Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
194 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

hvEEGNet: exploiting hierarchical VAEs on EEG data for neuroscience applications (2312.00799v1)

Published 20 Nov 2023 in eess.SP and cs.LG

Abstract: With the recent success of artificial intelligence in neuroscience, a number of deep learning (DL) models were proposed for classification, anomaly detection, and pattern recognition tasks in electroencephalography (EEG). EEG is a multi-channel time-series that provides information about the individual brain activity for diagnostics, neuro-rehabilitation, and other applications (including emotions recognition). Two main issues challenge the existing DL-based modeling methods for EEG: the high variability between subjects and the low signal-to-noise ratio making it difficult to ensure a good quality in the EEG data. In this paper, we propose two variational autoencoder models, namely vEEGNet-ver3 and hvEEGNet, to target the problem of high-fidelity EEG reconstruction. We properly designed their architectures using the blocks of the well-known EEGNet as the encoder, and proposed a loss function based on dynamic time warping. We tested the models on the public Dataset 2a - BCI Competition IV, where EEG was collected from 9 subjects and 22 channels. hvEEGNet was found to reconstruct the EEG data with very high-fidelity, outperforming most previous solutions (including our vEEGNet-ver3 ). Furthermore, this was consistent across all subjects. Interestingly, hvEEGNet made it possible to discover that this popular dataset includes a number of corrupted EEG recordings that might have influenced previous literature results. We also investigated the training behaviour of our models and related it with the quality and the size of the input EEG dataset, aiming at opening a new research debate on this relationship. In the future, hvEEGNet could be used as anomaly (e.g., artefact) detector in large EEG datasets to support the domain experts, but also the latent representations it provides could be used in other classification problems and EEG data generation.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (75)
  1. A review of machine learning and deep learning techniques for anomaly detection in iot data. Applied Sciences, 11, 5320.
  2. Convolutional autoencoder approach for eeg compression and reconstruction in m-health systems. In 2018 14th International Wireless Communications & Mobile Computing Conference (IWCMC) (pp. 370–375). doi:10.1109/IWCMC.2018.8450511.
  3. Wearable electroencephalography and multi-modal mental state classification: A systematic literature review. Computers in Biology and Medicine, (p. 106088).
  4. Indications of nonlinear deterministic and finite-dimensional structures in time series of brain electrical activity: Dependence on recording region and brain state. Physical review. E, Statistical, nonlinear, and soft matter physics, 64, 061907. doi:10.1103/PhysRevE.64.061907.
  5. Filter bank common spatial pattern (fbcsp) in brain-computer interface. In 2008 IEEE international joint conference on neural networks (IEEE world congress on computational intelligence) (pp. 2390–2397). IEEE.
  6. Correlation based dynamic time warping of multivariate time series. Expert Systems with Applications, 39, 12814–12823.
  7. Shared intelligence for robot teleoperation via bmi. IEEE Transactions on Human-Machine Systems, 52, 400–409. doi:10.1109/THMS.2021.3137035.
  8. Berger, H. (1929). On the eeg in humans. Arch. Psychiatr. Nervenkr, 87, 527–570.
  9. Eeg2vec: Learning affective eeg representations via variational autoencoders. In 2022 IEEE International Conference on Systems, Man, and Cybernetics (SMC) (pp. 3150–3157). doi:10.1109/SMC53654.2022.9945517.
  10. The non-invasive Berlin Brain-Computer Interface: Fast acquisition of effective performance in untrained subjects. NeuroImage, 37, 539–50. doi:10.1016/j.neuroimage.2007.01.051.
  11. Variational inference: A review for statisticians. Journal of the American statistical Association, 112, 859–877.
  12. Deep learning-based classification of fine hand movements from low frequency eeg. Future Internet, 13, 103.
  13. Neuronal oscillations in cortical networks. science, 304, 1926–1929.
  14. The need to separate the wheat from the chaff in medical informatics: Introducing a comprehensive checklist for the (self)-assessment of medical ai studies.
  15. Removal of muscle artifacts from the eeg: A review and recommendations. IEEE Sensors Journal, 19, 5353–5368.
  16. Real-time detection of eeg electrode displacement for brain-computer interface applications. In Proceedings of 5th International Conference on Wireless Communications, Vehicular Technology, Information Theory and Aerospace & Electronic Systems (Wireless VITAE) (pp. 1–5).
  17. Nearest neighbor pattern classification. IEEE transactions on information theory, 13, 21–27.
  18. Soft-dtw: a differentiable loss function for time-series. In International Conference on Machine Learning. URL: https://api.semanticscholar.org/CorpusID:9566599.
  19. Joint ecg–emg–eeg signal compression and reconstruction with incremental multimodal autoencoder approach. Circuits, Systems, and Signal Processing, 41, 6152–6181. URL: https://doi.org/10.1007/s00034-022-02071-x. doi:10.1007/s00034-022-02071-x.
  20. A longitudinal study investigating neural processing of speech envelope modulation rates in children with (a family risk for) dyslexia. Cortex, 93, 206–219. URL: https://www.sciencedirect.com/science/article/pii/S0010945217301624. doi:https://doi.org/10.1016/j.cortex.2017.05.007.
  21. Threaded ensembles of autoencoders for stream learning. Computational Intelligence, 34, 261–281.
  22. A simple system for detection of eeg artifacts in polysomnographic recordings. IEEE transactions on biomedical engineering, 50, 526–528.
  23. Autoencoding of long-term scalp electroencephalogram to detect epileptic seizure for diagnosis support system. Computers in Biology and Medicine, 110, 227–233. URL: https://www.sciencedirect.com/science/article/pii/S0010482519301933. doi:https://doi.org/10.1016/j.compbiomed.2019.05.025.
  24. Seizure detection by convolutional neural network-based analysis of scalp electroencephalography plot images. NeuroImage: Clinical, 22, 101684. URL: https://www.sciencedirect.com/science/article/pii/S2213158219300348. doi:https://doi.org/10.1016/j.nicl.2019.101684.
  25. A multi-artifact eeg denoising by frequency-based deep learning. arXiv preprint arXiv:2310.17335, .
  26. Automatic removal of various artifacts from eeg signals using combined methods. Journal of Clinical Neurophysiology, 27, 312–320.
  27. Training data distribution significantly impacts the estimation of tissue microstructure with machine learning. Magnetic Resonance in Medicine, 87, 932–947. URL: https://onlinelibrary.wiley.com/doi/abs/10.1002/mrm.29014. doi:https://doi.org/10.1002/mrm.29014. arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/mrm.29014.
  28. A review on machine learning for eeg signal processing in bioengineering. IEEE reviews in biomedical engineering, 14, 204--218.
  29. Eeg waveform analysis by means of dynamic time-warping. International journal of bio-medical computing, 17, 135--144.
  30. Moabb: trustworthy algorithm benchmarking for bcis. Journal of neural engineering, 15, 066011.
  31. Relationship between electrical brain responses to motor imagery and motor impairment in stroke. Stroke, 43, 2735--2740.
  32. A shallow autoencoder framework for epileptic seizure detection in eeg signals. Sensors, 23. URL: https://www.mdpi.com/1424-8220/23/8/4112. doi:10.3390/s23084112.
  33. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, .
  34. An introduction to variational autoencoders. arXiv preprint arXiv:1906.02691, .
  35. Thirty-minute motor imagery exercise aided by eeg sensorimotor rhythm neurofeedback enhances morphing of sensorimotor cortices: a double-blind sham-controlled study. Cerebral Cortex, 33, 6573--6584.
  36. Deap: A database for emotion analysis ;using physiological signals. IEEE Transactions on Affective Computing, 3, 18--31. doi:10.1109/T-AFFC.2011.15.
  37. Clinical evoked potentials in neurology: a review of techniques and indications. Journal of Neurology, Neurosurgery & Psychiatry, 88, 688--696.
  38. EEGNet: A compact convolutional network for EEG-based Brain-Computer Interfaces. Journal of Neural Engineering, 15. doi:10.1088/1741-2552/aace8c.
  39. Learning an autoencoder to compress eeg signals via a neural network based approximation of dtw. Procedia Computer Science, 222, 448--457. URL: https://www.sciencedirect.com/science/article/pii/S1877050923009481. doi:https://doi.org/10.1016/j.procs.2023.08.183. International Neural Network Society Workshop on Deep Learning Innovations and Applications (INNS DLIA 2023).
  40. Approximating dynamic time warping with a convolutional neural network on eeg data. Pattern Recognition Letters, 171, 162--169. URL: https://www.sciencedirect.com/science/article/pii/S0167865523001460. doi:https://doi.org/10.1016/j.patrec.2023.05.012.
  41. Densely feature fusion based on convolutional neural networks for motor imagery EEG classification. IEEE Access, 7, 132720--132730. doi:10.1109/ACCESS.2019.2941867.
  42. Constructing large-scale real-world benchmark datasets for aiops. arXiv:2208.03938.
  43. Eeg-based emotion classification using a deep neural network and sparse autoencoder. Frontiers in Systems Neuroscience, 14. URL: https://www.frontiersin.org/articles/10.3389/fnsys.2020.00043. doi:10.3389/fnsys.2020.00043.
  44. A review of classification algorithms for eeg-based brain--computer interfaces: a 10 year update. Journal of neural engineering, 15, 031005.
  45. Maghoumi, M. (2020). Deep Recurrent Networks for Gesture Recognition and Synthesis. Ph.D. thesis University of Central Florida Orlando, Florida.
  46. Deepnag: Deep non-adversarial gesture generation. In 26th International Conference on Intelligent User Interfaces (pp. 213--223).
  47. Machine learning approaches for anomaly detection of water quality on a real-world data set. Journal of Information and Telecommunication, 3, 294--307.
  48. Local or edge/cloud processing for data freshness, .
  49. Attempted arm and hand movements can be decoded from low-frequency eeg from persons with spinal cord injury. Scientific reports, 9, 7134.
  50. Dyslexia diagnosis by eeg temporal and spectral descriptors: An anomaly detection approach. International Journal of Neural Systems, 30, 2050029.
  51. Deep learning for anomaly detection: A review. ACM Comput. Surv., 54. URL: https://doi.org/10.1145/3439950. doi:10.1145/3439950.
  52. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12, 2825--2830.
  53. Iclabel: An automated electroencephalographic independent component classifier, dataset, and website. NeuroImage, 198, 181--197.
  54. Diverse super-resolution with pretrained deep hiererarchical vaes. arXiv preprint arXiv:2205.10347, .
  55. Kpi-tsad: A time-series anomaly detector for kpi monitoring in cloud applications. Symmetry, 11. URL: https://www.mdpi.com/2073-8994/11/11/1350. doi:10.3390/sym11111350.
  56. Denoising sparse autoencoder-based ictal eeg classification. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 26, 1717--1726. doi:10.1109/TNSRE.2018.2864306.
  57. Generating diverse high-fidelity images with vq-vae-2. In H. Wallach, H. Larochelle, A. Beygelzimer, F. d'Alché-Buc, E. Fox, & R. Garnett (Eds.), Advances in Neural Information Processing Systems. Curran Associates, Inc. volume 32. URL: https://proceedings.neurips.cc/paper_files/paper/2019/file/5f8e2fa1718d1bbcadf1cd9c7a54fb8c-Paper.pdf.
  58. MI-EEGNET: A novel convolutional neural network for motor imagery classification. Journal of Neuroscience Methods, 353, 109037. URL: https://www.sciencedirect.com/science/article/pii/S016502702030460X. doi:https://doi.org/10.1016/j.jneumeth.2020.109037.
  59. Learning temporal information for brain-computer interface using convolutional neural networks. IEEE Transactions on Neural Networks and Learning Systems, 29, 5619--5629.
  60. Dynamic programming algorithm optimization for spoken word recognition. IEEE Transactions on Acoustics, Speech, and Signal Processing, 26, 159--165. URL: https://api.semanticscholar.org/CorpusID:17900407.
  61. Finding a "kneedle" in a haystack: Detecting knee points in system behavior. In 2011 31st International Conference on Distributed Computing Systems Workshops (pp. 166--171). doi:10.1109/ICDCSW.2011.20.
  62. Deep learning with convolutional neural networks for EEG decoding and visualization. Human Brain Mapping, . doi:10.1002/hbm.23730.
  63. Neural tracking to go: auditory attention decoding and saliency detection with mobile eeg. Journal of neural engineering, 18, 066054.
  64. Teplan, M. et al. (2002). Fundamentals of eeg measurement. Measurement science review, 2, 1--11.
  65. Application of machine learning in epileptic seizure detection. Diagnostics, 12. URL: https://www.mdpi.com/2075-4418/12/11/2879. doi:10.3390/diagnostics12112879.
  66. Nvae: A deep hierarchical variational autoencoder. arXiv:2007.03898.
  67. Long-range forecasting in feature-evolving data streams. Knowledge-Based Systems, 206, 106405.
  68. An autoencoder-based approach to predict subjective pain perception from high-density evoked eeg potentials. In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) (pp. 1507--1511). doi:10.1109/EMBC44109.2020.9176644.
  69. Welch, P. (1967). The use of fast fourier transform for the estimation of power spectra: a method based on time averaging over short, modified periodograms. IEEE Transactions on audio and electroacoustics, 15, 70--73.
  70. Identifying data streams anomalies by evolving spiking restricted boltzmann machines. Neural Computing and Applications, 32, 6699--6713.
  71. veegnet: learning latent representations to reconstruct eeg raw data via variational autoencoders.
  72. CNN-based approaches for cross-subject classification in motor imagery: From the state-of-the-art to DynamicNet. In 2021 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB) (pp. 1--7). doi:10.1109/CIBCB49929.2021.9562821.
  73. veegnet: A new deep learning model to classify and generate eeg. In Proceedings of the 9th International Conference on Information and Communication Technologies for Ageing Well and e-Health - Volume 1: ICT4AWE, (pp. 245--252). INSTICC SciTePress. doi:10.5220/0011990800003476.
  74. Investigating critical frequency bands and channels for eeg-based emotion recognition with deep neural networks. IEEE Transactions on Autonomous Mental Development, 7, 162--175. doi:10.1109/TAMD.2015.2431497.
  75. Anomaly detection with robust deep autoencoders. In Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining KDD ’17 (p. 665–674). New York, NY, USA: Association for Computing Machinery. URL: https://doi.org/10.1145/3097983.3098052. doi:10.1145/3097983.3098052.
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now