature selection and deep learning ensembles method for emotion recognition from human brain EEG sensors IEEE Access 5 14797–806
  124. 124.Mohamed A K, Marwala T and John L R 2011 Single-trial EEG discrimination between wrist and finger movement imagery and execution in a sensorimotor BCI Ann. Int. Conf. Proc. IEEE Eng. Med. Biol. Soc. pp 6289–93
  125. 125.Moinnereau M A, Brienne T, Brodeur S, Rouat J, Whittingstall K and Plourde E 2018 Classification of auditory stimuli from EEG signals with a regulated recurrent neural network reservoir
  126. 126.Morabito F C, Campolo M, Ieracitano C, Ebadi J M, Bonanno L, Bramanti A, Desalvo S, Mammone N and Bramanti P 2016 Deep convolutional neural networks for classification of mild cognitive impaired and Alzheimer's disease patients from scalp EEG recordings IEEE 2nd Int. Forum on Research and Technologies for Society and Industry Leveraging a Better Tomorrow
  127. 127.Morabito F C et al 2017 Deep learning representation from electroencephalography of early-stage Creutzfeldt–Jakob disease and features for differentiation from rapidly progressive dementia Int. J. Neural Syst. 27 1650039
  128. 128.Naderi M A and Mahdavi-Nasab H 2010 Analysis and classification of EEG signals using spectral analysis and recurrent neural networks 17th Iranian Conf. of Biomedical Engineering (IEEE) pp 1–4
  129. 129.Narejo S, Pasero E and Kulsoom F 2016 EEG based eye state classification using deep belief network and stacked autoencoder Int. J. Electr. Comput. Eng. 6 3131–41
  130. 130.Niso G et al 2018 Meg-bids, the brain imaging data structure extended to magnetoencephalography Sci. Data 5 180110
  131. 131.Nolan H, Whelan R and Reilly R B 2010 FASTER: fully automated statistical thresholding for EEG artifact rejection J. Neurosci. Methods 192 152–62
  132. 132.Normand R and Ferreira H A 2015 Superchords: the atoms of thought (arXiv:150501228)
  133. 133.Nurse E, Mashford B S, Yepes A J, Kiral-Kornek I, Harrer S and Freestone D R 2016 Decoding EEG and LFP signals using deep learning: heading TrueNorth Proc. of the ACM Int. Conf. on Computing Frontiers pp 259–66
  134. 134.O 'shea A, Lightbody G, Boylan G and Temko A 2017 Neonatal seizure detection using convolutional neural networks (arXiv:1709.05849)
  135. 135.Omerhodzic I, Avdakovic S, Nuhanovic A and Dizdarevic K 2013 Energy distribution of EEG signals: EEG signal wavelet-neural network classifier 2 (arXiv:1307.7897)
  136. 136.Oord A V D et al 2017 Parallel wavenet: fast high-fidelity speech synthesis (arXiv:1711.10433)
  137. 137.O'reilly C, Gosselin N, Carrier J and Nielsen T 2014 Montreal archive of sleep studies: an open-access resource for instrument benchmarking and exploratory research J. Sleep Res. 23 628–35
  138. 138.O'Shea A, Lightbody G, Boylan G and Temko A 2018 Investigating the impact of CNN depth on neonatal seizure detection performance (arXiv: 1806.03044)
  139. 139.Padmanabh L, Shastri R and Biradar S 2017 Mental tasks classification using EEG signal, discrete wavelet transform and neural network Discovery 48 38–41
  140. 140.Paez A 2017 Gray literature: an important resource in systematic reviews J. Evidence-Based Med. 10 233–40
  141. 141.Page A, Shea C and Mohsenin T 2016 Wearable seizure detection using convolutional neural networks with transfer learning IEEE Int. Symp. on Circuits and Systems pp 1086–9
  142. 142.Palazzo S, Spampinato C, Kavasidis I, Giordano D and Shah M 2017 Generative adversarial networks conditioned by brain signals Proc. IEEE Int. Conf. on Computer Vision pp 3430–8
  143. 143.Pardede J, Turnip M, Manalu D R and Turnip A 2015 Adaptive recurrent neural network for reduction of noise and estimation of source from recorded EEG signals ARPN J. Eng. Appl. Sci. 10 993–7
  144. 144.Parekh V, Subramanian R, Roy D and Jawahar C V 2018 An EEG-based image annotation system Commun. Comput. Inf. Sci. 841 303–13
  145. 145.Patanaik A, Ong J L, Gooley J J, Ancoli-Israel S and Chee M W L 2018 An end-to-end framework for real-time automatic sleep stage classification Sleep 41 1–11
  146. 146.Patnaik S, Moharkar L and Chaudhari A 2017 Deep RNN learning for EEG based functional brain state inference Int. Conf. on Advances in Computing, Communication and Control
  147. 147.Perez-Benitez J L, Perez-Benitez J A and Espina-Hernandez J H 2018 Development of a brain computer interface interface using multi-frequency visual stimulation and deep neural networks Int. Conf. on Electronics, Communications and Computers pp 18–24
  148. 148.Perez L and Wang J 2017 The effectiveness of data augmentation in image classification using deep learning (arXiv:1712.04621)
  149. 149.Pernet C R, Appelhoff S, Flandin G, Phillips C, Delorme A and Oostenveld R 2019 Bids-EEG: an extension to the brain imaging data structure (bids) specification for electroencephalography Sci. Data 6 103
  150. 150.Phan H, Andreotti F, Cooray N, Chen O Y and De Vos M 2019 Joint classification and prediction CNN framework for automatic sleep stage classification IEEE Trans. Biomed. Eng. 66 1285–96
  151. 151.Prechelt L 1998 Automatic early stopping using cross validation: quantifying the criteria Neural Netw. 11 761–7
  152. 152.Radford A, Wu J, Child R, Luan D, Amodei D and Sutskever I 2018 Language models are unsupervised multitask learners Technical Report OpenAi
  153. 153.Raposo F, de Matos D M, Ribeiro R, Tang S and Yu Y 2017 Towards deep modeling of music semantics using EEG regularizers (arXiv:1712.05197)
  154. 154.Robbins H and Monro S 1951 A stochastic approximation method Statistics (Berlin: Springer) pp 102–9
  155. 155.Rosenblatt F 1958 The perceptron: a probabilistic model for information storage and organization Psychol. Rev. 65 386–408
  156. 156.Roy S, Asif U, Tang J and Harrer S 2019 Machine learning for seizure type classification: setting the benchmark
  157. 157.Roy S, Kiral-Kornek I and Harrer S 2018 ChronoNet: a deep recurrent neural network for abnormal EEG identification (arXiv:1802.00308)
  158. 158.Ruffini G, Ibanez D, Castellano M, Dubreuil L, Gagnon J F, Montplaisir J and Soria-Frisch A 2018 Deep learning with EEG spectrograms in rapid eye movement behavior disorder (bioRxiv: 240267)
  159. 159.Rumelhart D E, Hinton G E and Williams R J 1986 Learning representations by back-propagating errors Nature 323 533
  160. 160.Sajda P, Gerson A, Muller K R, Blankertz B and Parra L 2003 A data analysis competition to evaluate machine learning algorithms for use in brain-computer interfaces IEEE Trans. Neural Syst. Rehabil. Eng. 11 184–5
  161. 161.Sakhavi S and Guan C 2017 Convolutional neural network-based transfer learning and knowledge distillation using multi-subject data in motor imagery BCI Int. IEEE/EMBS Conf. on Neural Engineering pp 588–91
  162. 162.Sakhavi S, Guan C and Yan S 2015 Parallel convolutional-linear neural network for motor imagery classification 23rd European Signal Processing Conf. pp 2736–40
  163. 163.Salimans T, Goodfellow I, Zaremba W, Cheung V, Radford A and Chen X 2016 Improved techniques for training GANs Advances in Neural Information Processing Systems pp 2234–42
  164. 164.Saon G et al 2017 English conversational telephone speech recognition by humans and machines (arXiv:1703.02136)
  165. 165.Schalk G, McFarland D J, Hinterberger T, Birbaumer N and Wolpaw J R 2004 Bci2000: a general-purpose brain-computer interface (bci) system IEEE Trans. Biomed. Eng. 51 1034–43
  166. 166.Schirrmeister R T, Gemein L, Eggensperger K, Hutter F and Ball T 2017 Deep learning with convolutional neural networks for decoding and visualization of EEG pathology (arXiv:1708.08012)
  167. 167.Schirrmeister R T, Springenberg J T, Fiederer L D J, Glasstetter M, Eggensperger K, Tangermann M, Hutter F, Burgard W and Ball T 2017 Deep learning with convolutional neural networks for EEG decoding and visualization Hum. Brain Mapp. 38 5391–420
  168. 168.Schmidhuber J 2015 Deep learning in neural networks: an overview: read section 6.6 Neural Netw. 61 85–117
  169. 169.Schomer D L and Da Silva F L 2012 Niedermeyer's Electroencephalography: Basic Principles, Clinical Applications, and Related Fields (Baltimore, MD: Williams & Wilkins)
  170. 170.Schwabedal J T C, Snyder J C, Cakmak A, Nemati S and Clifford G D 2018 Addressing class imbalance in classification problems of noisy signals by using fourier transform surrogates (arXiv:1806.08675)
  171. 171.Shah V, Golmohammadi M, Ziyabari S, Von Weltin E, Obeid I and Picone J 2017 Optimizing channel selection for seizure detection Signal Processing in Medicine and Biology Symp. (IEEE) pp 1–5
  172. 172.Shamwell J, Lee H, Kwon H, Marathe A R, Lawhern V and Nothwang W 2016 Single-trial EEG RSVP classification using convolutional neural networks Micro-and Nanotechnology Sensors, Systems, and Applications VIII vol 9836
  173. 173.Shang J, Yuanyue H, Haixiang G, Yijing L, Mingyun G and Gong B 2017 Learning from class-imbalanced data: Review of methods and applications Expert Syst. Appl. 73 220–39
  174. 174.Shoeb A H and Guttag J V 2010 Application of machine learning to epileptic seizure detection Proc. of the 27th Int. Conf. on Machine Learning pp 975–82
  175. 175.Simonyan K and Zisserman A 2014 Very deep convolutional networks for large-scale image recognition (arXiv:1409.1556)
  176. 176.Snoek J, Larochelle H and Adams R P 2012 Practical bayesian optimization of machine learning algorithms Advances in Neural Information Processing Systems pp 2951–9
  177. 177.Soleymani M, Lichtenauer J, Pun T and Pantic M 2012 A multimodal database for affect recognition and implicit tagging IEEE Trans. Affective Comput. 3 42–55
  178. 178.Sors A, Bonnet S, Mirek S, Vercueil L and Payen J F 2018 A convolutional neural network for sleep stage scoring from raw single-channel EEG Biomed. Signal Process. Control 42 107–14
  179. 179.Spampinato C, Palazzo S, Kavasidis I, Giordano D, Souly N and Shah M 2017 Deep learning human mind for automated visual classification Proc.—30th IEEE Conf. on Computer Vision and Pattern Recognition pp 4503–11
  180. 180.Sree R A and Kavitha A 2017 Vowel classification from imagined speech using sub-band EEG frequencies and deep belief networks 4th Int. Conf. on Signal Processing, Communication and Networking (IEEE) pp 1–4
  181. 181.Stober S, Cameron D J and Grahn J A 2014 Using convolutional neural networks to recognize rhythm stimuli from electroencephalography recordings Neural Information Processing Systems pp 1–9
  182. 182.Stober S, Sternin A, Owen A M and Grahn J A 2015 Deep feature learning for EEG recordings (arXiv:1511.04306)
  183. 183.Sturm I, Bach S, Samek W and Müller K R 2016 Interpretable deep neural networks for single-trial EEG classification (arXiv:1604.08201)
  184. 184.Sun B and Saenko K 2016 Deep coral: correlation alignment for deep domain adaptation European Conf. on Computer Vision (Berlin: Springer) pp 443–50
  185. 185.Sun P and Qin J 2016 Neural networks based EEG-speech models 1–10 (arXiv:1612.05369)
  186. 186.Supratak A, Dong H, Wu C and Guo Y 2017 DeepSleepNet: a model for automatic sleep stage scoring based on raw single-channel EEG IEEE Trans. Neural Syst. Rehabil. Eng. 25 1998–2008
  187. 187.Sutskever I, Hinton G, Krizhevsky A and Salakhutdinov R R 2014 Dropout: a simple way to prevent neural networks from overfitting J. Mach. Learn. Res. 15 1929–58
  188. 188.Szegedy C, Vanhoucke V, Ioffe S, Shlens J and Wojna Z 2016 Rethinking the inception architecture for computer vision Proc. IEEE Conf. on Computer Vision and Pattern Recognition pp 2818–26
  189. 189.Tabar Y R and Halici U 2016 A novel deep learning approach for classification of EEG motor imagery signals J. Neural Eng. 14 16003
  190. 190.Talathi S S 2017 Deep recurrent neural networks for seizure detection and early seizure detection systems (arXiv:1706.03283)
  191. 191.Tang Z, Li C and Sun S 2017 Single-trial EEG classification of motor imagery using deep convolutional neural networks Optik 130 11–8
  192. 192.Taqi A M, Al-Azzo F, Mariofanna M and Al-Saadi J M 2017 Classification and discrimination of focal and non-focal EEG signals based on deep neural network Int. Conf. on Current Research in Computer Science and Information Technology pp 86–92
  193. 193.Teo J, Hou C L and Mountstephens J 2018 Preference classification using electroencephalography (EEG) and deep learning J. Telecommun. Electron. Comput. Eng. 10 87–91
  194. 194.Thodoroff P, Pineau J and Lim A 2016 Learning robust features using deep learning for automatic seizure detection (arXiv:1608.00220)
  195. 195.Thorsten O Z and Christian K 2011 Towards passive brain–computer interfaces: applying brain–computer interface technology to human–machine systems in general J. Neural Eng. 8 25005
  196. 196.Tibor S R, Tobias S J, Josef F L D, Martin G, Katharina E, Michael T, Frank H, Wolfram B and Tonio B 2017 Deep learning with convolutional neural networks for EEG decoding and visualization Hum. Brain Mapp. 38 5391–420
  197. 197.Tieleman T, Hinton G E, Srivastava N and Swersky K 2012 Lecture 6.5-rmsprop: divide the gradient by a running average of its recent magnitude COURSERA: Neural Netw. Mach. Learn. 4 26–31
  198. 198.Tripathy R K and Rajendra Acharya U 2018 Use of features from RR-time series and EEG signals for automated classification of sleep stages in deep neural network framework Biocybern. Biomed. Eng. 38 890–902
  199. 199.Truong N D, Kuhlmann L, Bonyadi M R and Kavehei O 2018 Semi-supervised seizure prediction with generative adversarial networks (arXiv:1806.08235)
  200. 200.Truong N D, Nguyen A D, Kuhlmann L, Bonyadi M R, Yang J, Ippolito S and Kavehei O 2018 Convolutional neural networks for seizure prediction using intracranial and scalp electroencephalogram Neural Netw. 105 104–11
  201. 201.Tsinalis O, Matthews P M, Guo Y and Zafeiriou S 2016 Automatic sleep stage scoring with single-channel EEG using convolutional neural networks (arXiv:1610.01683)
  202. 202.Tsiouris K M, Pezoulas V C, Zervakis M, Konitsiotis S, Koutsouris D D and Fotiadis D I 2018 A long short-term memory deep learning network for the prediction of epileptic seizures using EEG signals Comput. Biol. Med. 99 24–37
  203. 203.Turner J T, Page A, Mohsenin T and Oates T 2014 Deep belief networks used on high resolution multichannel electroencephalography data for seizure detection AAAI Spring Symp. Series pp 75–81
  204. 204.Ullah I, Hussain M, Qazi E U H and Aboalsamh H 2018 An automated system for epilepsy detection using EEG brain signals based on deep learning approach (arXiv:1801.05412)
  205. 205.Urigen J A and Garcia-Zapirain B 2015 EEG artifact removal state-of-the-art and guidelines J. Neural Eng. 12 031001
  206. 206.Van Putten M J A M, Olbrich S and Arns M 2018 Predicting sex from brain rhythms with deep learning Sci. Rep. 8
  207. 207.van Putten M J, de Carvalho R and Tjepkema-Cloostermans M C 2018 Deep learning for detection of epileptiform discharges from scalp EEG recordings Clin. Neurophysiol. 129 e98–9
  208. 208.Vanschoren J, van Rijn J N, Bischl B and Torgo L 2014 OpenML: networked science in machine learning SIGKDD Explorations 15 49–60
  209. 209.Vilamala A, Madsen K H and Hansen L K 2017 Deep convolutional neural networks for interpretable analysis of EEG sleep stage scoring (arXiv:1710.00633)
  210. 210.Volker M, Hammer J, Schirrmeister R T, Behncke J, Fiederer L D, Schulze-Bonhage A, Marusic P, Burgard W and Ball T 2018 Intracranial error detection via deep learning IEEE Int. Conf. on Systems, Man, and Cybernetics (IEEE) pp 568–75
  211. 211.Völker M, Schirrmeister R T, Fiederer L D J, Burgard W and Ball T 2017 Deep transfer learning for error decoding from non-invasive EEG 2018 6th Int. Conf. on Brain–Computer Interface (BCI) (IEEE) pp 1–6
  212. 212.Wang F, Zhong S H, Peng J, Jiang J and Liu Y 2018 Data augmentation for eeg-based emotion recognition with deep convolutional neural networks Lecture Notes Comput. Sci. 10705 LNCS 82–93 (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  213. 213.Wang S, Guo B, Zhang C, Bai X and Wang Z 2018 EEG detection and de-noising based on convolution neural network and Hilbert-Huang transform Proc.—2017 10th Int. Congress on Image and Signal Processing, BioMedical Engineering and Informatics pp 1–6
  214. 214.Waytowich N R, Lawhern V, Garcia J O, Cummings J, Faller J, Sajda P and Vettel J M 2018 Compact convolutional neural networks for classification of asynchronous steady-state visual evoked potentials (arXiv:1803.04566)
  215. 215.Wen T and Zhang Z 2018 Deep convolution neural network and autoencoders-based unsupervised feature learning of EEG signals IEEE Access 6 25399–410
  216. 216.Wilkinson M D 2016 Comment: The fair guiding principles for scientific data management and stewardship Sci. Data 3 160018
  217. 217.Wu Z, Wang H, Cao M, Chen Y and Xing E P 2018 Fair deep learning prediction for healthcare applications with confounder filtering 1–17
  218. 218.Wulsin D F, Gupta J R, Mani R, Blanco J A and Litt B 2011 Modeling electroencephalography waveforms with semi-supervised deep belief nets: fast classification and anomaly measurement J. Neural Eng. 8 036015
  219. 219.Xie S, Li Y, Xie X, Wang W and Duan X 2017 The analysis and classify of sleep stage using deep learning network from single-channel EEG signal Lecture Notes Comput. Sci. 10637 LNCS 752–8 (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
  220. 220.Xu H and Plataniotis K N 2016 Affective states classification using EEG and semi-supervised deep learning approaches IEEE 18th Int. Workshop on Multimedia Signal Processing pp 1–6
  221. 221.Yang B, Duan K and Zhang T 2016 Removal of EOG artifacts from EEG using a cascade of sparse autoencoder and recursive least squares adaptive filter Neurocomputing 214 1053–60
  222. 222.Yang B, Duan K, Fan C, Hu C and Wang J 2018 Automatic ocular artifacts removal in EEG using deep learning Biomed. Signal Process. Control 43 148–58
  223. 223.Yang H, Sakhavi S, Ang K K and Guan C 2015 On the use of convolutional neural networks and augmented CSP features for multi-class motor imagery of EEG signals classification Ann. Int. Conf. Proc. IEEE Eng. Med. Biol. Soc. pp 2620–3
  224. 224.Yang S, Golmohammadi M, Obeid I and Picone J 2016 Semi-automated annotation of signal events in clinical EEG data, Engineering Data Consortium, Temple University, Philadelphia, Pennsylvania, USA Signal Processing in Medicine and Biology Symp. pp 1–5
  225. 225.Yepes A J, Tang J and Mashford B S 2017 Improving classification accuracy of feedforward neural networks for spiking neuromorphic chips Int. Joint Conf. on Artificial Intelligence pp 1973–9
  226. 226.Yin Z and Zhang J 2016 Recognition of cognitive task load levels using single channel EEG and stacked denoising autoencoder Chinese Control Conf. (IEEE) pp 3907–12
  227. 227.Yin Z and Zhang J 2017 Cross-session classification of mental workload levels using EEG and an adaptive deep learning model Biomed. Signal Process. Control 33 30–47
  228. 228.Yin Z and Zhang J 2017 Cross-subject recognition of operator functional states via EEG and switching deep belief networks with adaptive weights Neurocomputing 260 349–66
  229. 229.Yogatama D, Dyer C, Ling W and Blunsom P 2017 Generative and discriminative text classification with recurrent neural networks (arXiv:1703.01898)
  230. 230.Yoon J, Lee J and Whang M 2018 Spatial and time domain feature of ERP speller system extracted via convolutional neural network Comput. Intell. Neurosci. 2018 6058065
  231. 231.Yuan Y, Xun G, Ma F, Suo Q, Xue H, Jia K and Zhang A 2018 A novel channel-aware attention framework for multi-channel EEG seizure detection via multi-view deep learning IEEE EMBS Int. Conf. on Biomedical & Health Informatics pp 4–7
  232. 232.Zafar R, Dass S C and Malik A S 2017 Electroencephalogram-based decoding cognitive states using convolutional neural network and likelihood ratio based score fusion PLoS ONE 12 e0178410
  233. 233.Zeiler M D 2012 ADADELTA: an adaptive learning rate method (arXiv:1212.5701)
  234. 234.Zhang C, Bengio S, Hardt M, Recht B and Vinyals O 2016 Understanding deep learning requires rethinking generalization (arXiv:1611.03530)
  235. 235.Zhang D, Yao L, Zhang X, Wang S, Chen W and Boots R 2018 Cascade and parallel convolutional recurrent neural networks on EEG-based intention recognition for brain computer interface 32nd AAAI Conf. on Artificial Intelligence pp 1703–10
  236. 236.Zhang G Q, Cui L, Mueller R, Tao S, Kim M, Rueschman M, Mariani S, Mobley D and Redline S 2018 The national sleep research resource: towards a sleep data commons. J. Am. Med. Inform. Assoc. 25 1351–8
  237. 237.Zhang J, Li S and Wang R 2017 Pattern recognition of momentary mental workload based on multi-channel electrophysiological data and ensemble convolutional neural networks Frontiers Neurosci. 11 1–16
  238. 238.Zhang Q and Liu Y 2018 Improving brain computer interface performance by data augmentation with conditional deep convolutional generative adversarial networks (arXiv:1806.07108)
  239. 239.Zhang T, Zheng W, Cui Z, Zong Y and Li Y 2018 Spatial-temporal recurrent neural network for emotion recognition IEEE Trans. Cybern. 1
  240. 240.Zhang X, Yao L, Chen K, Wang X, Sheng Q and Gu T 2017 DeepKey: an EEG and gait based dual-authentication system (arXiv:1706.01606)
  241. 241.Zhang X, Yao L, Huang C, Sheng Q Z and Wang X 2017 Intent recognition in smart living through deep recurrent neural networks International Conference on Neural Information Processing (Cham: Springer) pp 748–58
  242. 242.Zhang X, Yao L, Kanhere S S, Liu Y, Gu T and Chen K 2018 Mindid: Person identification from brain waves through attention-based recurrent neural network Proc. ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies vol 2 pp 149:1–23
  243. 243.Zhang X, Yao L, Sheng Q Z, Kanhere S S, Gu T and Zhang D 2018 Converting your thoughts to texts: enabling brain typing via deep feature learning of eeg signals IEEE Int. Conf. on Pervasive Computing and Communications (PerCom) pp 1–10
  244. 244.Zhang X, Yao L, Wang X, Zhang W, Zhang S and Liu Y 2018 Know your mind: adaptive brain signal classification with reinforced attentive convolutional neural networks (arXiv:1802.03996)
  245. 245.Zhang X, Yao L, Zhang D, Wang X, Sheng Q Z and Gu T 2017 Multi-person brain activity recognition via comprehensive EEG signal analysis Proc. of the 14th EAI Int. Conf. on Mobile and Ubiquitous Systems: Computing, Networking and Services (New York: ACM) pp 28–37
  246. 246.Zheng W L and Lu B L 2015 Investigating critical frequency bands and channels for EEG-based emotion recognition with deep neural networks IEEE Trans. Auton. Mental Dev. 7 162–75
  247. 247.Zheng W L, Liu W, Lu Y, Lu B L and Cichocki A 2019 Emotionmeter: a multimodal framework for recognizing human emotions IEEE Trans. Cybern. 49 1110–22
  248. 248.Zheng W L, Zhu J Y, Peng Y and Lu B L 2014 EEG-based emotion classification using deep belief networks Proc.—IEEE Int. Conf. on Multimedia and Expo pp 1–6
  249. 249.Zhou J and Xu W 2015 End-to-end learning of semantic role labeling using recurrent neural networks Proc. of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th Int. Joint Conf. on Natural Language Processing (Volume 1: Long Papers) vol 1 pp 1127–37

Citation

MLA
Roy, Y., et al. “Deep Learning-based Electroencephalography Analysis: A Systematic Review”. arXiv, 2019, http://arxiv.org/abs/1901.05498v2.
APA
Roy, Y., Banville, H., Albuquerque, I., Gramfort, A., Falk, T. H., & Faubert, J. (2019). Deep learning-based electroencephalography analysis: a systematic review. arXiv. http://arxiv.org/abs/1901.05498v2
Chicago
Roy, Y., H. Banville, I. Albuquerque, A. Gramfort, T. H. Falk, and J. Faubert. 2019. “Deep Learning-based Electroencephalography Analysis: A Systematic Review”. arXiv. http://arxiv.org/abs/1901.05498v2.
Harvard
Roy, Y. et al. (2019) “Deep learning-based electroencephalography analysis: a systematic review”, arXiv [Preprint]. Available at: http://arxiv.org/abs/1901.05498v2.
Vancouver
1. Roy Y, Banville H, Albuquerque I, Gramfort A, Falk TH, Faubert J (2019) Deep learning-based electroencephalography analysis: a systematic review. arXiv

BibTeX

@article{roy2019deep,
  title = {Deep learning-based electroencephalography analysis: a systematic review},
  author = {Roy, Yannick and Banville, Hubert and Albuquerque, Isabela and Gramfort, Alexandre and Falk, Tiago H. and Faubert, Jocelyn},
  year = {2019},
  journal = {arXiv},
  url = {http://arxiv.org/abs/1901.05498v2},
  eprint = {1901.05498}
}
Metadata:arXiv

Access the Paper

This paper is available from its original source. Click below to access the PDF.

Open PDF