Search

This work proposes a novel multivariate-multiscale approach for computing the spectral and temporal entropies from the multichannel electroencephalogram (EEG) signal. This facilitates the recognition of three human emotions: positive, neutral, and negative. The proposed approach is based on the application of the Fourier-Bessel series expansion based empirical wavelet transform (FBSE-EWT). We have extended the existing FBSE-EWT method for multichannel signals and derived FBSE-EWT based multivariate Hilbert marginal spectrum (MHMS) for computing spectral Shannon and K-nearest neighbor (K-NN) entropies. The multivariate FBSE-EWT decomposes the multichannel EEG signals into narrow band subband signals. The multiscaling operation adapted in the spectral domain is based on the selection of successive joint instantaneous amplitude and frequency functions of the subband signals. On the other hand, the time domain multiscale K-NN entropy is computed from the cumulatively added multidimensional subband signals. The extracted spectral and temporal entropy features are smoothed and fed to the sparse autoencoder based random forest (ARF) classifier architecture for emotion classification. The proposed approach is tested using multichannel EEG signals available in a public database (SJTU emotion EEG dataset (SEED)). The bivariate EEG signals from different channel pairs with distinct spatial locations over the scalp are considered as input to our proposed system. The obtained overall classification accuracy of 94.4% reveals that the proposed approach is useful in classifying human emotions. The method is also validated using DREAMER emotion EEG public database. The method outperforms the existing state-of-the-art methods evaluated in these databases.

According to various worldwide statistics, most car accidents occur solely due to human error. The person driving a car needs to be alert, especially when travelling through high traffic volumes that permit high-speed transit since a slight distraction can cause a fatal accident. Even though semiautomated checks, such as speed detecting cameras and speed barriers, are deployed, controlling human errors is an arduous task. The key causes of driver’s distraction include drunken driving, conversing with co-passengers, fatigue, and operating gadgets while driving. If these distractions are accurately predicted, the drivers can be alerted through an alarm system. Further, this research develops a deep convolutional neural network (deep CNN) models for predicting the reason behind the driver’s distraction. The deep CNN models are trained using numerous images of distracted drivers. The performance of deepCNNmodels, namely theVGG16,ResNet, and Xception network, is assessed based on the evaluation metrics, such as the precision score, the recall/sensitivity score, the F1 score, and the specificity score. The ResNet model outperformed all other models as the best detection model for predicting and accurately determining the drivers’ activities.

In an uncontrolled environment, many of the face detection algorithms lack robustness due to their design. The present research on unconstrained face detection is focused on handcrafted and visual features in isolation. We propose a novel approach to use handcrafted as well as visual features together for improvement in face detection to achieve robustness. The algorithm uses a side-view face detector, which divides the problem space into two: side view face detection and frontal face detection. For frontal faces, Discrete Wavelet Transform (DWT) followed by the encoding of Eyes like landmarks (ELL) pixels is proposed in this work. A Human trait that helps to make decisions even better when they are taken together with the help of more than one decision makers is modeled in this work. To achieve this, a combination of handcrafted and visual features is used. Further to improve classification and to provide a better decision, a faster second stage classification scheme is introduced. The result shows an improvement when handcrafted and visual features are combined instead of using them separately.

Monitoring the environment plays a significant role in the evolution of a smart city with enhanced safety, comfort, and security. One promising technology for smart environment monitoring is the radar-based moving object tracking and classification. Radar-based technology is known to provide better solutions compared to vision-based systems as the latter is prone to the adverse effects of lighting and weather conditions. Radars operating in mm-wave range are preferred due to small size, good angular resolution, small apertures, and low cost. The signature waveform received from the moving object is used to estimate the range and the velocity of multiple targets based on Doppler drift. Also, the micro-Doppler effect on the return signal that occurs due to the micromotion dynamics such as vibrations or rotations can be used to identify the specific type of the target. This book chapter provides an overview of different types of radars and their respective signal processing aspects used for detecting the human and animal activity. Moreover, this chapter explains the deep learning algorithms used for detection and classification of human and animal activity. Thus the intelligent radar systems add greater value to the smart environment compared to other non-radar methods.

Due to its interpretability and intuitiveness, survival tree based analysis is a powerful Artificial Intelligence method for modelling longitudinal survival data, its relationship with covariates and the interrelationship between covariates. Furthermore, it is being increasingly used for a range of applications including clustering, prognostication and classification. Phase type survival tree methods have been demonstrated to have important applications, including clustering patients into clinically meaningful groups, patient pathway prognostication and forecasting bed requirements. In this article, we critically investigate and assess several selection information criteria with regards to their suitability and limitations when used as splitting criteria in phase type survival tree construction. As shown in Table 12, the results of this analysis are compared and discussed. Furthermore, a text mining approach is utilised to further assess correlations, which have been extracted from hospital data, between the three underlying diseases and the two different types of population groups, namely age and gender groups. Its aim is to provide further investigative tools. In fact, due to its ability to analyse large volumes of textual data, text mining can provide a useful approach to this research area.

Emotion recognition from electroencephalography (EEG) signals is a very cost-effective method to monitor the general well-being of an individual, an employee of an organization, or to cater to mental health patients. But it is a challenging task owing to the non-stationarity of the EEG signals. Extracting relevant features through signal processing techniques that can be used to classify patterns in the EEG signal leading to different emotions is a difficult task. A dataset for emotion analysis with physiological signals DEAP [1] consists of EEG signals of 32 participants are categorized on the quadrant of valence, arousal, dominance, and liking, which signifies how they are associated with different emotions. In this paper, an efficient classifier for emotion/quadrant recognition from EEG signals with exceptional accuracy is presented. The data preprocessing strategy adapted is Empirical Mode Decomposition (EMD), which decomposes the signals into several oscillatory Intrinsic Mode Functions (IMF). Features are extracted from Second order difference plots (SODP) are area, mean, and central tendency measure of the elliptical region. Wilcoxson Test was performed to ensure the statistical significance of the extracted features with p

The aim of this paper is to demonstrate that small dataset can be used in machine learning for seizure monitoring and detection using smart organization of multichannel EEG sensor data. This reduces training time and improves computational performance in terms of space and time complexities on hardware implementations. The proposed approach has been tested and validated using CHB-MIT dataset containing EEG recordings of 24 clinically verified seizure and non-seizure pediatric patients. The predictability is discussed in terms of the latency and the required length of data for the proposed approach over the state-of-the-art method in the field of EEG-based seizure prediction.

This paper investigates the application of spectral graph decomposition method in improving the discriminative quality of spectral components of the common spatial pattern for seizure classification. The aim is to improve the variance between the normal and abnormal EEG patterns in feature extraction process. The affinity matrix of the graph spectral decomposition that derive the Laplacian matrix encodes the dataset of the CSP covariance matrix as a correlation instead of Euclidean (or Minkowski) distance on Gaussian kernel function. The feature vector containing pattern of abnormality is sorted in order of the magnitude of their simple statistical mean values. The results obtained show that the CSP-Spectral Graph Decompsotion approach seems to provide a better discriminative features than CSP feature extraction process.

Sequential pattern mining has been a popular data mining technique for extracting useful information from large databases and has successfully been used for numerous industrial and commercial problems. This paper presents a new mathematical modelling application to healthcare, providing important information to health service managers and policy makers to help them identify sequential patterns which require attention for efficiently managing scarce healthcare resources and developing effective healthcare management policies. In healthcare, these sequential patterns are analogous to the patient pathways. We present a non-homogeneous Markov model for identifying not only patient pathways which have high probability but also for identifying pathways which incur high cost or time. In order to have a more realistic model, we also consider time-dependent covariates and their impact on the pathways. An algorithm based on branch and bound global optimization is presented which can efficiently extract a required number of such patient pathways of interest. The approach is illustrated using historical data on geriatric patients from an administrative database of a London hospital.

Activity recognition is an important component of patient management in smart homes where high level activities can be learned from low level sensor data. Such activity recognition utilises sensor ID, task order and time of activation to learn about patient behavior, detect anomalies and provide prompts or other interventions. In this paper we use the sensor activation times to calculate durations and then investigate several model-based clustering approaches with a view to discretising the duration data and using such data to improve activity prediction. We explore several popular approaches to characterising such duration data, namely Coxian phase type distributions and Gaussian mixture distributions. We then show how we can utilise the learned clustering components for discretisation. Finally we use simulated data, based on a real smart kitchen deployment, to compare these approaches and evaluate the discretisation results with regard to activity prediction.