Anuradha Saha

EEG Analysis for Olfactory Perceptual-Ability Measurement Using a Recurrent Neural Classifier

A recurrent neural network model is designed to classify (pretrained) aromatic stimuli and discriminate noisy stimuli of both similar and different genres, using EEG analysis of the experimental subjects. The design involves determining the weights of the selected recurrent dynamics so that for a given base stimulus, the dynamics converges to one of several optima (local attractors) on the given Lyapunov energy surface. Experiments undertaken reveal that for small noise amplitude below a selected threshold, the dynamics essentially converges to fixed stable attractor. However, with a slight increase in noise amplitude above the selected threshold, the local attractor of the dynamics shifts in the neighborhood of the attractor obtained for the noise-free standard stimuli. The other important issues undertaken in this paper include a novel algorithm for evolutionary feature selection and data-point reduction from multiple experimental EEG trials using principal component analysis. The confusion matrices constructed from experimental results show a marked improvement in classification accuracy in the presence of data point reduction algorithm. Statistical tests undertaken indicate that the proposed recurrent classifier outperforms its competitors with classification accuracy as the comparator. The importance of this paper is illustrated with a tea-taster selection problem, where an olfactory perceptual-ability measure is used to rank the tasters.

Olfaction recognition by EEG analysis using differential evolution induced Hopfield neural net

The paper proposes a novel approach to recognize smell stimuli from the electroencephalogram (EEG) signals acquired during the period of inhalation. The main contribution of the paper lies in feature selection by an evolutionary algorithm and pattern classification by Differential Evolution induced Hopfield neural network. One additional merit of the work lies in data point reduction by Principal component analysis. Experiments undertaken on 25 subjects with 10 smell stimuli indicate that the proposed scheme of feature selection, data point reduction and classification outperforms the traditional approach by a wide margin. Experimental results confirm that the smell stimuli excites the pre frontal lobe of the human brain and is responsible for a special type of brain rhythms (EEG signal) in alpha-band, theta-band and delta-band.

EEG analysis for cognitive failure detection in driving using neuro-evolutionary synergism

The paper proposes a solution to reduce accidents in driving by alarming specific cognitive failures to the drivers on occurrence of the failures. Three different types of cognitive failures that might occur due to lapse of i) visual alertness, ii) cognitive planning and iii) motor execution are studied, and suitable classifiers have been employed to classify these failures. Force exerted by the driver during turning or sudden tracking is also measured to detect his level of cognitive load during significant changes in the driving environment. Recurrent neural networks are introduced here as classifiers to decode cognitive tasks performed by the driver from his acquired EEG. For each recurrent neural net, we use a Lyapunov energy surface the minima of which denote the cognitive tasks during one of three cognitive activities mentioned above. Given the features of the measured EEG for a cognitive tasks, the recurrent net converges to one of several optima, describing a specific cognitive failure. Experimental results obtained by employing a driving simulator and an EEG system are encouraging.

EEG classification to determine the degree of pleasure levels in touch-perception of human subjects

This paper introduces a novel approach to examine the scope of touch perception as a possible modality of treatment of patients suffering from certain mental disorder using a Radial Basis function induced Back Propagation Neural Network. Experiments are designed to understand the perceptual difference of schizophrenic patients from normal and healthy subjects with respect to four different touch classes, including soft touch, rubbing, massaging and embracing and their three typical subjective responses such as pleasant, acceptable, and unpleasant. Experiments undertaken indicate that that the frontal part of the scalp map of healthy subjects carry more blood during touch perception than those obtained for the schizophrenic patients. Further, for normal subjects and schizophrenic patients, the average percentage accuracy in classification of all the three classes including pleasant, acceptable or unpleasant is comparable with their respective oral responses. In addition, for schizophrenic patients, the percentage accuracy for acceptable class is very poor of the order of below 10%, which for normal subjects is quite high (46%). Performance analysis reveals that the proposed classifier outperforms its competitors with respect to classification accuracy in all the above three classes. A well known statistical test confirms that the proposed classifier outperforms all its competitors along with principal component analysis as feature selector by a large margin.

An EEG-based cognitive failure detection in driving using two-stage motor intension classifier

Cognitive failure of drivers during driving becomes nowadays the most alarming issue for traffic fatalities, and hence, it is important to intend correct motor actions during driving to avoid accidents. This paper proposes a novel two-stage motor intension classifier to detect the correct motor planning of the drivers during car driving. Common spatial pattern filters are introduced here for feature extraction as well as artifact removal from the raw electroencephalographic signals. Experiments undertaken further confirm that the basic motor intensions including i) steering control, ii) accelerator, iii) brake, and iv) no action and as well as the sub-classification of individual control are best classified by Support vector machine with polynomial kernel from a list of standard classifiers with an average classification accuracy of 85.70%.

Data-point and feature selection of motor imagery EEG signals for neural classification of cognitive tasks in car-driving

This paper proposes novel algorithms for data-point and feature selection of motor imagery electroencephalographic signals for classifying motor plannings involved in car- driving including braking, acceleration, left steering control and right steering control. Variants of neural network classifiers such as linear support vector machines, and kernel-based support vector machines including radial basis function kernel, polynomial kernel and hyperbolic kernel have been applied to classify the various cognitive tasks. Experimental finding reveals that the proposed data-point and feature selection technique altogether provides better classification accuracies (more than 88%) for all cognitive tasks in comparison with using factor analysis for data-point reduction and feature selection. It is also observed that power spectral density and discrete wavelet transform features are selected among the list of electroencephalographic features for holding the top two rank values for cognitive task classification during car-driving. From the experimental result, it is confirmed that support vector machines with radial basis function along with power spectral density outperforms the remaining feature-classifier pairs in terms of average classification accuracy.

A type-2 fuzzy approach towards cognitive load detection using fNIRS signals

The main notion of this paper is to identify the cognitive load during a mental arithmetic task experiment using fNIRS signals. The first objective is to classify the difficulty level and the state of inactivity during the given task. To identify the classes, the feature vectors have to undergo all the possible steps of a pattern classification problem. In this paper, we have developed a novel Feature Selection technique to reduce the dimension of the feature vectors by omitting the redundant features. For this purpose, an objective function depending upon the class density or likelihood functions is optimized using the well-known Differential Evolution algorithm. General type-2 fuzzy classifier is used for subsequent classification step. The proposed Feature Selection technique gives a satisfactory accuracy results over principal component analysis. The fuzzy classifier outperforms the other well-known classifier like support vector machine, k-nearest neighborhood. Experimental result reveals that the proposed likelihood-based FS induced type 2 fuzzy classifier attains the highest classification accuracy (above 90% in each case) as compared to its standard competitors. The load of a subject undergoing the experiment is measured at a particular class relying upon the mean type- 1 fuzzy value of all feature entities. A clear discrimination in concentration level from 16 channels has been observed for each distinct feature set.

Decoding of motor imagery potentials in driving using DE-induced fuzzy-neural classifier

This paper presents a novel feature selection and fuzzy-neural classification scheme to decode motor imagery signals during driving. To perform this, we would consider the fuzziness involved in sudden left bent, where the driver is supposed to take sudden 90o left turn during acceleration. This requires classification of motor imagery signals during acceleration and steering left control. The fuzzy-recurrent neural network classifier offers better performance using proposed differential evolution-induced feature selection technique, when compared with principal component analysis in such situation and provides the highest classification accuracy of 98.472%. In addition, false classification rate/misclassification rate is also found much higher when using principal component analysis instead of proposed differential evolution-induced feature selection algorithm. The performance of the proposed differential evolution-induced fuzzy recurrent neural network classifier has been compared with a list of standard classifiers including linear support vector machines, k-nearest neighbor and support vector machines with radial basis function kernel, where fuzzy-recurrent neural network classifier outperforms its competitors with an average classification accuracy of 95.472% and 95.647 for steering left and acceleration motor intensions respectively.

EEG Analysis for Cognitive Failure Detection in Driving Using Type-2 Fuzzy Classifiers

This paper aims at detecting online cognitive failures in driving by decoding the electroencephalography (EEG) signals acquired during visual alertness, motor planning and motor-execution phases of the driver. Visual alertness of the driver is detected by classifying the preprocessed EEG signals obtained from his prefrontal and frontal lobes into two classes: alert and nonalert. Motor planning performed by the driver using the preprocessed parietal signals is classified into four classes: braking, acceleration, steering control, and no operation. Cognitive failures in motor planning are determined by comparing the classified motor-planning class of the driver with the ground truth class obtained from the copilot through a hand-held rotary switch. Lastly, failure in motor execution is detected, when the time delay between the onset of motor imagination and the electromyogram response exceeds a predefined duration. The most important aspect of the present research lies in cognitive failure classification during the planning phase. The complexity in subjective plan classification arises due to possible overlap of signal features involved in braking, acceleration, and steering control. A specialized interval/general type-2 fuzzy set induced neural classifier is employed to eliminate the uncertainty in classification of motor planning. Experiments undertaken reveal that the proposed neuro-fuzzy classifier outperforms traditional techniques in presence of external disturbances to the driver. Decoding of visual alertness and motor execution are performed with kernelized support vector machine classifiers. An analysis reveals that at a driving speed of 64 km/h, the lead time is more than 600 ms, which offer a safe distance of 10.66 m.

EEG based gesture mimicking by an artificial limb using cascade-correlation learning architecture

Patients with prosthesis defects find it is very difficult to perform day-to-day basic tasks which involve employment of their limbs. This motivates us to develop a system where an artificial limb is employed to mimic the arm gestures of the patients for assisting them. Towards developing this system, we have taken the help from the electroencephalography (EEG) signals acquired from the brain of the patients to build a bypass network (BPN) to direct the artificial limb. Since difficulties are already present in the arm movements of the patients (here subjects), thus only gestures of those subjects are not sufficient to build the proposed system. This research finds tremendous applications in rehabilitative aid for the disable persons. To concretize our goal we have developed an experimental setup, where the target subject (for training phase healthy subjects are taken into account) is asked to catch a ball while his/her brain (occipital, parietal and motor cortex) signals using EEG acquisition device and body gestures using Kinect sensor are simultaneously acquired. These data are mapped using four cascade-correlation learning architecture (CCLA) to train artificial limb (we have used Jaco robot arm) to move accordingly. Utilizing the mapping results obtained from these four CCLAs, a BPN is developed. When a rehabilitative patient is unable to catch the ball, then in that scenario, the artificial limb is helpful for assisting the patient to catch the ball with a high accuracy of 85.65%. The proposed system can be implemented not only for ball catching experiment but also in several applications where an artificial limb needs to perform a locomotive task based on EEG and body gesture.