Feature Selection Using Extreme Gradient Boosting Bayesian Optimization to upgrade the Classification Performance of Motor Imagery signals for BCI

Thenmozhi, T.; Helen, R.

doi:10.1016/j.jneumeth.2021.109425

Cited by 25 publications

(8 citation statements)

References 62 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Furthermore, when targeting mobile EEG devices, removing movement artifacts can lead to better EEG signal quality and improved performance [ 46 ]. Another possible improvement could be to tailor the proposed decoder to individual subjects by using some form of automatic hyperparameter tuning—for example, applying Bayesian optimization for feature selection [ 47 ] and hyperparameter optimization for the network structures [ 48 ]. Moreover, other BCI paradigms might lead to more user-friendly designs; for example, one could use imagined speech to switch between walking, rotation, and standing, while using IM for rotating only.…”

Section: Discussionmentioning

confidence: 99%

Real-Time Navigation in Google Street View® Using a Motor Imagery-Based BCI

Yang

Hulle

2023

Sensors

View full text Add to dashboard Cite

Navigation in virtual worlds is ubiquitous in games and other virtual reality (VR) applications and mainly relies on external controllers. As brain–computer interfaces (BCI)s rely on mental control, bypassing traditional neural pathways, they provide to paralyzed users an alternative way to navigate. However, the majority of BCI-based navigation studies adopt cue-based visual paradigms, and the evoked brain responses are encoded into navigation commands. Although robust and accurate, these paradigms are less intuitive and comfortable for navigation compared to imagining limb movements (motor imagery, MI). However, decoding motor imagery from EEG activity is notoriously challenging. Typically, wet electrodes are used to improve EEG signal quality, including a large number of them to discriminate between movements of different limbs, and a cuedbased paradigm is used instead of a self-paced one to maximize decoding performance. Motor BCI applications primarily focus on typing applications or on navigating a wheelchair—the latter raises safety concerns—thereby calling for sensors scanning the environment for obstacles and potentially hazardous scenarios. With the help of new technologies such as virtual reality (VR), vivid graphics can be rendered, providing the user with a safe and immersive experience; and they could be used for navigation purposes, a topic that has yet to be fully explored in the BCI community. In this study, we propose a novel MI-BCI application based on an 8-dry-electrode EEG setup, with which users can explore and navigate in Google Street View®. We pay attention to system design to address the lower performance of the MI decoder due to the dry electrodes’ lower signal quality and the small number of electrodes. Specifically, we restricted the number of navigation commands by using a novel middle-level control scheme and avoided decoder mistakes by introducing eye blinks as a control signal in different navigation stages. Both offline and online experiments were conducted with 20 healthy subjects. The results showed acceptable performance, even given the limitations of the EEG set-up, which we attribute to the design of the BCI application. The study suggests the use of MI-BCI in future games and VR applications for consumers and patients temporarily or permanently devoid of muscle control.

show abstract

Section: Discussionmentioning

confidence: 99%

Real-Time Navigation in Google Street View® Using a Motor Imagery-Based BCI

Yang

Hulle

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…This operation reduces the memory usage. (10) where includes the key operations of multihead ProbSparse self-attention and attention block, is the 1D convolution on the time series, is the activation function, and is the maximum-pooling layer with a step length of 2. With the increase in the number of encoder layers after the self-attention distilling operation, the cumulative time-series length at each layer was half of that in the upper layer; this restricted the data volume of the model parameters and reduced the demand for video memory of the GPU.…”

Section: Figure 5 Eeg Generated In Different Parts Of the Brain Is Su...mentioning

confidence: 99%

“…Another method is motor imagery (MI), in which users imagine limb motions to input data into the BCI. Thenmozhi et al improved the classification performance of MI signals for BCI [10]. Attallah et al used MI-BCI to assist people with limb motor disabilities by enabling them to control assistive devices through their brain signals [11].…”

Section: Introductionmentioning

confidence: 99%

Super-Resolution Level Separation: A Method for Enhancing Electroencephalogram Classification Accuracy Through Super-Resolution Level Separation

Sun,

Li,

Zhang

2024

IEEE Access

View full text Add to dashboard Cite

Accurate identification of electroencephalogram (EEG) signals forms the basis for the development and application of brain-computer interface (BCI) devices. Signal preprocessing is an essential part of most EEG classification and recognition systems.In this study, we proposed a method called Superresolution level separation (SRLS) to add dimensions of information to the classification model. First, we calculated the correlation between the EEG signals acquired by each channel and divided these channels into multiple levels. Next, we used a super-resolution method to calculate common EEGs (C-EEGs) acquired by the channels. In addition, we designed an EEG-split-informer (ES-informer) model based on an informer model to enable small-sample users to obtain highly fitting C-EEGs. We then calculated the difference between the C-EEG and true EEG (T-EEG) to obtain the unique EEGs (U-EEGs) acquired by each channel, thus, adding dimensions to the data inputted to the classification model. Utilizing the 2008 2a motor imagery (MI) EEG dataset from the BCI Competition and the P300 paradigm data collected, EEGNet was employed as the classification model to validate the efficacy of the proposed SRLS method. The results of the experiments indicated that the SRLS method augmented the input dimension of the model and amplified the classification accuracy by over 7% for MI and 3.6% for P300. These findings demonstrate that SRLS is capable of enhancing the recognition accuracy of EEG.

show abstract

“…Filters focused on small data in classification to improve performance [34]. Some studies used fast fourier transform (FFT) [35]. In comparison, other studies use Wavelet [31].…”

Section: Introductionmentioning

confidence: 99%

Multi-channel of electroencephalogram signal in multivariable brain-computer interface

Djamal

Sury

2023

IJ-AI

View full text Add to dashboard Cite

Brain-computer interface (BCI) usually uses Electroencephalogram (EEG) signals as an intermediate device to drive external devices directly from the brain. The development of BCI capabilities is carried out by involving multivariable EEG signals as movement commands. EEG signals are recorded using multi-channel, enriching information if it uses the suitable method and architecture. This research proposed a two-dimensional convolutional neural networks (CNN) method to recognize multi-channel EEG signals. The vertical dimension is the channel, while the horizontal is the signal sequence. Hence, the signal is connected with the information time series of the same channel and between channels simultaneously. BCI was arranged with multivariable signals, specifically motor imagery and emotion. Both variables have different characteristics, and the information is from different channels. Therefore, it needs multiple CNNs to recognize the two variables in the EEG signal. The experiment showed that the accuracy of multiple 2D-CNN increased to 94.62% compared to 85.44% of single 2D CNN. Multiple 2D-CNN gave accuracy from 82.04% to 94.62% more than multiple 1D-CNN. 2D-CNN makes the channel extraction perfect into vectors to maintain the signal sequence. Signal extraction is essential, so the used Wavelet filter upgraded accuracy from 73.75% to 94.62%.

show abstract

Feature Selection Using Extreme Gradient Boosting Bayesian Optimization to upgrade the Classification Performance of Motor Imagery signals for BCI

Cited by 25 publications

References 62 publications

Real-Time Navigation in Google Street View® Using a Motor Imagery-Based BCI

Real-Time Navigation in Google Street View® Using a Motor Imagery-Based BCI

Super-Resolution Level Separation: A Method for Enhancing Electroencephalogram Classification Accuracy Through Super-Resolution Level Separation

Multi-channel of electroencephalogram signal in multivariable brain-computer interface

Contact Info

Product

Resources

About