One of the major problems associated with the motor imagery (MI) electroencephalogram (EEG) based brain–computer interface (BCI) classifications is the informative ambiguities mainly caused by interferences of artifacts and nonstationarities in EEG signals. Other factors containing mislabeling or misleading MI EEG trials might also cause more uncertainties in training datasets that lead to decline in classification performance. This paper proposes a new framework to achieve more efficient classification in multiclass MI EEG-based BCIs. An artifact rejected common spatial pattern (AR-CSP) method is proposed for feature extraction in order to cope with the interferences of artifacts. A self-regulated adaptive resonance theory based neuro-fuzzy classifier that is referred to as “self-regulated supervised Gaussian fuzzy adaptive system Art (SRSG-FasArt)” is introduced to deal with EEG nonstationarities. A metacognitive self-regulatory-based learning algorithm is also proposed to more efficiently deal with the uncertainties. The algorithm captures the training data samples by priority and automatically creates, upgrades, or prunes the fuzzy rules by scanning the knowledge content existing in the data patterns and the created rules. The mechanism improves the generalization capability of the SRSG-FasArt and prevents over-training. The performance of the proposed cooperative framework of AR-CSP and SRSG-FasArt is evaluated using the BCI competition IV dataset 2a. The results indicate more accurate and efficient BCI classification compared to the existing frameworks.

A New Self-Regulated Neuro-Fuzzy Framework for Classification of EEG Signals in Motor Imagery BCI

The strength of the adaptive neuro-fuzzy system (ANFIS) involves two contradictory requirements in a common fuzzy modeling problem, i.e. interpretability and accuracy. It is known that simultaneous optimization of accuracy and interpretability will improve performance of the system and avoid over-fitting of data. The objective of this study is the integration of particle swarm optimization (PSO) with ANFIS using modified linguistic and threshold values. This integration is expected to enhance the performance of the ANFIS system in classification problems. PSO is used to tune ANFIS parameters, to improve its classification accuracy. It is also used to find the optimal number of rules and their optimal interpretability. The proposed method has been tested on six standard data sets with different inputs of real and integer data types. The findings indicate that the proposed ANFIS---PSO integration provides a better result for classification, both in interpretability and accuracy.

Particle swarm optimization for ANFIS interpretability and accuracy

This paper presents the influence of the voting strategy to enhance the classification rates in motor imagery of brain–computer interface (BCI) systems. The motor imagery is the three-class problem of left-hand movement imagination, right-hand movement imagination, and word generation. An algorithm based on neural networks and fuzzy theory (S-dFasArt) is used to classify spontaneous mental activities from electroencephalogram signals, in order to operate a noninvasive BCI. This algorithm allows obtaining several prediction models. The voting among these prediction results improves the success rates of the classifier method. The number of models and the size of the data set have been analyzed obtaining some recommendation rules for practitioners. An improvement of more than 12% can be expected.

Voting Strategy to Enhance Multimodel EEG-Based Classifier Systems for Motor Imagery BCI

A fuzzy-based m-PSK attention detector for multi-talker scenarios is proposed.The approach outperformed the performance of previous works (ITR and accuracy).This outcome could have relevant impact on BCI community. The automatic and online detection of auditory attention in multi-talker scenarios (e.g., cocktail party paradigm) is a current topic in electroencephalography (EEG)-based brain-computer interfaces (BCIs). Recent works have demonstrated a way to make it possible by means of a model based on an m-ary phase shift keying (m-PSK) detector. However, this attention detection model lacks of relevant information such as the non-stationary nature of EEG signals, the neuro-plasticity/habituation effects or the nonlinearities of the attention. In this paper we propose an enriched version of the attention detection model constituted by an automatic adaptive m-PSK detector implemented on fuzzy logic. In it, the relevant information mentioned before is modeled as two inputs that feed the fuzzy-based attention detection model. The output provides the detection. Our enriched model outperformed the results of previous works in terms of mean information transfer rate (ITR) (4-PSK: 5.41bpm; 6-PSK: 6.03bpm) and accuracy (4-PSK: 0.54; 6-PSK: 0.39) after only 4.63 (4-PSK) and 2.93 (6-PSK) seconds of processing. The proposed model for the automatic detection of auditory attention can have relevant impact on several areas such as education, public transport, jobs, industry, attention disorders, ubiquitous systems, sports and art.

Detection of attention in multi-talker scenarios

/pdf/a-novel-adaptive-method-for-robotic-wheelchair-navigation-3jng7sb3v7.pdf

A Novel Adaptive Method for Robotic Wheelchair Navigation Using Brain Signals

This paper presents the first results of online classification using a model based on a neuro-fuzzy architecture called S-dFasArt, in order to recognize in real time and with sufficient reliability between two mental tasks in a Brain Computer Interface (BCI). Spontaneous brain activity recorded with non-invasive techniques and processed through the Fast Fourier Transform (FFT) has been used to test the classifier. The dynamic characteristics and the ability of the online classification of neuro-fuzzy algorithm make it very suitable to interpret EEG signals. The classifier designed is based on the creation and combination of diferent classification models S-dFasArt, alterning sessions of EEG signals in the adjustment phase and performing a complete study to find the best values of the classifier parameters. In the paper, the adjustment phase of each classification model has been described. New voting strategies and levels of uncertainty have been incorporated to improve the success rate in the online classification. The experimental results with different users have been reported.

/pdf/neuro-fuzzy-classifier-to-recognize-mental-tasks-in-a-bci-41odb2warl.pdf

Neuro-fuzzy classifier to recognize mental tasks in a BCI

espanolEl accidente cerebrovascular (ACV) es una de las principales causas en el mundo que provoca discapacidad. Por ello, la comunidad cientifca esta investigando la manera de ayudar a los afectados de ACV. Este trabajo utiliza las interfaces cerebro-maquina (BMI) y un exoesqueleto de miembro inferior para mejorar el proceso de rehabilitacion de dichas personas. De esta manera, el futuro paciente, estara mas involucrado en su terapia de rehabilitacion. El objetivo de este trabajo es estudiar si el ruido que general el movimiento del exoesqueleto afecta al EEG de las senales adquiridas del usuario y por ende verse influenciado en el desempeno de la BMI. Se ha realizado un estudio del espectro de la potencia de las senales utilizando la Transformada rapida de Fourier. Los resultados indican que dicho movimiento, no genera una diferencia significativa en la potencia obtenida de los electrodos seleccionados. Sin embargo, un grupo de electrodos en el area occipital si que presentaba diferencias significativas. En el futuro, se tendra en cuenta dicha informacion cuando se utilice esta tecnologia con personas que han sufrido un ACV. EnglishCerebrovascular accident (CVA) or stroke is one of the most common causes in the world which provokes motor impairment. For this reason, the scientific community is investigating ways to help those affected by stroke. This work uses brain-machine interfaces (BMI) and lower limb exoskeleton technologies to improve the rehabilitation process of CVA patients.In this way, the future patient, will be more involved in his/her rehabilitation therapy. The aim of this work is to study if the generated noise by the exoskeleton affects the EEG signals of the user and therefore, the performance of the BMI can be influenced by it. A power spectrum study of the EEG signals using Fast Fourier Transform was performed.Results show that the movement of the exoskeleton does not produce a significant difference in the power obtained of the selected electrodes. However, a group of electrodes in the occippital area do present significant differences. In the future, we will take this information into account when we use this technology with people that have suffered a CVA.

/pdf/influencia-del-exoesqueleto-de-miembro-inferior-en-senales-5eldxj5ph4.pdf

Influencia del exoesqueleto de miembro inferior en señales eeg

The combination of rehabilitation technologies and better control systems for the assistive technologies allows us to create brain-machine interfaces for motion control by using exoskeleton devices. In this work, the creation of a protocol and analysis method is developed for human decision-making decoding in tasks related to motor activity. The aims of this research is to be able to develop a real-time interface for the control of a lower-limb exoskeleton in the future. The proposed analysis tries to decode the patterns when subjects have the intention to change their current speed on their own will. 4 subjects performed the experiments obtaining results around 65%.

Jose M. Cano

Papers

Neuro-fuzzy classifier to recognize mental tasks in a BCI

Influencia del exoesqueleto de miembro inferior en señales eeg

Analysis of frequency bands and channels configuration for detecting intention of change speed through EEG