We propose DoReFa-Net, a method to train convolutional neural networks that have low bitwidth weights and activations using parameter gradients. In particular, during backward pass, gradients are stochastically quantized numbers before being propagated layers. As convolutions forward/backward passes can now operate on activations/gradients respectively, DoReFa-Net use bit convolution kernels accelerate both training inference. Moreover, as be efficiently implemented CPU, FPGA, ASIC GPU,...

Previous approaches for scene text detection have already achieved promising performances across various benchmarks. However, they usually fall short when dealing with challenging scenarios, even equipped deep neural network models, because the overall performance is determined by interplay of multiple stages and components in pipelines. In this work, we propose a simple yet powerful pipeline that yields fast accurate natural scenes. The directly predicts words or lines arbitrary...

Metering data from the advanced metering infrastructure can be used to find abnormal electricity behavior for detection of theft, which causes huge financial losses electric companies every year. This article proposes an theft detector using based on extreme gradient boosting (XGBoost). The are preprocessed, including recover missing and erroneous values normalization. classification model XGBoost is trained both benign malicious samples. Simulations done by Irish Smart Energy Trails set...

The weak grids may be unstable due to the negative resistance behavior of grid-connected inverter (GCI) caused by synchronization unit with phase-locked loop (PLL). This article proposes a passivity enhancement method attain positive output GCI in qq channel. Initially, impedance decomposition is utilized reveal reason for brought PLL. Secondly, decomposed impedance, novel PLL adding prefilter and an phase regulator normalized amplitude input traditional synchronous reference frame (SRF-PLL)...

The interpolated discrete Fourier transform (IpDFT) is one of the most popular techniques to estimate parameters a damped real-valued sinusoidal signal (DRSS). However, its accuracy affected by strong noise presence and short observation windows. To this end, letter proposes novel two-point IpDFT method, called I2pZDFT, for parameter estimation DRSS. proposed I2pZDFT uses zero-padding technique increase sampling rate in frequency domain. conjugate symmetry parity zero-padded are utilized...

Harmonic estimation is an important topic in power system signal processing. Windowed interpolation fast Fourier transformation (WIFFT) efficient algorithm for harmonic estimation, which can eliminate the errors caused by spectral leakage and picket fence effect. However, fitting polynomial procedure contains both even odd terms, this increases computational burden. This paper proposes a simple symmetrical FFT algorithm, where terms are removed from based on triangular self-convolution...

Previous approaches for scene text detection have already achieved promising performances across various benchmarks. However, they usually fall short when dealing with challenging scenarios, even equipped deep neural network models, because the overall performance is determined by interplay of multiple stages and components in pipelines. In this work, we propose a simple yet powerful pipeline that yields fast accurate natural scenes. The directly predicts words or lines arbitrary...

We report results of formaldehyde gas (CH2O) detection under dark conditions and UV light irradiation with pristine tungsten trioxide nanowires (WO3 NWs) metal nanoparticles decorated WO3 NWs sensing layers. The resistive layers were deposited by one step aerosol assisted chemical vapor deposition (AACVD) on commercial alumina substrates 10-pair interdigitated platinum electrodes. elaborated sensors, based Au, Pt, Au/Pt, Ni Fe nanoparticles, investigated towards three concentrations (5, 10...

Electricity theft has been a growing concern for the smart grid. It can be defined as follows: illegal customers use energy from electric utilities without contract or manipulate their meter readings to pay less not electricity bill. Over past decade, significant studies have done prevent and combat theft. This article aims provide general overview of progress detection, including threat models, datasets input features adopted, methodologies techniques, evaluation metrics. We also make...

The phase-locked loop (PLL)-caused couplings on the grid-connected converter (GCC) are strengthened by increasing its bandwidth and then incur synchronization instability. This article proposes a constant-coupling-effect-based PLL (CCE–PLL) to resolve aforementioned issues. Initially, feature of PLL-caused GCC is explored. It exhibits that varied along with hence jeopardizing system's stability. Subsequently, CCE–PLL constant low coupling regardless proposed. illustrates does not influence...

For interpolated discrete Fourier transform (IpDFT)-based phasor estimators, the out-of-band interference (OOBI) test is among most challenging ones. The typical iterative-interpolated DFT (i-IpDFT) estimator utilizes a two-step iterative framework to eliminate effects of negative frequency and OOBI. However, speed estimation limited by adopted redundant iterations. To this end, article proposes fast i-IpDFT (FiIpDFT) method for an OOBI contaminated signal, which three-point IpDFT (I3pDFT)...

Deep learning has been widely used in recent years for printed circuit board (PCB) defect detection because of its excellent performance. However, deep-learning-based approaches often suffer from the over-fitting problem due to lack sufficient training data real applications. Meanwhile, these still have some challenges detect defects with small sizes and irregular shapes. To address problems, this article developed a novel deep Siamese semantic segmentation network which integrates...

The wide bandwidth of phase-locked loop (PLL) will increase the negative real part output impedance grid-connected inverter (GCI), thus destroying stability weak grids. This article proposes a novel method to improve system through decoupling PLL and grid impedance. Initially, coupling relationship between is revealed. It can explain why choice magnitude limited. Then, proposed where primary control structure GCI does not need change uses structured voltage, product PCC voltage admittance,...

We study the exact set similarity join problem, which, given two collections of sets, finds out all similar pairs from collections. Existing methods generally utilize prefix filter based framework. They generate a for each and prune whose prefixes are disjoint. However pruning power is limited, because if dissimilar sets share common element in their prefixes, they cannot be pruned. To address this we propose partition-based design partition scheme to into several subsets guarantee that only...

This paper focuses on the accurate frequency estimation of power signals corrupted by a stationary white noise. The noneven item interpolation FFT based triangular self-convolution window is described. A simple analytical expression for variance noise contribution derived, which shows variances are proportional to energy adopted window. Based proposed method, level measurement channel can be estimated, and optimal parameters (e.g., sampling length) algorithm that minimize thus determined....

Reducing bit-widths of weights, activations, and gradients a Neural Network can shrink its storage size memory usage, also allow for faster training inference by exploiting bitwise operations. However, previous attempts quantization RNNs show considerable performance degradation when using low bit-width weights activations. In this paper, we propose methods to quantize the structure gates interlinks in LSTM GRU cells. addition, balanced further reduce degradation. Experiments on PTB IMDB...

This paper presents a symmetric three points IpDFT (S3IpDFT) method which can be associated with all combined cosine windows. The influence of spectral interference from image component and noise on the power system frequency estimation provided by S3IpDFT is investigated. expression for absolute error normalized caused presented. Also, analytical variance has been derived. derived results are validated means computer simulations experiments.

Many of the actions that we take with our hands involve self-contact and occlusion: shaking hands, making a fist, or interlacing fingers while thinking. This use illustrates importance tracking through occlusion for many applications in computer vision graphics, but existing methods faces are not designed to treat extreme amounts self-occlusion exhibited by common hand gestures. By extending recent advances vision-based physically based animation, present first algorithm capable...

Tracking the frequency of power systems accurately and promptly is cornerstone industry applications. The excellence windowed interpolation discrete Fourier transform (DFT)–based methods has been proved in estimation by suppressing both short- long-range spectral leakage. However, due to nonnegligible image component, their behavior deteriorates dramatically when short time windows are adopted. To this purpose, a novel estimator based on two-point DFT Hanning window proposed for estimating...

The power coupling issue in grid-forming converters (GFMs) is an emerging concern within systems dominated by inverter-based resources. Specifically, the phase difference between two rotational frames, i.e., system/grid and controller has not been associated with effect. It demonstrated that such causes interaction inner control loops. To this end, we first developed a detailed model of GFM system to study sensitivity parameters against issues. In particular, found voltage loop could impact...