Search

This paper deals with the problem of controller design for normal nonlinear switched systems with unknown dynamics. The situation assumes that the nonlinear component of the system dynamics is unknown and there are extra uncertain terms and external perturbations affecting the system responses. Unlike previous methods that usually apply artificial neural networks to obtain an approximate model for an unknown dynamics, we use simple adaptive laws to surmount such fluctuations as well as to circumvent the steady state errors in the approaches provided by neural networks or fuzzy logics. No information on the bounds and parameters of the uncertain parts is needed. We design a twofold sliding mode control methodology to stabilize the whole switched system with arbitrary switching signals. To handle the consequences of the inherent chattering, a smooth integral type control signal is constructed. It is proved that once the system states attain the developed sliding manifold, the system becomes insensitive to both the lumped uncertainties and the dynamics of the switched system. The common assumption of a known Lyapunov functions for the subsystems is relaxed and the derived adaptive approach provides nonconservative circumstances for guaranteeing the global stability of the equilibrium state. The theoretical results of this paper are generalized for the canonical nonlinear systems with mismatched uncertainties and a novel adaptive sliding manifold is proposed to suppress the effects of the system mismatching uncertain parts. A comparative computer simulation is developed to examine the effective performance of the introduced controllers. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs), including convolutional neural networks (CNNs) and residual networks (ResNets) models, are able to learn abstract representations from the input data by considering a deep hierarchy of layers that perform advanced feature extraction. The combination of these models with visual attention techniques can assist with the identification of the most representative parts of the data from a visual standpoint, obtained through more detailed filtering of the features extracted by the operational layers of the network. This is of significant interest for analyzing remotely sensed hyperspectral images (HSIs), characterized by their very high spectral dimensionality. However, few efforts have been conducted in the literature in order to adapt visual attention methods to remotely sensed HSI data analysis. In this paper, we introduce a new visual attention-driven technique for the HSI classification. Specifically, we incorporate attention mechanisms to a ResNet in order to better characterize the spectral–spatial information contained in the data. Our newly proposed method calculates a mask that is applied to the features obtained by the network in order to identify the most desirable ones for classification purposes. Our experiments, conducted using four widely used HSI data sets, reveal that the proposed deep attention model provides competitive advantages in terms of classification accuracy when compared to other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

This paper focuses on the shingled magnetic recording (SMR) as the recording system for further high-density hard disk drives. Previously, we applied a log-likelihood ratio (LLR) modulator to the low-density parity check coding and iterative decoding system in SMR to improve the decoding performance. In this paper, we propose a new LLR modulator using neural network. Furthermore, we clarify that the proposed LLR modulator using neural network realizes effective iterative decoding in a read/write channel with pattern-dependent medium noise. [ABSTRACT FROM AUTHOR]

This paper investigates the feasibility of novel data-driven deep learning (DL) models to predict the solution of Maxwell’s equations for low-frequency electromagnetic (EM) devices. With ground truth (empirical evidence) data being generated from a finite-element analysis solver, a deep convolutional neural network is trained in a supervised manner to learn a mapping for magnetic field distribution for topologies of different complexities of geometry, material, and excitation, including a simple coil, a transformer, and a permanent magnet motor. Preliminary experiments show DL model predictions in close agreement with the ground truth. A probabilistic model is introduced to improve the accuracy and to quantify the uncertainty in the prediction, based on Monte Carlo dropout. This paper establishes a basis for a fast and generalizable data-driven model used in the analysis, design, and optimization of EM devices. [ABSTRACT FROM AUTHOR]

In this paper, a model identification method based on a long short-term memory (LSTM) neural network composed of a network structure and training algorithm is used to build a thermal field model that accurately simulates the crystal growth process. The support vector machine (SVM) approach is then adopted to identify model order and lag to determine network input and to improve precision. The thermal field model reflecting the growth process in the Czochralski crystal furnace is simulated. Experimental results and comparative analysis results both suggest that the method proposed by this paper can build an efficient thermal field model which outperforms other methods in terms of precision. [ABSTRACT FROM AUTHOR]

Linear spectral unmixing is the practice of decomposing the mixed pixel into a linear combination of the constituent endmembers and the estimated abundances. This paper focuses on unsupervised spectral unmixing where the endmembers are unknown a priori. Conventional approaches use either geometrical- or statistical-based approaches. In this paper, we address the challenges of spectral unmixing with unsupervised deep learning models, in specific, the autoencoder models, where the decoder serves as the endmembers and the hidden layer output serves as the abundances. In several recent attempts, part-based autoencoders have been designed to solve the unsupervised spectral unmixing problem. However, the performance has not been satisfactory. In this paper, we first discuss some important findings we make on issues with part-based autoencoders. By proof of counterexample, we show that all existing part-based autoencoder networks with nonnegative and tied encoder and decoder are inherently defective by making these inappropriate assumptions on the network structure. As a result, they are not suitable for solving the spectral unmixing problem. We propose a so-called untied denoising autoencoder with sparsity, in which the encoder and decoder of the network are independent, and only the decoder of the network is enforced to be nonnegative. Furthermore, we make two critical additions to the network design. First, since denoising is an essential step for spectral unmixing, we propose to incorporate the denoising capacity into the network optimization in the format of a denoising constraint rather than cascading another denoising preprocessor in order to avoid the introduction of additional reconstruction error. Second, to be more robust to the inaccurate estimation of a number of endmembers, we adopt an $l_{21}$ -norm on the encoder of the network to reduce the redundant endmembers while decreasing the reconstruction error simultaneously. The experimental results demonstrate that the proposed approach outperforms several state-of-the-art methods, especially for highly noisy data. [ABSTRACT FROM AUTHOR]

The earth’s magnetosphere is an intricate input–output system, where the inputs are the solar wind parameters and the output measure are the geomagnetic. The disturbance storm time (DST) index is one such quantity measure for the intensity of disturbance in the geomagnetic field due to the magnetic storm. The geomagnetic activities have catastrophic effects on several communication networks and harmful for biological livings. Therefore, it is significant to develop a model for the prediction of disturbance in the geomagnetic field. This paper presents linear as well as nonlinear parametric techniques to model the complex magnetosphere dynamics. Neural network (NN)-based modeling techniques, such as feed-forward NN (FFNN), NN integrated with nonlinear autoregression with exogenous (NARX) inputs, adaptive neuro-fuzzy inference system (ANFIS), and recurrent NN are developed for the prediction of the magnetic storm. Global search algorithms, such as particle swarm optimization and genetic algorithm, also train the developed FFNNs. All the models are trained on 15 years’ real-time series data of solar wind parameters and DST index on a frequency of 1 h. The accuracy of the predicted DST index for multiple intensity levels of the magnetic storm using all developed linear and nonlinear modeling techniques is presented, compared, and analyzed thoroughly. The performance of NN integrated with NARX inputs and ANFIS is higher than rest of methods developed in the paper. Both techniques have the capability to predict mild, moderate, and intense magnetic storm with high degree of accuracy. The comparative analysis with state of the art shows the enhanced accuracy and robustness of developed models in this paper. [ABSTRACT FROM AUTHOR]

The paper proposes a novel block-wise prediction paradigm based on Convolutional Neural Networks (CNNs) for lossless video coding. A deep neural network model which follows a multi-resolution design is employed for block-wise prediction. Several contributions are proposed to improve neural network training. A first contribution proposes a novel loss function formulation for an efficient network training based on a new approach for patch selection. Another contribution consists in replacing all HEVC-based angular intra-prediction modes with a CNN-based intra-prediction method, where each angular prediction mode is complemented by a CNN-based prediction mode using a specifically trained model. Another contribution consists in an efficient adaptation of the CNN-based intra-prediction residual for lossless video coding. Experimental results on standard test sequences show that the proposed coding system outperforms the HEVC standard with an average bitrate improvement of around 5%. To our knowledge, the paper is the first to replace all the traditional HEVC-based angular intra-prediction modes with an intra-prediction method based on modern Machine Learning techniques for lossless video coding applications. [ABSTRACT FROM AUTHOR]

This paper aims to investigate how adequate cognitive functions for recognizing, predicting, and generating a variety of actions can be developed through iterative learning of action-caused dynamic perceptual patterns. Particularly, we examined the capabilities of mental simulation of one’s own actions as well as the inference of others’ intention because they play a crucial role, especially in social cognition. We propose a dynamic neural network model based on predictive coding which can generate and recognize dynamic visuo-proprioceptive patterns. The proposed model was examined by conducting a set of robotic simulation experiments in which a robot was trained to imitate visually perceived gesture patterns of human subjects in a simulation environment. The experimental results showed that the proposed model was able to develop a predictive model of imitative interaction through iterative learning of large-scale spatio-temporal patterns in visuo-proprioceptive input streams. Also, the experiment verified that the model was able to generate mental imagery of dynamic visuo-proprioceptive patterns without feeding the external inputs. Furthermore, the model was able to recognize the intention of others by minimizing prediction error in the observations of the others’ action patterns in an online manner. These findings suggest that the error minimization principle in predictive coding could provide a primal account for the mirror neuron functions for generating actions as well as recognizing those generated by others in a social cognitive context. [ABSTRACT FROM AUTHOR]

In this paper, the stability problem of delayed neural networks is investigated by adopting some switching methods. First, the delay interval is divided into many smaller variable intervals, and when the smaller variable interval is regarded as a mode of the delay, delayed neural networks are modeled as switched systems. Then, by using some switching methods, less conservative stability criteria are derived to ensure the stability of delayed neural networks. Finally, one example is provided to show the effectiveness of the obtained criteria. [ABSTRACT FROM AUTHOR]

Being able to detect, identify, and diagnose a fault is a key feature of industrial supervision systems, which enables advance asset management, in particular, predictive maintenance, which greatly increases efficiency and productivity. In this paper, an Industrial Internet app for real-time fault detection and diagnosis is implemented and tested in a pilot scale industrial motor. Real-time fault detection and identification is based on dynamic incremental principal component analysis (DIPCA) and reconstruction-based contribution (RBC). When the analysis indicates that one of the vibration measurements is responsible for the fault, a convolutional neural network (CNN) is used to identify the unbalance or bearing fault type. The application was evaluated in its three functionalities: fault detection, fault identification, and fault identification of vibration-related faults, yielding a fault detection rate over 99%, a false alarm rate below 5%, and an identification accuracy over 90%. Note to Practitioners—This paper focuses on designing and evaluating a real-time fault diagnosis application in an industrial setup. To this end, this paper also tackles the problem of developing a methodology for implementing advanced state-of-the-art fault detection techniques in real machinery, following industry standards and using a modern informatics architecture. The application here developed uses a statistical data-driven fault diagnosis technique, hence it requires a training stage using historical data to learn patterns and estimate parameters. A proof of concept in fault diagnosis for industrial motors is given; however, it should be noted that both the methodology and the deployed architecture are scalable and flexible enough to facilitate the implementation in other industrial environments. The implementation here presented was deployed using only open-source tools, which allows evaluating this tool without incurring in high expenses. [ABSTRACT FROM AUTHOR]

This paper proposes and experimentally evaluates a method, which is able to measure the refrigeration capacity of compressors in extremely short times. Traditionally, the measurement of this parameter is done using specialized and expensive test rigs and can take up to 4.5 h for the measurement of a single compressor. To overcome this long measurement time, it is proposed to use a simpler measuring system, in which, instead of measuring the refrigeration capacity directly, the rate of pressure rise imposed by the compressor to a cylinder with fixed volume is measured. This paper shows that both variables have a high correlation factor when the test and compressor conditions are taken into account. As the correlation among the variables is nonlinear, it is not possible to use simple mathematical correlation functions, so neural networks are used for that purpose. By using the proposed method, the refrigeration capacity of a compressor can be determined in less than 7 s, which is approximately 0.05% of the time required by traditional methods. To guarantee the metrological reliability of the results presented by the proposed neural model, this paper proposes a novel approach based on the Monte Carlo and the bootstrap methods. [ABSTRACT FROM AUTHOR]

Distribution system state estimation (DSSE) is a core task for monitoring and control of distribution networks. Widely used algorithms such as Gauss–Newton perform poorly with the limited number of measurements typically available for DSSE, often require many iterations to obtain reasonable results, and sometimes fail to converge. DSSE is a non-convex problem, and working with a limited number of measurements further aggravates the situation, as indeterminacy induces multiple global (in addition to local) minima. Gauss–Newton is also known to be sensitive to initialization. Hence, the situation is far from ideal. It is therefore natural to ask if there is a smart way of initializing Gauss–Newton that will avoid these DSSE-specific pitfalls. This paper proposes using historical or simulation-derived data to train a shallow neural network to “learn to initialize,” that is, map the available measurements to a point in the neighborhood of the true latent states (network voltages), which is used to initialize Gauss–Newton. It is shown that this hybrid machine learning/optimization approach yields superior performance in terms of stability, accuracy, and runtime efficiency, compared to conventional optimization-only approaches. It is also shown that judicious design of the neural network training cost function helps to improve the overall DSSE performance. [ABSTRACT FROM AUTHOR]

Most of the existing single object trackers track the target in a unitary local search window, making them particularly vulnerable to challenging factors such as heavy occlusions and out-of-view movements. Despite the attempts to further incorporate global search, prevailing mechanisms that cooperate local and global search are relatively static, thus are still sub-optimal for improving tracking performance. By further studying the local and global search results, we raise a question: can we allow more dynamics for cooperating both results? In this paper, we propose to introduce more dynamics by devising a dynamic attention-guided multi-trajectory tracking strategy. In particular, we construct dynamic appearance model that contains multiple target templates, each of which provides its own attention for locating the target in the new frame. Guided by different attention, we maintain diversified tracking results for the target to build multi-trajectory tracking history, allowing more candidates to represent the true target trajectory. After spanning the whole sequence, we introduce a multi-trajectory selection network to find the best trajectory that deliver improved tracking performance. Extensive experimental results show that our proposed tracking strategy achieves compelling performance on various large-scale tracking benchmarks. The project page of this paper can be found at https://sites.google.com/view/mt-track/. [ABSTRACT FROM AUTHOR]

Load forecasting is one of the most important and studied topics in modern power systems. Most of the existing research on day-ahead load forecasting try to build a good model to improve the forecasting accuracy. The forecasted load is then used as the input to generation scheduling with the ultimate goal of minimizing the cost of generation schedules. However, existing day-ahead load forecasting models do not consider this ultimate goal at the training/forecasting stage. This paper proposes a task-based day-ahead load forecasting model labeled as LfEdNet that combines two individual layers in one model, including a load forecasting layer based on deep neural network (Lf layer) and a day-ahead stochastic economic dispatch (SED) layer (Ed layer). The training of LfEdNet aims to minimize the cost of the day-ahead SED in the Ed layer by updating the parameters of the Lf layer. Sequential quadratic programming (SQP) is used to solve the day-ahead SED in the Ed layer. The test results demonstrate that the forecasted results produced by LfEdNet can lead to lower cost of day-ahead SED at the expense of slight reduction in forecasting accuracy. [ABSTRACT FROM AUTHOR]

The articles in this special section focus on situation awareness in intelligent human-computer interaction for critical decision making (HCI). HCI is recognized as an ctive field that focuses on the various interactions of human with machines. The HCI has been widely applied in multiple domains, such as artificial intelligence, computer vision, image and multimedia analysis, and cognitive and behavioral sciences. The objective of the HCI is to make the computer smart via receiving enough knowledge about the environment where it is deployed and reduce the human intervention aspect toward decision making. This enables development of high-end computers that are context aware and smart in making decisions with reference to the context. Situation awareness of an intelligent HCI will decide the success and application of the solution across the real world environment. The aim of this special issue is to provide a platform on the topic of situation awareness in intelligent HCI for time critical decision making. [ABSTRACT FROM AUTHOR]

Object detection is important in car sharing services. Accuracy, efficiency, and low memory consumption are desirable for object detection in car sharing services. This paper presents a network system that satisfies all these requirements. Our approach first divides the object detection task into multiple simpler local regression tasks. Then, we propose the generalized Haar filter-based convolutional neural network to reduce the consumption of memory and computing resource. To achieve real-time performance, we introduce a sparse window generation strategy to reduce the number of input image patches without sacrificing accuracy. We perform experiments on both vehicle and pedestrian data sets. Experimental results demonstrate that our approach can accurately detect objects under challenging conditions. Note to Practitioners—Object detection is an important part of intelligent vehicle technologies, which play an important role in car sharing services. Object detection provides metadata for collision avoidance, self-driving systems, and driver-assistance systems, which can result in better safety and consumer experiences in car sharing services. Although deep learning has achieved an excellent performance in object detection, they consume a large amount of storage and computing resource, which makes them difficult to be deployed for car sharing services. This paper suggests a novel approach which is based on the generalized Haar filter and the local regression strategy. Our approach is accurate, efficient, and light. The experimental results verify the effectiveness of the proposed approach in car sharing services. [ABSTRACT FROM AUTHOR]

This paper presents an efficient method, which reconstructs the temperature field around the tool/chip interface from infrared (IR) thermal images, for online monitoring the internal peak temperature of the cutting tool. The tool temperature field is divided into two regions; namely, a far field for solving the heat-transfer coefficient between the tool and ambient temperature, and a near field where an artificial neural network (ANN) is trained to account for the unknown heat variations at the frictional contact interface. Methods to extract physics-based feature points from the IR image as ANN inputs are discussed. The effects of image resolution, feature selection, chip occlusion, contact heat variation, and measurement noises on the estimated contact temperature are analyzed numerically and experimentally. The proposed method has been verified by comparing the ANN-estimated surface temperatures against “true values” experimentally obtained using a high-resolution IR imager on a custom-designed testbed as well as numerically simulated using finite-element analysis. The concept feasibility of the temperature monitoring method is demonstrated on an industrial lathe-turning center with a commercial tool insert. Note to Practitioners—The internal tool-temperature field around the tool/chip interface during cutting offers essential information to monitor tool-wear and ensure surface-quality particularly in finishing cuts, which is difficult to be monitored because of the stringent real-time requirements (including low cost and high accuracy) in harsh working environments. This paper presents a potentially low-cost solution that combines noncontact surface temperature measurements with high-fidelity physics-based computational models for monitoring the internal peak temperature at the tool/chip frictional contact during cutting. This physics-based method requires only a small number of preselected features, and uses thermal isotherms and streamlines to detect and substitute any occlusions in the IR images. The method does not rely on high-resolution (HR) images to infer the steep temperature gradient near the tool-tip where the peak temperature occurs, and thus can be implemented with a relatively low-cost IR imager. Unlike traditional least-square methods that bases solutions to a heat-conduction partial differential equation (PDE), the ANN is trained with precomputed physics-based models based on a novel dual-field (far-field and near-field) approach. The ANN-based method not only eliminates the need to solve the time-demanding PDE in real time, but also effectively accounts for parameter variations due to the uncertain frictional heat-fluxes at the contact interface during cutting. [ABSTRACT FROM AUTHOR]

In this paper, we propose a training framework for an inverter-based memristive neuromorphic hardware. The framework, which is called PHAX, is a physical characteristics aware one relying on an ex-situ training approach. The considered neuromorphic circuit is highly energy efficient hybrid CMOS-memristive implementation of neuromorphic circuits. To solve the problem of high sensitivity of the training to the mismatches between the high-level mathematical modeling of the neurons and the corresponding physical characteristics, an approach for analytical yet accurate modeling of the memristive crossbar and neuron circuits is suggested. The approach, which is based on SPICE simulations, models the inverter-based neurons using a hyperbolic tangent function. To increase the training efficacy, the backpropagation training algorithm is modified by considering some constraints based on the physical characteristics of the memristive circuit. This modification along with the accurate back-annotation of the physical characteristics considerably improve the effectiveness of the ex-situ training method of the neuromorphic circuit. The results of this paper show an average reduction of $1805{\times }$ in the training runtime compared to that of the in-situ training approach. Furthermore, the results of applying the approach on the kernels of some applications such as image recognition, image processing, and financial analysis reveal that the designed neuromorphic circuits provide an average power saving (speed up) of $1478{\times }$ ( $5.2{\times }$ ) over the ASIC implementation in a 90-nm CMOS technology. [ABSTRACT FROM AUTHOR]

The joint deformation contributes significantly to final end-effector displacement of a manipulator, especially when the manipulator is in the form a serial kinematics with long links. When the manipulators employed in the DEMO are under the heavy payload, the deformation of manipulator is inevitable and the magnitude is significant. In order to maneuver the large object through several via points with high positioning accuracy in the remote handling process of DEMO, the real-time computation of manipulator deformation has to be conducted, which is crucial to the DEMO adaptive position control system for the displacement compensation. Three computation-effective deformation modeling methods are proposed in this paper, which are parametric modeling method, nonparametric deterministic artificial neural network (ANN) modeling method, and nonparametric Bayesian ANN modeling method, respectively. A specific joint in a boom equipped in a telescopic articulated remote mast is taken as the study object in this paper. A nodal deformation in the joint is investigated by three modeling methods, respectively. The parametric deformation model is derived by using the structural mechanics, whose parameters are identified by using the Markov chain Monte Carlo (MCMC) method; the deformation model of deterministic ANN is trained by using the Levenberg–Marquardt method; and the deformation model of the Bayesian ANN is trained by using the MCMC method. The results show that the parametric model from the structural mechanics is linear and is incompetent in the deformation modeling when the nonlinearity presents; both the deterministic and Bayesian ANNs are capable of model the nodal deformation of joint. The performance of both the deterministic network and Bayesian network cannot rival for one another in the application scenario of this paper. The training of the Bayesian network can provide the criterions for estimation of possible ranges of the modeling outputs from its probabilistic distribution curves, and the judgment of proper size of network. [ABSTRACT FROM AUTHOR]

This paper presents a new non-deterministic approach for multiyear distribution network reinforcement planning (DNRP) considering the uncertainty sources pertaining to loads, electricity prices, investment costs, and operation costs. Accordingly, the underlying idea of the information-gap decision theory (IGDT) is used to obtain a robust solution protected against different realizations of each uncertain variable lying in its robust region. The proposed model is capable of adjusting the robustness of the optimal solution in terms of a specific parameter designated as the budget of uncertainty. As the uncertain loads, investment and operation costs competitively tend to maximize their robust regions for a particular value of the budget of uncertainty, the normalized normal constraint (NNC) method as a proficient multi-objective optimization method is exploited in this paper to solve the proposed multiobjective IGDT-based DNRP (IGDT-DNRP) model. Mainly, the NNC method presents a set of Pareto optimal solutions rather than a single optimal solution. Accordingly, a posteriori out-of-sample analysis is introduced in this paper to find the best solution among the set of Pareto optimal solutions. The proposed IGDT-DNRP model is implemented on the IEEE 33-bus distribution network under different circumstances. Simulation results illustrate the effectiveness of the proposed nondeterministic approach. [ABSTRACT FROM AUTHOR]

Accurate digital maps play a crucial role in various location-based services and applications. However, store information is usually missing or outdated in current maps. In this paper, we propose CrowdGIS, an automatic store self-updating system for digital maps that leverages street views and sensing data crowdsourced from mobile users. We first develop a new weighted artificial neural network to learn the underlying relationship between estimated positions and real positions to localize user’s shooting positions. Then, a novel text detection method is designed by considering two valuable features, including the color and texture information of letters. In this way, we can recognize complete store name instead of individual letters as in the previous study. Furthermore, we transfer the shooting position to the location of recognized stores in the map. Finally, CrowdGIS considers three updating categories (replacing, adding, and deleting) to update changed stores in the map based on the kernel density estimate model. We implement CrowdGIS and conduct extensive experiments in a real outdoor region for 1 month. The evaluation results demonstrate that CrowdGIS effectively accommodates store variations and updates stores to maintain an up-to-date map with high accuracy. Note to Practitioners—This paper was motivated by the problem of automatically updating digital maps in a manner of mobile crowdsensing. Existing approaches can update stores in maps through a manual survey or update roads automatically from mobile crowdsensing data. Since the store information is a crucial component in digital map, this paper suggests a novel approach to automatically update stores in digital maps through mobile crowdsensing. This is necessary, in general, because the accuracy of digital map will directly affect the quality of various location-based services. Therefore, the system proposed in this paper is useful for engineers and developers to obtain precise digital maps for localization, navigation, automatic drive, etc. [ABSTRACT FROM PUBLISHER]

The objective of this paper is to monitor complex process dynamics manifest in multivariate (multidimensional) time series data using a spectral (algebraic) graph theoretic approach. We test the hypothesis that the spectral graph-based topological invariants detect incipient process drifts earlier [lower average run length (ARL1)] and with higher fidelity (consistency of detection) when compared with the conventional statistics-based approaches. The presented approach maps a multidimensional sensor data stream \mathcal X^N\times d (visualize $N$ as time and d$ as the number of sensors) as an unweighted and undirected network graph G(V,E)$ , indexed by its vertices V$ and edges E . The rationale is that the graph-based topological invariants are surrogate representatives of the system state. We compare the monitoring performance of spectral graph theoretic invariants with conventional statistical features in an exponentially weighted moving average control chart setting. The practical utility of the approach is substantiated in the context of process monitoring in two advanced manufacturing scenarios, namely, ultraprecision machining (UPM) and semiconductor chemical mechanical planarization. These studies corroborate the hypothesis that graph theoretic invariants, when used as monitoring statistics, lead to lower ARL1 and more consistent detections in contrast to conventional statistical features. For instance, in the UPM case, the fault detection delay using graph theoretic invariants is less than 160 ms, compared with over 8 s of delay with statistical features.

Note to Practitioners—This paper addresses the critical problem of capturing process drifts from multidimensional (multisensor) data. The novelty of this paper is the development of a graph theoretic approach that combines signals from multiple in situ sensors for detecting abnormal process drifts. We show that this approach, which invokes graph-based topological invariants instead of statistical feature mining, is capable of capturing process drifts at an earlier stage (in terms of detection delay or lower average run length) and higher consistency of detection than conventional statistical features. As a practical consequence of this research, the operator can track the status of a complex process in a tractable control chart setting with only two graph theoretic topological invariants, as opposed to complex black-box models involving several features. [ABSTRACT FROM AUTHOR]

This paper describes digital circuit architectures for automatic speech recognition (ASR) and voice activity detection (VAD) with improved accuracy, programmability, and scalability. Our ASR architecture is designed to minimize off-chip memory bandwidth, which is the main driver of system power consumption. A SIMD processor with 32 parallel execution units efficiently evaluates feed-forward deep neural networks (NNs) for ASR, limiting memory usage with a sparse quantized weight matrix format. We argue that VADs should prioritize accuracy over area and power, and introduce a VAD circuit that uses an NN to classify modulation frequency features with 22.3- \mu \textW power consumption. The 65-nm test chip is shown to perform a variety of ASR tasks in real time, with vocabularies ranging from 11 words to 145 000 words and full-chip power consumption ranging from 172 \mu \textW to 7.78 mW. [ABSTRACT FROM AUTHOR]

In this paper, we present and analyze an event distribution system for brain–computer interfaces. Events are commonly used to mark and describe incidents during an experiment and are therefore critical for later data analysis or immediate real-time processing. The presented approach, called Tools for brain–computer interaction interface D (TiD), delivers messages in XML format via a buslike system using transmission control protocol connections or shared memory. A dedicated server dispatches TiD messages to distributed or local clients. The TiD message is designed to be flexible and contains time stamps for event synchronization, whereas events describe incidents, which occur during an experiment. TiD was tested extensively toward stability and latency. The effect of an occurring event jitter was analyzed and benchmarked on a reference implementation under different conditions as gigabit and 100-Mb Ethernet or Wi-Fi with a different number of event receivers. A 3-dB signal attenuation, which occurs when averaging jitter influenced trials aligned by events, is starting to become visible at around 1–2 kHz in the case of a gigabit connection. Mean event distribution times across operating systems are ranging from 0.3 to 0.5ms for a gigabit network connection for 106 events. Results for other environmental conditions are available in this paper. References already using TiD for event distribution are provided showing the applicability of TiD for event delivery with distributed or local clients. [ABSTRACT FROM PUBLISHER]

Light-fidelity (LiFi) is a fully-networked bidirectional optical wireless communication (OWC) technology that is considered as a promising solution for high-speed indoor connectivity. In this paper, the joint estimation of user 3D position and user equipment (UE) orientation in indoor LiFi systems with unknown emission power is investigated. Existing solutions for this problem assume either ideal LiFi system settings or perfect knowledge of the UE states, rendering them unsuitable for realistic LiFi systems. In addition, these solutions consider the non-line-of-sight (NLOS) links of the LiFi channel gain as a source of deterioration for the estimation performance instead of harnessing these components in improving the position and the orientation estimation performance. This is mainly due to the lack of appropriate estimation techniques that can extract the position and orientation information hidden in these components. In this paper, and against the above limitations, the UE is assumed to be connected with at least one access point (AP), i.e., at least one active LiFi link. Fingerprinting is employed as an estimation technique and the received signal-to-noise ratio (SNR) is used as an estimation metric, where both the line-of-sight (LOS) and NLOS components of the LiFi channel are considered. Motivated by the success of deep learning techniques in solving several complex estimation and prediction problems, we employ two deep artificial neural network (ANN) models, one based on the multilayer perceptron (MLP) and the second on the convolutional neural network (CNN), that can map efficiently the instantaneous received SNR with the user 3D position and the UE orientation. Through numerous examples, we investigate the performance of the proposed schemes in terms of the average estimation error, precision, computational time, and the bit error rate. We also compare this performance to that of the k-nearest neighbours (KNN) scheme, which is widely used in solving wireless localization problems. It is demonstrated that the proposed schemes achieve significant gains and are superior to the KNN scheme. [ABSTRACT FROM AUTHOR]

In underwater acoustic (UWA) adaptive communication system, due to time-varying channel, the transmitter often has outdated channel state information (CSI), which results in low efficiency. UWA channels are much more difficult to estimate and predict than terrestrial wireless channels, given the more severe multipath environments with varying propagation speeds in different locations, non-linear propagation paths, several-order higher propagation latency, mobile transceiver and obstacles in the sea, etc. To handle the complexity, this paper proposes an efficient online CSI prediction model for UWA CSI prediction considering the complicated correlationship of UWA channels in both the time and frequency domains. This paper designs a learning model called CsiPreNet, which is an integration of a one-dimensional convolutional neural network (CNN) and a long short term memory (LSTM) network. The performance is compared with the widely used recursive least squares (RLS) predictor, the approximate linear dependency recursive kernel least-squares (ALD-KRLS), and two common conventional deep neural networks (DNN) predictors, i.e., back propagation neural network (BPNN) and LSTM network using the measured data recorded in the South China Sea. To validate the efficacy of prediction, we investigate the prediction of CSI in simulated downlink UWA orthogonal frequency division multiple access (OFDMA) systems. Specifically, the measured UWA channel is used in the OFDMA system. A joint subcarrier-bit-power adaptive allocation scheme is used for resource allocation. To further improve the performance, we develop an offline-online prediction scheme, enabling the prediction results to be more stable. Simulation and experimental results show that the CsiPreNet has superior performance than the existing solutions, thanks to its capability in capturing both the temporal and frequency correlation of the UWA CSIs. [ABSTRACT FROM AUTHOR]