Search

Faults occur in electric transmission and distribution lines may accompany with a destructive phenomenon named partial discharge. If left untreated for an extended period of time, the insulation would be damaged. Technical University of Ostrava and ENET Centre utilize a cost effective, simple meter to measure the stray electric field voltage signals for covered conductors of medium voltage overhead power lines. The data could be utilized to detect faults through continuously, and economically monitoring power lines. In recent years, various unique, and complicated methods have been proposed to detect the partial discharge in the aforementioned signals. Most of them are in need of huge amount of computing and equipment support. This article is dedicated to building an effective long short-term memory neural network classifier and reducing its requirements on hardware conditions, such as storage memory, CPU, and GPU, through efficient data processing and feature extraction. The proposed method is tested on the ENET Centre’s public voltage signal data. And it is compared to various classifiers. It demonstrates light calculation, superior performance, and great practicality. Such an automated monitoring method would reduce the maintenance costs of power systems. And it would efficiently prevents power outages.

Glaucoma is a chronic eye disease, which causes gradual vision loss and eventually blindness. Accurate glaucoma screening at early stage is critical to mitigate its aggravation. Extracting high-quality features are critical in training of classification models. In this paper, we propose a deep ensemble network with attention mechanism that detects glaucoma using optic nerve head stereo images. The network consists of two main sub-components, a deep Convolutional Neural Network that obtains global information and an Attention-Guided Network that localizes optic disc while maintaining beneficial information from other image regions. Both images in a stereo pair are fed into these sub-components, the outputs are fused together to generate the final prediction result. Abundant image features from different views and regions are being extracted, providing compensation when one of the stereo images is of poor quality. The attention-based localization method is trained in a weakly-supervised manner and only image-level annotation is required, which avoids expensive segmentation labelling. Results from real patient images show that our approach increases recall (sensitivity) from the state-of-the-art 88.89% to 95.48%, while maintaining precision and performance stability. The marked reduction in false-negative rate can significantly enhance the chance of successful early diagnosis of glaucoma.

This paper describes an open access electroencephalography (EEG) data set for multitasking mental workload activity induced by a single-session simultaneous capacity (SIMKAP) experiment with 48 subjects. To validate the database, EEG spectral activity was evaluated with EEGLAB and the significant channels and activities for the experiment are highlighted. Classification performance was evaluated by training a support vector regression model on selected features from neighborhood component analysis based on a nine-point workload rating scale. With a reduced feature dimension, 69% classification accuracy was obtained for 3 identified workload levels from the rating scale with Cohen's kappa of 0.46. Accurate discrimination of mental workload is a desirable outcome in the field of operator performance analysis and BCI development; thus, we hope that our provided database and analyses can contribute to future investigations in this research field.

We study the three-dimensional turbulent Kolmogorov flow, i.e. the Navier-Stokes equations forced by a low-single-wave-number sinusoidal force in a periodic domain, by means of direct numerical simulations. This classical model system is a realization of anisotropic and non-homogeneous hydrodynamic turbulence. Boussinesq's eddy viscosity linear relation is checked and found to be approximately valid over half of the system volume. A more general nonlinear quadratic Reynolds stress development is proposed and its parameters estimated at varying the Taylor scale-based Reynolds number in the flow up to the value 200. The case of a forcing with a different shape, here chosen Gaussian, is considered and the differences with the sinusoidal forcing are emphasized., Comment: 11 pages

Nowadays huge amount of crowd data captured by surveillance cameras gets shared publicly in the form of images or videos through television or the internet. Although many of these videos are meant to provide public security, they also lead to a widespread concern towards privacy protection, since a lot of personal information about subjects gets revealed through this video/image data. Hence, "de-identifying" people (i.e., obscuring identity information) captured by the surveillance cameras is of utmost importance for providing privacy along with security. Traditional identity obfuscation techniques such as blurring, warping, and filtering lead to the loss of vital non-biometric information. More recent k-same-based as well as generative model-based de-identification techniques eliminate the above problem to a certain extent. Still, the visual quality of the generated images produced by these methods is not realistic. Also, these approaches are unable to maintain structural integrity and cannot preserve the required non-biometric information at a high resolution. As an improvement, in this paper, we propose a new network termed as EPD-Net and train it with suitable loss functions to maximize the emotion similarity and minimize the identity similarity. Experimental results verify the effectiveness of our approach and its superiority over other popular face de-identification techniques.

The statistical properties of species undergoing chemical reactions in a turbulent environment are studied. We focus on the case of reversible multi-component reactions of second and higher orders, in a condition close to chemical equilibrium sustained by random large-scale reactant sources, while the turbulent flow is highly developed. In such a state a competition exists between the chemical reaction that tends to dump reactant concentration fluctuations and enhance their correlation intensity and the turbulent mixing that on the contrary increases fluctuations and remove relative correlations. We show that a unique control parameter, the Damkh\"{o}ler number ($Da_\theta$) that can be constructed from the scalar Taylor micro-scale, the reactant diffusivity and the reaction rate characterises the functional dependence of fluctuations and correlations in a variety of conditions, i.e., at changing the reaction order, the Reynolds and the Schmidt numbers. The larger is such a Damkh\"{o}ler number the more depleted are the scalar fluctuations as compared to the fluctuations of a passive scalar field in the same conditions, and vice-versa the more intense are the correlations. A saturation in this behaviour is observed beyond $Da_\theta \simeq \mathcal{O}(10)$. We provide an analytical prediction for this phenomenon which is in excellent agreement with direct numerical simulation results., Comment: 17 pages , 9 figures

This paper presents a method based on a population in which the parameters of individuals can be processed by operators from various population-based algorithms. The mechanism of selecting operators is based on the introduction of an additional binary parameters vector located in each individual, on the basis of which it is decided which operators are to be used to modify individuals’ parameters. Thus, in the proposed approach, many operators can be used simultaneously for this purpose. As part of the paper various methods of initializing binary parameters, various population sizes, and their impact on the operation of the algorithm were tested. The simulation was carried out on a well-known set of benchmark functions.

An affective brain-computer interface (aBCI) is a direct communication pathway between human brain and computer, via which the computer tries to recognize the affective states of its user and respond accordingly. As aBCI introduces personal affective factors into human-computer interaction, it could potentially enrich the user’s experience during the interaction. Successful emotion recognition plays a key role in such a system. The state-of-the-art aBCIs leverage machine learning techniques which consist in acquiring affective electroencephalogram (EEG) signals from the user and calibrating the classifier to the affective patterns of the user. Many studies have reported satisfactory recognition accuracy using this paradigm. However, affective neural patterns are volatile over time even for the same subject. The recognition accuracy cannot be maintained if the usage of aBCI prolongs without recalibration. Existing studies have overlooked the performance evaluation of aBCI during long-term use. In this paper, we propose SAFE—an EEG dataset for stable affective feature selection. The dataset includes multiple recording sessions spanning across several days for each subject. Multiple sessions across different days were recorded so that the long-term recognition performance of aBCI can be evaluated. Based on this dataset, we demonstrate that the recognition accuracy of aBCIs deteriorates when re-calibration is ruled out during long-term usage. Then, we propose a stable feature selection method to choose the most stable affective features, for mitigating the accuracy deterioration to a lesser extent and maximizing the aBCI performance in the long run. We invite other researchers to test the performance of their aBCI algorithms on this dataset, and especially to evaluate the long-term performance of their methods.

The rapid advancements in machine learning, graphics processing technologies and the availability of medical imaging data have led to a rapid increase in the use of deep learning models in the medical domain. This was exacerbated by the rapid advancements in convolutional neural network (CNN) based architectures, which were adopted by the medical imaging community to assist clinicians in disease diagnosis. Since the grand success of AlexNet in 2012, CNNs have been increasingly used in medical image analysis to improve the efficiency of human clinicians. In recent years, three-dimensional (3D) CNNs have been employed for the analysis of medical images. In this paper, we trace the history of how the 3D CNN was developed from its machine learning roots, we provide a brief mathematical description of 3D CNN and provide the preprocessing steps required for medical images before feeding them to 3D CNNs. We review the significant research in the field of 3D medical imaging analysis using 3D CNNs (and its variants) in different medical areas such as classification, segmentation, detection and localization. We conclude by discussing the challenges associated with the use of 3D CNNs in the medical imaging domain (and the use of deep learning models in general) and possible future trends in the field., Comment: Published in Sensors Journal (https://www.mdpi.com/1424-8220/20/18/5097)

Extracting intra- and inter-subject parameters from Electroencephalogram (EEG) representing different Situation Awareness (SA) status is a critical challenge for objective SA recognition. Most of the existing work focuses on the subject-dependent classification that applies power spectrum density (PSD) features. In this paper, we propose a novel spectral-spatial (S-S) model for cross-subject fatigue-related SA recognition. The S-S model not only considers the biological topology across different brain regions to capture both local and global relations among different EEG channels, but also extracts spectral features for each EEG channel. Specifically, we firstly model the topological structure of EEG channels via an adjacency matrix which is built based on the Euclidean distance between EEG channels. Then, the graph convolution operation is employed to perform the neighbourhood aggregation for extracting spatial features. We test our model on a public dataset collected during driver's task performance. The subject-independent performance of the model is explored. Results demonstrate (1) the superior performance of our model compared with the state-of-the-art models on SA recognition from EEG signals. Specifically, our S-S model achieves 70.6% accuracy which is higher than traditional machine learning methods by 2.7%-6.8% and deep learning methods by 10.3%-11.6%; (2) EEG signal at the occipital region can better reflect the change of SA. National Research Foundation (NRF) Accepted version This research is supported by the National Research Foundation, Singapore under its International Research Centres in Singapore Funding Initiative. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of National Research Foundation, Singapore.

Learning deep neural networks requires huge hardware resources and takes a long time. This is due to the need to process huge data sets multiple times. One type of neural networks that are particularly useful in practice are denoising autoencoders. It is, therefore, necessary to create new algorithms that reduce the training time for this type of networks. In this work, we propose a method that, in contrast to the classical approach, where each data element is repeatedly processed by the network, is focused on processing only the most difficult to analyze elements. In the learning process, subsequent data may lose their significance and others may become important. Therefore, an additional algorithm has been used to detect such changes. The method draws inspiration from boosting algorithms and drift detectors.

In turbulent combustion simulations, the flow structure at the unresolved scale level needs to be reasonably modeled. Following the idea of turbulent flamelet equation for the non-premixed flame case, which was derived based on the filtered governing equations (L. Wang, Combust. Flame 175 , 259 (2017)), the scalar dissipation term for tabulation can be directly computed from the resolved flowing quantities, instead of solving species transport equations. Therefore, the challenging source term closure for the scalar dissipation or any assumed probability density functions can be avoided; meanwhile the chemical sources are closed by scaling relations. The general principles are discussed in the context of large eddy simulation with case validation. The new model predictions of the bluff-body flame show sufficiently improved results, compared with these from the classic progress-variable approach.

The present work focuses on the flame–wall interaction (FWI) based on direct numerical simulations (DNS) of a head-on premixed flame quenching configuration at the statistically stationary state. The effects of FWI on the turbulent flame temperature, wall heat flux, flame dynamics and flow structures were investigated. In turbulent head-on quenching, particularly for high turbulence intensity, the distorted flames generally consist of the head-on flame part and the entrained flame part. The flame properties are jointly influenced by turbulence, heat generation from chemical reactions and heat loss to the cold wall boundary. For the present FWI configuration, as the wall is approached, the ‘influence zone’ can be identified as the region within which the flame temperature, scalar gradient and flame dilatation start to decrease, whereas the wall heat flux tends to increase. As the distance to the wall drops below the flame-quenching distance, approximately where the wall heat flux reaches its maximum value, chemical reactions become negligibly weak inside the ‘quenching zone’. A simplified counter-flow model is also proposed. With the reasonably proposed relation between the flame speed and the flame temperature, the model solutions match well with the DNS results, both qualitatively and quantitatively. Moreover, near-wall statistics of some important flame properties, including the flame dilatation, reaction progress variable gradient, tangential strain rate and curvature were analysed in detail under different wall boundary conditions.

Flame–wall interaction (FWI) in premixed turbulent combustion has been analysed based on a counter-flow like configuration at the statistically stationary state. For the present configuration, the ...

Direct numerical simulations (DNS) of oblique wall quenching of a turbulent V-flame and head-on quenching (HOQ) of a statistically planar flame by isothermal inert walls have been utilized to analy...

The present work focuses on the geometrical features of a zero-pressure-gradient turbulent boundary layer based on vectorline segment analysis. In a turbulent vector field, tracing from any non-singular point, along either the vector or the inverse direction, one will reach a local extremum of the vector magnitude. The vectorline between the two local extrema is defined as the vectorline segment corresponding to the given spatial point. Specifically the vectorline segment can be the streamline segment for the velocity vector case, or the vorticity line segment for the vorticity vector case. Such a quantitatively defined and space-filling vectorline segment structure reflects the natural vectorial topology. Because of inhomogeneity in the wall-normal direction, vectorline segments corresponding to the grid points at specified wall-normal distances are sampled for statistics. For streamline segments, the probability density function (p.d.f.) of the normalized segment length in different flow regions matches a model solution, and for vorticity line segments such a p.d.f. in the log-law region and beyond matches the same model solution, which indicates some universality of different flow regions and different vector field structures. Typically the joint p.d.f. of the characteristic parameters of streamline segments presents clear asymmetry, which is closely related to the skewness of the velocity derivative. Moreover, the orientation statistics of vectorline segments, characterized by the coordinate difference between the segment starting point and ending point, have been provided to quantify the flow anisotropy.

The three-dimensional Navier-Stokes characteristic boundary conditions (3D-NSCBC), although physically reasonable and popular in many applications, may encounter the instability problem in simulating complex flows, especially for large Reynolds number reactive turbulence where locally the strong reversed flow appears at the outflow boundary surfaces. In the present work, a revised 3D-NSCBC strategy is proposed based on the kinematic relation in different moving coordinate systems. Following this strategy, a systematic formulation is presented for the outflow surface with local reversed flow and can be easily extended to the coupled edge and corner boundaries. Direct numerical simulation (DNS) tests of flow with different turbulence intensities are carried out. Compared with the conventional 3D-NSCBC, the newly proposed method exhibits satisfactory performance to confine numerical instability in the strong reversed flow region. The results confirm the robustness and effectiveness of this newly proposed algorithm.

The tremendous success of machine learning algorithms at image recognition tasks in recent years intersects with a time of dramatically increased use of electronic medical records and diagnostic imaging. This review introduces the machine learning algorithms as applied to medical image analysis, focusing on convolutional neural networks, and emphasizing clinical aspects of the field. The advantage of machine learning in an era of medical big data is that significant hierarchal relationships within the data can be discovered algorithmically without laborious hand-crafting of features. We cover key research areas and applications of medical image classification, localization, detection, segmentation, and registration. We conclude by discussing research obstacles, emerging trends, and possible future directions. Published version