EEG-based mental workload estimation using deep BLSTM-LSTM network and evolutionary algorithm

Chakladar, Debashis Das; Dey, Shubhashis; Roy, Partha Pratim; Dogra, Debi Prosad

doi:10.1016/j.bspc.2020.101989

Cited by 111 publications

(42 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Conversely, other SI based optimisers explored for DL include the salp-swarm optimiser [64], harmony search optimiser on variational stacked autoencoders [65], whale optimisation algorithm(WOA) using bidirectional RNN [66], the Artificial Bee Colony (ABC) for optimizing hyperparameters for LSTM models [67], the AC-Parametric WOA (ACP-WOA) [68] for predicting biomedical images, symbiotic organisms search (SOS) algorithm [69], lion swarm optimiser(LSO) [70], and many others. Nevertheless, a comparison of the performance of these genetic algorithms shows that the GWO convergence rate is fastest compared to the Genetic Algorithm (GA), and PSO [71]. However, there is no holistic comparison of the performance of these SI optimisers found in the literature.…”

Section: Particle Swarm Optimisers (Pso)mentioning

confidence: 99%

Advances in Optimisation Algorithms and Techniques for Deep Learning

Nwankpa¹

2020

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

In the last decade, deep learning(DL) has witnessed excellent performances on a variety of problems, including speech recognition, object recognition, detection, and natural language processing (NLP) among many others. Of these applications, one common challenge is to obtain ideal parameters during the training of the deep neural networks (DNN). These typical parameters are obtained by some optimisation techniques which have been studied extensively. These research have produced state-of-art(SOTA) results on speed and memory improvements for deep neural networks(NN) architectures. However, the SOTA optimisers have continued to be an active research area with no compilations of the existing optimisers reported in the literature. This paper provides an overview of the recent advances in optimisation algorithms and techniques used in DNN, highlighting the current SOTA optimisers, improvements made on these optimisation algorithms and techniques, alongside the trends in the development of optimisers used in training DL based models. The results of the search of the Scopus database for the optimisers in DL provides the articles reported as the summary of the DL optimisers. From what we can tell, there is no comprehensive compilation of the optimisation algorithms and techniques so far developed and used in DL research and applications, and this paper summarises these facts.

show abstract

Section: Particle Swarm Optimisers (Pso)mentioning

confidence: 99%

Advances in Optimisation Algorithms and Techniques for Deep Learning

Nwankpa¹

2020

Adv. sci. technol. eng. syst. j.

View full text Add to dashboard Cite

show abstract

“…Although implanted electrodes inside the human brain (invasive approach) show more effectiveness in recognizing the state of the brain [5][6][7], it has a hazardous risk of serious contagious gangrene and lethal death tolling outcomes. Therefore, researchers are more focused on noninvasive methods for EEG measurement [8] and gradually application spectrum of the EEG signal is increasing with recent including of mental workload estimation through EEG, brain-machine-based multi-target cursor movement, detection of emotions, and many more [9][10][11].…”

Section: Introductionmentioning

confidence: 99%

Common Spatial Pattern in Frequency Domain for Feature Extraction and Classification of Multichannel EEG Signals

Saha

Rahman²,

Alam

et al. 2021

SN COMPUT. SCI.

View full text Add to dashboard Cite

The extraction methodology of the significant features from the signals is one of the most important prerequisite steps for EEG signal classification. Common spatial pattern (CSP) is a widely used feature extraction method for EEG signal but with a lacking of failing to maintain discriminative features between classes in the time domain, and further as a consequence, ends up in inconvenience with erroneous output. To overcome the limitations of the convention CSP, this research work proposes a novel frequency domain CSP (FCSP) method for feature extraction. This method proposes to convert the time domain EEG signal to its power spectral density (PSD) so that the event-related variation can be found in the frequency domain. After that, the CSP method is applied to the PSD values of the selected channels to extract the variation based on the spatial pattern of the channels for the events. The output of this method helps to extract simple features from the FCSP-PSD data for the classification. The proposed method is applied to motor imagery data from BCI competition IV. To check the applicability of the proposed method, a complex environment was created considering the same lobe events such as combined left and right feet (Class#1) versus right-hand (Class#2) imagery movement. To compare the performance of the proposed work, the method is also applied to the conventional classification problem (left-hand vs right-hand imagery movement) and found very promising results of 91% accuracy on average.

show abstract

“…Beyindeki bilişsel aktiviteler sonucu oluşan değişimleri algılama hassasiyetinin yüksek olması nedeniyle, EEG sinyalleri zihinsel iş yükü değerlendirmelerinde sıklıkla kullanılan bir yöntemdir [13][14]. Zihinsel iş yükü seviyelerinin çok seviyeli sınıflandırılması üzerine çeşitli çalışmalar yürütülmüştür [15][16][17][18][19]. Wang ve arkadaşları [15], dört spesifik EEG alt bandı için hesapladıkları, entropi, sinyal gücü, morfolojik ve istatistiksel öznitelikleri içeren 658 öznitelikten oluşan öznitelik vektörünü kullanarak, mRMR tabanlı öznitelik seçimi ve destek vektör makineleri tabanlı sınıflandırma sonucu %84 sınıflandırma doğruluğuna erişmişlerdir.…”

Section: Gi̇ri̇ş (Introduction)unclassified

“…Hızlı Fourier Dönüşümü ile alt bantlarına ayrıştırılmış EEG sinyallerinin her alt bant için güç spektral yoğunluklarını öznitelik olarak kullanarak, düşük, orta ve yüksek zihinsel iş yükünü %69 doğrulukla belirlemişlerdir. Chakladar ve arkadaşları [19]…”

Section: Gi̇ri̇ş (Introduction)unclassified

Classification of Mental Workload Levels by Using EEG Signals

Aydın

2021

Politeknik Dergisi

View full text Add to dashboard Cite

 Fraktal boyut algoritmaları zihinsel iş yükü seviyelerinin sınıflandırılmasında kullanılabilir./Fractal dimension algorithms can be used for classifying mental workload levels.  Higuchi fraktal boyut (HFB) algoritması destek vektör makineleri ile birlikte zihinsel iş yükü seviyelerinin sınıflandırılmasında en yüksek sınıflandırma doğruluğu sağlamıştır./Higuchi fractal dimension (HFB) algorithm combined with support vector machines provided the highest classification accuracy in classifying mental workload levels.

show abstract

EEG-based mental workload estimation using deep BLSTM-LSTM network and evolutionary algorithm

Cited by 111 publications

References 21 publications

Advances in Optimisation Algorithms and Techniques for Deep Learning

Advances in Optimisation Algorithms and Techniques for Deep Learning

Common Spatial Pattern in Frequency Domain for Feature Extraction and Classification of Multichannel EEG Signals

Classification of Mental Workload Levels by Using EEG Signals

Contact Info

Product

Resources

About