Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

In this conceptual paper, we discuss the concept of hospital of the future (HoF) and the requirements for its wireless connectivity. The HoF will be mostly wireless, connecting patients, healthcare professionals, sensors, computers and medical devices. Spaces of the HoF are first characterized in terms of communicational performance requirements. In order to fulfil the stringent requirements of future healthcare scenarios, such as enhanced performance, security, safety, privacy, and spectrum usage, we propose a flexible hybrid optical-radio wireless network to provide efficient, high-performance wireless connectivity for the HoF. We introduce the concept of connected HoF exploiting reconfigurable hybrid optical-radio networks. Such a network can be dynamically reconfigured to transmit and receive optical, radio or both signals, depending on the requirements of the application. We envisage that HoF will consist of numerous communication devices and hybrid optical-radio access points to transmit data using radio waves and visible light. Light-based communications exploit the idea of visible light communications (VLC), where solid-state luminaries, white light-emitting diodes (LEDs) provide both room illumination as well as optical wireless communications (OWC). The hybrid radio-optical communication system can be used in principle in every scenario of the HoF. In addition to the hybrid access, we also propose a reconfigurable optical-radio communications wireless body area network (WBAN), extending the conventional WBAN to more generic and highly flexible solution. As the radio spectrum is becoming more and more congested, hybrid wireless network approach is an attractive solution to use the spectrum more efficiently. The concept of HoF aims at enhancing healthcare while using hospital resources efficiently. The enormous surge in novel communication technologies such as internet of things (IoT) sensors and wireless medical communications devices could be undermined by spectral congestion, security, safety and privacy issues of radio networks. The considered solution, combining optical and radio transmission network could increase spectral efficiency, enhancing privacy while reducing patient exposure to radio frequency (RF). Parallel radio-optical communications can enhance reliability and security. We also discuss possible operation scenarios and applications that can be introduced in HoF as well as outline potential challenges.

/pdf/wireless-communications-for-the-hospital-of-the-future-4x2rzyqg2i.pdf

Wireless Communications for the Hospital of the Future: Requirements, Challenges and Solutions

Forecast of annual water consumption in 31 regions of China considering GDP and population

The classification of electroencephalogram (EEG) signals is of significant importance in brain-computer interface (BCI) systems. Aiming to achieve intelligent classification of EEG types with high accuracy, a classification methodology using sparse representation (SR) and fast compression residual convolutional neural networks (FCRes-CNNs) is proposed. In the proposed methodology, EEG waveforms of classes 1 and 2 are segmented into subsignals, and 140 experimental samples were achieved for each type of EEG signal. The common spatial patterns algorithm is used to obtain the features of the EEG signal. Subsequently, the redundant dictionary with sparse representation is constructed based on these features. Finally, the samples of the EEG types were imported into the FCRes-CNN model having fast down-sampling module and residual block structural units to be identified and classified. The datasets from BCI Competition 2005 (dataset IVa) and BCI Competition 2003 (dataset III) were used to test the performance of the proposed deep learning classifier. The classification experiments show that the recognition averaged accuracy of the proposed method is 98.82%. The experimental results show that the classification method provides better classification performance compared with sparse representation classification (SRC) method. The method can be applied successfully to BCI systems where the amount of data is large due to daily recording.

/pdf/an-intelligent-eeg-classification-methodology-based-on-2eevg21830.pdf

An Intelligent EEG Classification Methodology Based on Sparse Representation Enhanced Deep Learning Networks.

For brain computer interfaces (BCIs) research, the classification of motor imagery brain signals is a major and challenging step. Based on the traditional sparse representation classification, a classification algorithm of electroencephalogram (EEG) based on sparse representation and convolution neural network is proposed by this paper. For the EEG signal, firstly, the features of the signal are obtained through the common spatial pattern (CSP) algorithm, and then the redundant dictionary with sparse representation is constructed based on these features. The sparse representation of the EEG signal is completed and the sparse features can be obtained. Finally, the sparse features are transformed into two dimensional signals, and the convolution neural network is used to complete the classification of EEG signals. Using the dataset downloaded from the website of BCI competition III (dataset IVa), for two types of EEG signals, the experiments show that the recognition accuracy of the method is over 80, and the recognition accuracy is better than that of the traditional SRC algorithm.

EEG Classification Based on Sparse Representation and Deep Learning

Based on the advancements in pervasive healthcare our perception of healthcare is changing quite fast. For any standard existing Hospital Information System (HIS) there are several major problems that hinders automation like, fixed information point or inflexible networking mode. In our paper, Internet of Things (IOT) based platform has been used for a better automation and faster decision making related executions in the smart hospital. Hidden Markov Models (HMM) offers scope for better predictive analysis for its dynamic probabilistic nature and helped us developing a dense sensing approach based on traces of activities in a hospital. A need-based HMM-IOT hybrid has been used for the smart hospital scheme proposed here, and it continuously synchronizes itself with the current spectral analysis data of the bio-signals of the admitted patients acquired by non-invasive processes like Electroencephalogram, Electromyogram, Electrocardiogram etc. The results of the experimentation are really promising.

Hidden Markov model and Internet of Things hybrid driven smart hospital

Time-series data optimized AR/ARMA model for frugal spectrum estimation in Cognitive Radio

Non-invasive process like Electroencephalogram or Electromyogram are much preferred among various Brain-Computer Interfaces techniques. Parametric method for spectral density calculation could be beneficial for such signal due to certain advantages over non-parametric methods like avoiding side-lobe leakages. Auto-regressive and Crosscorrelative analysis of EEG or EMG signals are performed to infer their nature. Nowadays, doctors can prescribe medicines or further investigations simply based on visual analysis of the EEG or EMG graphs, where power spectral density of such signals are assumed stationary after being filtered. This paper proposes a new technique of statistically inferring the health status based on power analysis of neurophysiological signals. Modified Covariance, Burg, and/or Yule-Walker methods are used to calculate spectral densities. Individual weights from such methods are computed to infer a single electrode characteristic. Similar procedure is repeated for each electrode for concluding overall neurophysiological health status. Experimental results thus obtained are quite encouraging.

Brain computer interfacing: A spectrum estimation based neurophysiological signal interpretation

Recently Spectrum estimation has become an interesting topic for the researchers. Non-parametric methods generally do not have any knowledge about the process being observed. They also suffer from serious drawbacks like sidelobe leakages and unrealistic windowing methods. The second approach being known as parametric method overcomes these shortcomings. In parametric approach initially a suitable model is selected based on apriori knowledge about how the process is generated and then followed by estimating the parameters from the observed data. After calculation of parameters the power spectrum is estimated. In this paper we have studied thoroughly the Autoregressive method of spectrum estimation. We perform both simulation as well as real time implementations on FPGA based radio prototype board known as Wireless Open Access Research Platform (WARP) of RICE University. Various algorithms like Yule-Walker, Burg, Covariance and Modified Covariance have been studied with real time estimation of the statistical parameters by which they are described in AR technique.

Performance analysis of different autoregressive methods for spectrum estimation along with their real time implementations

A two-stage data-driven Spectrum Estimation (SE) technique has been introduced for a Cognitive Radio (CR) system involving Null-Hypothesis approach using robust Chi-Square Goodness of Fit (GoF) for...

Debashis Chakraborty

Papers

Hidden Markov model and Internet of Things hybrid driven smart hospital

Time-series data optimized AR/ARMA model for frugal spectrum estimation in Cognitive Radio

Brain computer interfacing: A spectrum estimation based neurophysiological signal interpretation

Performance analysis of different autoregressive methods for spectrum estimation along with their real time implementations

Testbed Implementation of a Scalable ARIMA Model for Spectrum Estimation in Cognitive Radio-A Null Hypothesis Approach