Ideal Outputs Research Articles

CritiCoder: An End-to-End Uncertain Regression Network for Robust Macroscopic Pressure Models in Water Distribution Systems

Due to the massive uncertain disturbances in water distribution systems (WDSs), it is rather challenging to construct robust macroscopic pressure models. In this article, we propose the CritiCoder, an end-to-end uncertain regression neural network, to build macroscopic pressure models in WDSs and estimate the distribution of uncertain disturbances. By separating the normal regression process into two parts, the CritiCoder decomposes the output into two parts brought by observable and unobservable variates separately. Two subnetworks, Coder and Critic, make up of the CritiCoder. Through the reconstruction of data flow, the Coder is expected to approximate ideal outputs from observable variates. Meanwhile, the loss function of the Critic is redesigned to consider the effectiveness of uncertain disturbances in output brought by unobservable variates. Experiments on a practical application of WDSs in a Chinese mega-city show superior performance of CritiCoder. Especially for pressure-monitoring nodes more severely impacted by disturbances, the performance decline of the CritiCoder is less compared with other baseline methods.

Leader learning loss function in neural network classification

Deep learning, based on Empirical Risk Minimization (ERM), typically aims to fit the ideal outputs of all samples due to its large capacity. However, models trained based on empirical losses like cross entropy (CE) or mean square error (MSE), often learn unnecessary information during classification, leading to premature overfitting. On the other hand, the result-focused loss functions, i.e., zero–one loss or hinge loss, are hard to optimize and thus are rarely applied directly in neural network. This paper proposes a novel leader learning in classification, where CE is gradually trained by classification results using sample-dependent cost-sensitive learning. As complementary, the stepwise-changed CE covers the deficiency on classification error while preserving the advantage of fast convergence. In this way, the deviation between CE and classification error can be corrected. Experimental results demonstrate that the proposed leader learning has a more significant convergence trend than the baseline algorithms. Moreover, the loss function learned from a specific dataset has broad generality that can be transferred to other models as prior knowledge.

Output Layer Structure Optimization for Weighted Regularized Extreme Learning Machine Based on Binary Method

In this paper, we focus on the redesign of the output layer for the weighted regularized extreme learning machine (WRELM). For multi-classification problems, the conventional method of the output layer setting, named “one-hot method”, is as follows: Let the class of samples be r; then, the output layer node number is r and the ideal output of s-th class is denoted by the s-th unit vector in Rr (1≤s≤r). Here, in this article, we propose a “binarymethod” to optimize the output layer structure: Let 2p−1<r≤2p, where p≥2, and p output nodes are utilized and, simultaneously, the ideal outputs are encoded in binary numbers. In this paper, the binary method is employed in WRELM. The weights are updated through iterative calculation, which is the most important process in general neural networks. While in the extreme learning machine, the weight matrix is calculated in least square method. That is, the coefficient matrix of the linear equations we solved is symmetric. For WRELM, we continue this idea. And the main part of the weight-solving process is a symmetry matrix. Compared with the one-hot method, the binary method requires fewer output layer nodes, especially when the number of sample categories is high. Thus, some memory space can be saved when storing data. In addition, the number of weights connecting the hidden and the output layer will also be greatly reduced, which will directly reduce the calculation time in the process of training the network. Numerical experiments are conducted to prove that compared with the one-hot method, the binary method can reduce the output nodes and hidden-output weights without damaging the learning precision.

Implementation of a Lightweight Semantic Segmentation Algorithm in Road Obstacle Detection.

Due to deep learning’s accurate cognition of the street environment, the convolutional neural network has achieved dramatic development in the application of street scenes. Considering the needs of autonomous driving and assisted driving, in a general way, computer vision technology is used to find obstacles to avoid collisions, which has made semantic segmentation a research priority in recent years. However, semantic segmentation has been constantly facing new challenges for quite a long time. Complex network depth information, large datasets, real-time requirements, etc., are typical problems that need to be solved urgently in the realization of autonomous driving technology. In order to address these problems, we propose an improved lightweight real-time semantic segmentation network, which is based on an efficient image cascading network (ICNet) architecture, using multi-scale branches and a cascaded feature fusion unit to extract rich multi-level features. In this paper, a spatial information network is designed to transmit more prior knowledge of spatial location and edge information. During the course of the training phase, we append an external loss function to enhance the learning process of the deep learning network system as well. This lightweight network can quickly perceive obstacles and detect roads in the drivable area from images to satisfy autonomous driving characteristics. The proposed model shows substantial performance on the Cityscapes dataset. With the premise of ensuring real-time performance, several sets of experimental comparisons illustrate that SP-ICNet enhances the accuracy of road obstacle detection and provides nearly ideal prediction outputs. Compared to the current popular semantic segmentation network, this study also demonstrates the effectiveness of our lightweight network for road obstacle detection in autonomous driving.

A Novel Low-Frequency Piezoelectric Motor Modulated by an Electromagnetic Field

For expanding the driving mode of the piezoelectric motor, a novel piezoelectric motor modulated by a magnetic field is proposed. This driving system combines piezoelectric driving and magnetic modulation together and can transform the reciprocating swing of the stator into step running of the rotor via the intermittent magnetic clamping between the rotor and stator. For investigating the inherent character of dynamics, the dynamic equations of key parts of the driving system are established. The natural frequencies and mode functions of the driving system are solved. A prototype was fabricated to prove the dynamic analysis and measure the output characteristic. The results show that the nature of the frequency measured from the test is coincident with theoretical analysis. In addition, by applying the driving frequency of 3 Hz, the voltage of the modulating signal of 4.5 V, the phase difference α between driving signal and modulating signal of 30°, the ideal outputs are 0.1046 rad/min for velocity and 0.405 Nmm for torque.

Learning Hadamard-Product-Propagation for Image Dehazing and Beyond

Image dehazing has evolved into an attractive research field in the computer vision community in the past few decades. Previous traditional approaches attempt to design energy-based objective functions. However, they cannot accurately express the intrinsic characteristics of the images, posing weak adaptation ability for real-world complex scenarios. More recently, deep learning techniques for image dehazing have matured and become more reliable, showing outstanding performance. Nevertheless, these methods heavily depend on training data, restricting their application ranges. More importantly, both traditional and deep learning approaches all ignore a common issue, noises/artifacts always appear in the recovery process. To this end, a new Hadamard-Product (HP) model is proposed, which consists of a series of data-driven priors. Based on this model, we derive a Learnable Hadamard-Product-Propagation (LHPP) by cascading a series of principle-inspired guidance and recovery modules. In which, the principle-inspired guidance related to transmission is endowed the smoothness property, the other recovery module satisfies the distribution of natural images. The Hadamard-product-based propagations is generated in our developed learnable framework for the task of image dehazing. In this way, we can eliminate noises/artifacts in the recovery procedure to obtain the ideal outputs. Subsequently, since the generality of our HP model, we successfully extend our LHPP to settle low-light image enhancement and underwater image enhancement problems. A series of analytical experiments are performed to verify our effectiveness. Plenty of performance evaluations on three complex tasks fully reveal our superiority against multiple state-of-the-art methods.

Binary Output Layer of Extreme Learning Machine for Solving Multi-class Classification Problems

Considered in this paper is the design of output layer nodes of extreme learning machine (ELM) for solving multi-class classification problems with r ( $$r\ge 3$$ ) classes of samples. The common and conventional setting of output layer, called “one-to-one approach” in this paper, is as follows: The output layer contains r output nodes corresponding to the r classes. And for an input sample of the ith class ( $$1\le i\le r$$ ), the ideal output is 1 for the ith output node, and 0 for all the other output nodes. We propose in this paper a new “binary approach”: Suppose $$2^{q-1}< r\le 2^q$$ with $$q\ge 2$$ , then we let the output layer contain q output nodes, and let the ideal outputs for the r classes be designed in a binary manner. Numerical experiments carried out in this paper show that our binary approach does equally good job as, but uses less output nodes and hidden-output weights than, the traditional one-to-one approach.

Binary Output Layer of Feedforward Neural Networks for Solving Multi-Class Classification Problems

Considered in this short note is the design of output layer nodes of feedforward neural networks for solving multiple-class classification problems with $r$ ( $r\geq 3$ ) classes of samples. The common and conventional setting of the output layer called “ $one - to - one~approach$ ” in this paper, is as follows: The output layer contains $r$ output nodes corresponding to the $r$ classes. And for an input sample of the $i$ -th class ( $1\leq i\leq r$ ), the ideal output is 1 for the $i$ -th output node, and 0 for all the other output nodes. We propose in this paper a new “ $binary~approach$ ”: Suppose $2^{q-1} with $q\geq 2$ , then we let the output layer contain $q$ output nodes, and let the ideal outputs for the $r$ classes be designed in a binary manner. This idea of binary output is also applied for other classifiers, such as support vector machines and associative pulsing neural networks. Numerical simulations are carried out on eight real-world data sets, showing that our binary approach performs as well as, but uses less output nodes and hidden-output weights than, the traditional one-to-one approach.

In brief

PAGE 995 The use of long coherent integration times in CZT-based algorithms could have applications for GPS systems operating in weak signal environments. A down-sampling data operation introduced by researchers in China demonstrates that the same signal acquisition can be achieved with a lower carrier-to-noise ratio and reduced computational effort. PAGE 991 Researchers in the UK working with semiconductor-based terahertz emitters have observed that temperature dependent polarity reversal within the material is an intrinsic property. Terahertz emission from various emitters over a range of temperatures revealed the semiconductor's inversion is not influenced by metallic contact, and creates potential for these devices to compete with commercial antennas. Signal detection probability improves with extended coherent integration times. PAGE 1007 A novel design for a simpler, more flexible, high performance microstrip power-divider (PD) has been implemented and tested by researchers in China. The device is composed of two dual-mode filtering PDs lacking any additional matching network, and provides significant design utility and size reduction with applications in wireless communications systems. Spectrometer set up for temperature dependence measurement of metal emitters. PAGE 967 Researchers from China have developed a frequency calibration method for a crystal-less current-controlled ring oscillator utilising a serial data signal from another chip as a reference. This newly presented auto-correlation method outperforms similar literature designs in terms of calibration accuracy, power consumption, and circuit area. Measured and simulated performances of power-divider device channels. PAGE 969 A purely digital approach to ultra-narrow pulse control for use in the picosecond and gigahertz domains has been presented by researchers in China. The digital synthesis method converts pulse-width control to the relative delay control between two signals, and can be applied to ultra-wideband communication systems and pulse radar. Frequency histograms before and after chip signal calibration. Diagram of ideal and real sharpener edge signal outputs.

Load frequency control of power systems with non‐linearities

Load frequency control (LFC) for power systems with governor deadband (GDB) and generation rate constraint (GRC) non-linearities is discussed in this study, with the objective to add anti-windup schemes for any designed LFC controller, so that the stability and performance of the existing LFC controller can be retained when there are non-linearities. Effects of GDB and GRC on the performance of LFC controllers are first analysed via describing function method. Anti-windup schemes are then proposed to compensate the GDB and GRC non-linearities. For GDB, the error between the realistic and the ideal outputs of the generator is added to the output of LFC controller for rejection, and for GRC, the error between the realistic and the ideal outputs of the turbine is fed back to the observer of the LFC controller for estimation and elimination. Simulation results show that the compensation schemes are effective in overcoming the adverse effects of GDB and GRC.

Wind velocity retrieval and field widening techniques of fringe-imaging Mach-Zehnder interferometer for Doppler lidar

Compared with the traditional Doppler frequency discriminator Fabry-Perot interferometer, Mach-Zehnder interferometer (MZI) has the advantages in high transmittance, linear parallel fringes instead of circular rings and wide field of view. An incoherent Doppler wind lidar system is demonstrated based on a fringe-imaging MZI. The mathematical model of wind velocity retrieval is derived, and the field widening of MZI is also used to optimize the performance of lidar system. The ideal fringe pattern outputs are obtained by numerical simulation, and the SineSqr function fitting method is proposed to determine the high-precision displacement of fringes after a Doppler shift. Furthermore, the field widening of MZI can be realized by field compensation which reduces the effect of the incidence angle on optical path difference. The results show that the wind velocity error obtained by SineSqr function fitting method is less than 0.45 m·s-1 at the line-of-sight wind velocity in a range from -100 m·s-1 to 100 m·s-1, which overcomes the instability of wind velocity retrieval by the gravity method. A full-angle field of view of 1° for MZI frequency discriminator without significant performance reduction is also achieved. These supplied discussion of the application techniques for Doppler lidar based on MZI fringe technique would promote the practical development of Doppler lidar for wind velocity measurement of the middle and upper atmosphere.

Indistinguishability and identifiability of kinetic models for the MurC reaction in peptidoglycan biosynthesis

An important question in Systems Biology is the design of experiments that enable discrimination between two (or more) competing chemical pathway models or biological mechanisms. In this paper analysis is performed between two different models describing the kinetic mechanism of a three-substrate three-product reaction, namely the MurC reaction in the cytoplasmic phase of peptidoglycan biosynthesis. One model involves ordered substrate binding and ordered release of the three products; the competing model also assumes ordered substrate binding, but with fast release of the three products. The two versions are shown to be distinguishable; however, if standard quasi-steady-state assumptions are made distinguishability cannot be determined. Once model structure uniqueness is ensured the experimenter must determine if it is possible to successfully recover rate constant values given the experiment observations, a process known as structural identifiability. Structural identifiability analysis is carried out for both models to determine which of the unknown reaction parameters can be determined uniquely, or otherwise, from the ideal system outputs. This structural analysis forms an integrated step towards the modelling of the full pathway of the cytoplasmic phase of peptidoglycan biosynthesis.

Injury alliance in Canada

To achieve greater societal impact within a highly competitive charitable marketplace, the four major, national, injury prevention organisations have decided to collaborate toward the following purposes:Long-term goalTo save more lives...

A Computer Software – Epitimulator® for Simulating Temporal Dynamics of Plant Disease Epidemic Progress

The objective of the present study was to develop a computer software for simulating the temporal development of plant disease epidemics using Richards, logistic, Gompertz, monomolecular, and exponential functions, respectively, and for predicting disease with a fitted model. The software was programmed using Visual Basic (VB6.0) and packaged with the Wise Installation System. The Fibonacci (‘0.618') section strategy was used to find out the most appropriate value for the shape parameter ( m) in Richards function simulation through looping procedures. The software program was repeatedly tested, debugged and edited until it was run through favorably and produced ideal outputs. It was named Epitimulator based on the phrase ‘epidemic time simulator' and has been registered by the National Copyright Department of China (Reg. no. 2007SR18489). It can be installed and run on personal computers with all versions of Windows operational systems. Data of disease index and survey time are keyed in or imported from Access files. The output of fitted models and related data of parameters can be pasted into Microsoft Excel worksheet or into Word document for editing as required and the simulated disease progress curves can be stored in separate graphic files. After being finally tested and completed, Epitimulator® was applied to simulate the epidemic progress of corn northern leaf blight ( Exserohilum turcicum) with recorded data from field surveys of corn crops and the fitted models were output. Comparison of the simulation results showed that the disease progress was always best described by Richards function, which resulted in the most accurate simulation model. Result also showed that forecast of northern leaf blight development was highly accurate by using the computed progress model from Richards function.

Iterative LBG clustering for SIMO channel identification

This paper deals with the problem of channel identification for Single Input Multiple Output (SIMO) slow fading channels using clustering algorithms. Due to the intrinsic memory of the discrete-time model of the channel, over short observation periods, the received data vectors of the SIMO model are spread in clusters because of the AWGN noise. Each cluster is practically centered around the ideal channel output labels without noise and the noisy received vectors are distributed according to a multivariate Gaussian distribution. Starting from the Markov SIMO channel model, simultaneous maximum-likelihood estimation of the input vector and the channel coefficients reduce to one of obtaining the values of this pair that minimizes the sum of the Euclidean norms between the received and the estimated output vectors. Viterbi algorithm can be used for this purpose provided the trellis diagram of the Markov model can be labeled with the noiseless channel outputs. The problem of identification of the ideal channel outputs, which is the focus of this paper, is then equivalent to designing a Vector Quantizer (VQ) from a training set corresponding to the observed noisy channel outputs. The Linde-Buzo-Gray (LBG)-type clustering algorithms [1] could be used to obtain the noiseless channel output labels from the noisy received vectors. One problem with the use of such algorithms for blind time-varying channel identification is the codebook initialization. This paper looks at two critical issues with regards to the use of VQ for channel identification. The first has to deal with the applicability of this technique in general; we present theoretical results for the conditions under which the technique may be applicable. The second aims at overcoming the codebook initialization problem by proposing a novel approach which attempts to make the first phase of the channel estimation faster than the classical codebook initialization methods. Sample simulation results are provided confirming the effectiveness of the proposed initialization technique.

Automated tuning of an electronic circuit board using the artificial neural network approach

Manufacturing of electronic circuits for microwave communication boards often requires tuning of different circuit characteristics by manual adjustment of several trimmer components, including the trimmer's resistance and capacitance. This manual tuning process was automated by applying the artificial neural network modeling approach. In the considered tuning process, which required manual adjustment of a set of trimmers, multiple specification criteria had to be satisfied by several trimmer rotations. The tuning process was described in terms of three independent steps: the circuit output measurement, trimmer selection, and trimmer rotation. The trimmer selection was performed by a semi-supervised neural network, which learned the patterns of circuit characteristics and the deviations between the ideal and practical outputs. Another network was developed for determination of trimmer rotation rate. The results, based on computer simulation of the tuning process, showed that the developed system improved performance of the tuning process, allowing for automation of the microwave circuit board tuning task in a real manufacturing environment.

A quality-of-service specification for multimedia presentations

The bandwidth limitations of multimedia systems force trade-offs between presentation-data fidelity and real-time performance. For example, digital video is commonly encoded with lossy compression to reduce bandwidth, and frames may be skipped during playback to maintain synchronization. These trade-offs depend on device performance and physical data representations that are hidden by a database system. If a multimedia database is to support digital video and other continuous media data types, we argue that the database should provide a quality-of-service (QOS) interface to allow application control of presentation timing and information-loss trade-offs. This paper proposes a data model for continuous media that preserves device and physical data independence. We show how to define formal QOS constraints from a specification of ideal presentation outputs. Our definition enables meaningful requests for endto-end service guarantees, while leaving the database system free to optimize resource management. We propose one set of QOS parameters that constitute a complete model for presentation error, and we show how this error model extends the opportunities for resource optimization.

Application of structured singular value synthesis to a fighter aircraft

The results of a design study to examine the control of a fighter aircraft at high angles of attack are presented. The flight condition considered represents an extreme point along a demanding large angle maneuver. Control laws are developed for a linear model of the aircraft using structure singular value synthesis to take the parameter uncertainty structure into account. The control objective is to attain acceptable flying qualities despite variations and uncertainty in the aircraft's aerodynamic coefficients. Desired flying qualities are embedded in ideal models of the angle of attack, sideslip angle, and stability axis roll rate responses to pilot inputs. The control objective is met by minimizing the weighted error between the ideal model outputs and the plant outputs while including structured parameter uncertainty in the design model.

On the optimum synthesis of sampled data multipole filters with random and nonrandom inputs

This paper considers the synthesis of optimum sampled data multipole filters with n inputs and m outputs. The signal portion of each input is assumed to consist of a stationary random component and a polynomial with unknown coefficients but known maximum order. Each signal is corrupted by stationary random noise. The filter under investigation is linear, time-invariant, and has finite memory. Each input to the filter consists of a sequence of impulses with a constant period T. Each impulse is assumed to have an area equal to the value of the signal plus noise at the sampling instant. The synthesis procedure to be developed is to specify the weighting functions of the filter such that the system error, which is defined as the difference between the actual and ideal outputs, has zero ensemble mean and the system ensemble mean square error is minimum. The weighting functions thus obtained will have, in general, abrupt jumps at the sampling instants but they are continuous within the sampling intervals. The synthesis procedure is extended to the case shown in Appendix A where each of the nonrandom signals can be expressed as an arbitrary linear combination of a set of known time functions. Further generalization is possible to the synthesis of time-varying filter with sampled nonstationary random inputs as given in Appendix B.

Techniques for the optimum synthesis of multipole control systems with random processes as inputs

This report considers the general problem of obtaining the optimum multipole system when the inputs to the system are stationary random processes. The system under investigation is linear and time invariant. The input to each terminal shall consist of signal and noise. The synthesis procedure is carried out under the basis that a fixed plant must be compensated in order to perform certain desired tasks. The design criterion employed is the minimum mean square error between actual outputs and ideal outputs of the system. A set of integral equations shall be obtained, which can be converted into algebraic equations through transformation. By solving these equations and using the method of undetermined coefficients, the transfer functions of the compensation can be uniquely determined.