A common assumption in traditional supervised learning is the similar probability distribution of data between training phase and testing/operating phase. When transitioning from to testing phase, a shift input known as covariate shift. Covariate shifts commonly arise wide range real-world systems such electroencephalogram-based brain–computer interfaces (BCIs). In systems, there necessity for continuous monitoring process behavior, tracking state decide about initiating adaptation timely...

The non-stationary nature of electroencephalography (EEG) signals makes an EEG-based brain-computer interface (BCI) a dynamic system, thus improving its performance is challenging task. In addition, it well-known that due to non-stationarity based covariate shifts, the input data distributions BCI systems change during inter- and intra-session transitions, which poses great difficulty for developments online adaptive data-driven systems. Ensemble learning approaches have been used previously...

Automatic vehicle following is an important feature of a fully or partially automated highway system (AHS). The on-board control should be able to accept and process inputs from the driver, infrastructure, other vehicles, perform diagnostics, provide appropriate commands actuators so that resulting motion safe compatible with AHS objectives. purpose this article design test in order achieve full automation longitudinal direction for several modes operation, where infrastructure manages...

Appropriately combining mental practice (MP) and physical (PP) in a post-stroke rehabilitation is critical for ensuring substantially positive outcome.Here we present protocol incorporating separate active PP stage followed by MP stage, using hand exoskeleton brain-computer interface (BCI).The was mediated force sensor feedback based assist-as-needed control strategy, whereas the provided BCI multimodal neurofeedback anthropomorphic visual proprioceptive of impaired extension attempt.A 6...

In this study, seven different types of regression-based predictive modelling techniques are used to predict the product gas composition (H2, CO, CO2, CH4) and yield (GY) during gasification biomass in a fluidised bed reactor. The performance models is compared with gradient boosting model (GB) show relative merits demerits technique. Additionally, SHapley Additive exPlanations (SHAP)-based explainable artificial intelligence (XAI) method was utilised explain individual predictions. This...

A major issue in electroencephalogram (EEG)-based brain-computer interfaces (BCIs) is the intrinsic nonstationarities brain waves, which may degrade performance of classifier, while transitioning from calibration to feedback generation phase. The nonstationary nature EEG data cause its input probability distribution vary over time, often appear as a covariate shift. To adapt shift, we had proposed an adaptive learning method our previous work and tested it on offline standard datasets. This...

Due to the non-stationary nature of electroencephalography (EEG) signals, a Brain-computer Interfacing (BCI) system requires frequent calibration. This leads inter session inconsistency which is one main reason that impedes widespread adoption non-invasive BCI for real-world applications, especially in rehabilitation and medicine. Domain adaptation deep learning-based techniques have gained relevance designing calibration-free BCIs solve this issue. EEGNet such net architecture has been...

Research and development of new machine learning techniques to augment the performance Brain-computer Interfaces (BCI) have always been an open area interest among researchers. The need develop robust generalised classifiers has one vital requirements in BCI for realworld application. EEGNet is a compact CNN model that had reported be different paradigms. In this paper, we aimed at further improving architecture by employing Neural Structured Learning (NSL) taps into relational information...

The objective is to evaluate the impact of EEG referencing schemes and spherical surface Laplacian (SSL) methods on classification performance motor-imagery (MI)-related brain-computer interface systems. Two schemes: common average three methods: current source density (CSD), finite difference method, SSL using realistic head model were implemented separately for pre-processing signals recorded at scalp. A combination filter bank spatial features extraction support vector machine was used...

Abstract Recent advancements in magnetoencephalography (MEG)-based brain-computer interfaces (BCIs) have shown great potential. However, the performance of current MEG-BCI systems is still inadequate and one main reasons for this unavailability open-source datasets. MEG are expensive hence datasets not readily available researchers to develop effective efficient BCI-related signal processing algorithms. In work, we release a 306-channel data recorded at 1KHz sampling frequency during four...

A major problem in a brain-computer interface (BCI) based on electroencephalogram (EEG) recordings is the varying statistical properties of signals during inter- or intra-session transfers that often lead to deteriorated BCI performances. filter bank CSP (FBCSP) algorithm typically uses all features from bands extract and select robust features. In this paper, we evaluate performance four methods for frequency band selection applied binary motor imagery classification: forward-addition (FA),...

Numbers and letters are the fundamental building blocks of our everyday social interactions. Previous studies have focused on determining cortical pathways shaped by numeracy literacy in human brain, partially supporting hypothesis distinct perceptual neural circuits involved visual processing two categories. In this study, we aim to investigate temporal dynamics for number letter processing. We present magnetoencephalography (MEG) data from experiments (N = 25 each). first experiment,...

Dataset shift is a major challenge in the non-stationary environments wherein input data distribution may change over time. Detecting dataset point time-series data, where of changes its properties, utmost interest. exists broad range real-world systems. In such systems, there need for continuous monitoring process behavior and tracking state so as to decide about initiating adaptive corrections timely manner. This paper presents an algorithm detect shift-point data. The proposed method...

A major issue for bringing brain-computer interface (BCI) based on electroencephalogram (EEG) recordings outside of laboratories is the non-stationarities EEG signals. Varying statistical properties signals during inter- or intra-session transfers can lead to deteriorated BCI performances over time. These variations may cause input data distribution shift when transitioning from training phase (calibration session) testing/operating resulting in a covariate shift. We propose handle this...

The brake subsystem is one of the most significant parts a vehicle with respect to safety. A computer-controlled system has capability acting faster than human driver during emergencies, and therefore potential improving safety vehicles used in future intelligent transportation systems (ITSs). In this paper we consider problem modeling control system. model developed using series experiments conducted on test bench which contains full scale Lincoln town car actuator designed by Ford Motor...

Learning with dataset shift is a major challenge in non-stationary environments wherein the input data distribution may over time. Detecting point time-series data, where of shifts its properties, utmost interest. Dataset exists broad range real-world systems. In such systems, there need for continuous monitoring process behavior and tracking state so as to decide about initiating adaptation timely manner. This paper presents an adaptive learning algorithm shift-detection using exponential...

Simultaneously occurring random events are often reported by multiple nodes. However, the accuracy of event detection at every node is dependent on node's relative position from event, and hence, not reliable. Moreover, factors influencing inference so many, that such an compromised. Targeting problem accurate in priority events, as forest fire, a fuzzy rule-based method proposed. Four parameters identified for which fuzzyfied values obtained membership function variable. A set 256 rules...

Non-invasive Brain-Computer Interface (BCI) has appeared as a new hope for large population of disabled people, who were waiting communication means that would translate some brain responses into actions. After several decades research in fields such neuroscience and machine learning, the performance remains too low due to signal noise ratio EEG signal, time be dedicated recording responses. Hybrid BCIs consider combination modalities, including responses, systems. The creation BCI requires...

In this paper our objective is to analyze the cortico-muscular coupling for hand finger motion and its possible use in control of an exoskeleton based neurorehabilitation system stroke sufferers. Cortical activity alone often not sufficient reliably a device such as hence, focus ascertain connectivity between motor cortex forearm muscles, controlling fingers, terms coherence electroencephalogram (EEG) electromyogram (EMG) signals. We have analyzed signals separately three different kinds...