Feature selection in machine learning is subject to the intrinsic randomness of feature algorithms (e.g., random permutations during MDA). Stability selected features with respect such essential human interpretability a algorithm. The authors propose rank-based stability metric called <i>instability index</i> compare stabilities three algorithms—MDA, LIME, and SHAP—as applied forests. Typically, are by averaging many iterations Although variability does decrease as number increases, it not...

Weather forecasting is a crucial task for meteorologic research, with direct social and economic impacts. Recently, data-driven weather models based on deep learning have shown great potential, achieving superior performance compared traditional numerical prediction methods. However, these often require massive training data computational resources. In this paper, we propose EWMoE, an effective model accurate global forecasting, which requires significantly less Our incorporates three key...

A polarisation rotation reflective surface (PRRS) is proposed to achieve radar cross‐section (RCS) reduction of a patch antenna at broadband. The PRRS consists two pairs L‐shaped patches, which are printed on substrate backed by metallic ground. By arranging the around in four orthogonal directions, wideband RCS can be obtained frequency band 10–18 GHz. low‐RCS fabricated and measured an anechoic chamber. experimental results show that designed makes dramatically reduced. What more,...

We present a novel code-aided carrier recovery method for low-density parity-check (LDPC) coded systems, which comprises two supporting algorithms: 1) coarse synchronization by maximizing cost function, the mean absolute value of soft outputs LDPC decoder, followed simple interpolation operation to improve estimation accuracy and 2) fine based on decisions produced decoder. With moderate computational complexity, this proposed algorithm is designed synchronize signals with large frequency...

Weather forecasting is a long-standing computational challenge with direct societal and economic impacts. This task involves large amount of continuous data collection exhibits rich spatiotemporal dependencies over long periods, making it highly suitable for deep learning models. In this paper, we apply pre-training techniques to weather propose W-MAE, model Masked AutoEncoder forecasting. W-MAE pre-trained in self-supervised manner reconstruct spatial correlations within meteorological...

Weather forecasting is a crucial task for meteorologic research, with direct social and economic impacts. Recently, data-driven weather models based on deep learning have shown great potential, achieving superior performance compared traditional numerical prediction methods. However, these often require massive training data computational resources. In this paper, we propose EWMoE, an effective model accurate global forecasting, which requires significantly less Our incorporates three key...

Clock gating is a power efficient technique by switching off unnecessary clock signals to the registers. The condition under which registers can be safely gated checked using EXOR of current and next state values. Due extra consumed logics consisting latch an AND gate, we have proposed optimum sharing method controls based on BDD (Binary Decision Diagram) with single-stage for optimization. In this paper, enhance optimization including multi-stage compare structural approach. By gating,...

We present an improved code-aided symbol timing recovery method for low-density parity-check (LDPC) coded systems, which comprises two supporting algorithms: 1) a coarse estimation by utilizing cost function, the mean absolute value of soft output LDPC decoder and 2) fine based on Expectation-Maximization (EM) algorithm with simple maximization step. With low computational complexity, this proposed can achieve at signal-to-noise ratio (SNR) large range, eliminating biased results many...

The detection problem in impulsive noise modeled by the symmetric alpha stable <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mrow><mml:mrow><mml:mfenced open="(" close=")" separators="|"><mml:mrow><mml:mtext>S</mml:mtext><mml:mi>α</mml:mi><mml:mtext>S</mml:mtext></mml:mrow></mml:mfenced></mml:mrow></mml:mrow></mml:math> distribution is studied. traditional detectors based on second or higher order moments fail...

While scene text image super-resolution (STISR) has yielded remarkable improvements in accurately recognizing text, prior methodologies have placed excessive emphasis on optimizing performance, rather than paying due attention to efficiency - a crucial factor ensuring deployment of the STISR-STR pipeline. In this work, we propose novel Efficient Scene Text Image Super-resolution (ESTISR) Network for resource-limited platform. ESTISR's functionality primarily depends two critical components:...

Clock gating is the insertion of control signal for registers to switch off unnecessary clock signals selectively without violating functional correctness original design so as reduce dynamic power consumption. Commercial EDA tools usually have a mechanism generate logic based on structural method where specified by designers are used, and effectiveness depends signals. In research, we focus automatic generation propose candidate extraction selection. We formalize selection using linear...

In this paper, an analytical method for predicting the armature-reaction field distribution of surface-mounted permanent-magnet machine accounting slots in stator and rotor was presented. The model is divided into three types subdomains. And solution each subdomai n obtained by applying interface boundary conditions. non-overlapping or overlapping windings are also considered. prototype with 4-poles/24-stator slots/20-rotor used validation. FE (finite element) results confirm validity proposed model.

We present a frame synchronization algorithm for low-density parity-check (LDPC) coded burst transmissions, which combines conventional pilots-assisted and code-aided based on the mean magnitude of soft outputs from LDPC decoder. With moderate computational complexity, proposed is more efficient in bandwidth than algorithms. When compared with other algorithms, offers better trade-off between complexity performance. Simulation results case an 8-PSK system (1944, 972) code show that can...

Clock gating is supported by commercial tools as a power optimization feature based on the guard signal described in HDL (structural method). However, identification of control signals for gated registers hard and designer-intensive work. Besides, since clock cells also consume power, it imperative to minimize number inserted their switching activities optimization. In this paper, we propose an automatic multi-stage algorithm with ILP (Integer Linear Programming) formulation, including...

The problem of multichannel synthetic aperture radar image recovery in the presence sidelobe narrow‐band interference (NBI) is addressed. NBI steering matrix accurately estimated with signal received passive mode. A beamforming technique then utilised to minimise energy loss target echoes and suppress a satisfactory level. By establishing linear model for removed, reconstructed via sparse recovery. experimental results demonstrate that proposed method can both effectively generate high...

Feature selection in machine learning is subject to the intrinsic randomness of feature algorithms (for example, random permutations during MDA). Stability selected features with respect such essential human interpretability a algorithm. We proposes rank based stability metric called instability index compare stabilities three MDA, LIME, and SHAP as applied forests. Typically, are by averaging many iterations Though we find that variability does decrease number increases, it not go zero, do...

In this paper, we propose robust stochastic algorithms for solving convex compositional problems of the form $f(\E_\xi g(\cdot; \xi)) + r(\cdot)$ by establishing {\bf sub-Gaussian confidence bounds} under weak assumptions about tails noise distribution, i.e., heavy-tailed noise} with bounded second-order moments. One can achieve goal using an existing boosting strategy that boosts a low probability convergence result into high result. However, piecing together results suffers from several...

A digital closed-loop control quartz accelerometer with new structures is performed, which overcomes the disadvantages of restriction measurement range and sensitivity QVBA (quartz vibrating beam accelerometer). It has not only same precision as in theory, but also large without reducing sensitivity. The proposed to be used high-precision acceleration.

In this paper, we propose a modified superposition modulation (MSM) and new demodulation method based on decorrelation. Compared to the traditional (SM) signal, constellation of MSM signal is asymmetric, which makes it hard be demodulated by un-cooperative receiver. Traditionally, iterative processing adopted for SM computational complexity amazing. Aiming at proposed decorrelation, calculation can reduced remarkably.