Electronics Letters最新文献

A novel batch gradient descent algorithm for parameterized quantum circuits that significantly reduces the time complexity in terms of batch size for training quantum neural networks is proposed. Batch data constructed to quantum random access memory (qRAM) structure is mapped to one circuit that estimates average loss. As the number of circuits decreases, the range to which quantum amplitude estimation can be applied increases, speeding up with a quadratic scale in batch size.

A significant challenge in the domain of anti-drone warfare is the identification of enemies or own aircraft through the analysis of data broadcast by drones (e.g. ADS-B). This issue can be conceptualized as an open set recognition (OSR) problem. This paper proposes a DV-OSR-QSED framework for the purpose of data visualization-based OSR (DV-OSR). Phase-based 2D high-importance features are extracted, the DV-OSR framework is designed and mapped to 2D, and the 5th and 95th quantile selection-Euclidean distance (QSED) strategy is proposed. Experiments show that by using the proposed framework, the correct classification rate for known and unknown samples is 96.04% and 95.79%, the recall rate and F1 value are 89.00% and 92.27%, and the AUC is 0.9630.

The neural network proposed here specializes in herbarium image segmentation. The encoder of the proposed model contains multiple kernels of different sizes to address the complex structures of plant components, such as tangled roots and stems. By employing multiple kernel sizes, the convolution block enables multiscale learning, which is underexplored in previous approaches. This design effectively extracts and fuses local and global features, enabling both broad and narrow perspectives on complex structures within herbarium images and thereby improves segmentation performance. The experimental results demonstrate that the proposed model outperforms three conventional models. The source code can be accessed at https://github.com/tkdgur658/herbarim_segmentation_network

Urban complexes often feature a mix of commercial, entertainment and recreational space serving a wide range of services. Pedestrian intent classification is hence crucial to identify their different destinations and understanding their needs. Moreover, circadian effects generally influence pedestrian behaviour. This paper proposes a deep circadian-informed probability refinement network for pedestrian intent classification (CIPRNet). It incorporates circadian information using a multiplexer network architecture to refine preliminary classification probabilities generated by a preliminary deep learning-based trajectory classifier. A joint loss function is used to co-optimize both the preliminary baseline trajectory classifier and the CIPRNet. Experimental results using real pedestrian trajectories captured from 3D range sensors at the Osaka Asia and Pacific Trade Centre (ATC) on a sunny day and cloudy day show that the CIPRNet can improve the state-of-the-art prediction of pedestrian paths by long short term memory classifier and trajectory unified transformer by approximately 13% and 10%, respectively. The CIPRNet is also extended to trajectory prediction and it outperformed various state-of-the-art algorithms in terms of average and final displacement error reduction. It may serve as an attractive alternative for pedestrian intent classification for urban complexes.

Here, for the first time, a gated diode that can be turned on/off by its single insulated gate is introduced. The novel device is a combination of a metal-oxide-semiconductor field-effect transistor (MOSFET) and a diode. It has a simple structure and can be fabricated by the regular complementary metal-oxide-semiconductor (CMOS) technology at low cost. The insulated-gate unipolar diode (IGUD) is simulated by device simulator tools. Simulations show the output curve of the IGUD is not only similar to a regular diode but also can be shifted by the gate. The idea of IGUD has been evaluated by experimental tests. The experimental data are in good agreement with the simulation results. The IGUD can be used as a fast switch in high-current low-voltage applications. Also, it can be used to achieve controlled rectification without synchronisation to the AC input.

Inverse Synthetic Aperture Radar (ISAR) is a vital radar imaging technique that leverages the relative motion between the radar and the target to generate high-resolution images. Traditional ISAR methods; however, are highly sensitive to inaccuracies in estimating rotational parameters, roll, pitch, and yaw, leading to image degradation. This article proposes a novel Differential Semblance Optimization (DSO) criterion for imaging dynamically rotating targets in a multistatic ISAR configuration. Unlike the Intensity Criterion (IC), which requires a precise initial parameter range, DSO enables broader exploration of value ranges, offering greater flexibility. Although the experiments focus on yaw rotation, the method is versatile and extendable to other rotational parameters. Tests with varying transmitter and receiver configurations demonstrate that DSO maintains robust performance even with fewer receivers. Comparisons with IC show that DSO produces sharper, more focused images and performs robustly in noisy environments, underscoring its potential for enhancing ISAR imaging in complex and dynamic scenarios.

The effectiveness of deep learning-based methods for insulator defect detection has been proven. However, in practical applications of power transmission lines, the complex and variable backgrounds in insulator images, coupled with the difficulty in labeling insulator defects, pose challenges to improving the robustness of such methods. Existing studies often utilize generative adversarial networks or forcefully combine foreground and background to augment training samples, but they overlook the rich semantic information in complex scenes, leading to distorted generated adversarial samples. To address this challenge, an innovative multi-semantic contrast enhancement method that significantly enhances the robustness of defect detection by deeply integrating high-level semantic knowledge and low-level signal priors is proposed. Moreover, through adversarial training using generated samples with diverse semantics and real samples, the robustness of the method is further improved. Experimental results demonstrate that this method surpasses state-of-the-art models, achieving significant performance on three independent cross-scene datasets.

The time step functions as a crucial temporal unit for simulating neuronal dynamics within spiking neural networks, which play a significant role in neuromorphic computing systems. Efficient management of these time steps is vital to ensure model accuracy while optimizing overall system performance. As system scale increases, variations in hardware across subsystems and their asynchronous operations create challenges in achieving effective time step control. To address this issue, this paper proposes an innovative framework for managing time steps in large-scale neuromorphic systems. This framework allows subsystems to dynamically adjust their time step lengths according to computational loads and to perform look-ahead computations. Such a strategy effectively reduces the overhead related to time step synchronization, enhancing system efficiency. Additionally, the paper introduces a safeguard mechanism to ensure the system's reliability. Experimental results indicate that the proposed framework sustains the correct long-term operation of the system and improves model execution performance by 8.88% to 27.15% when compared to existing methods.

Metastability in successive-approximation register analogue-to-digital converters (ADCs) degrades the ADC's signal-to-noise and distortion ratio and causes error propagation through the digital equalizers of ADC-based receivers. To mitigate these issues, a charge-pump-based successive-approximation register metastability calibration method is proposed. This approach operates independently of a fixed voltage or time reference. The calibration process is executed in the background with pipelining, requiring minimal additional power. Comprehensive testing shows that the proposed calibration consistently enhances ADC SNDR and reduces the code error rate across a wide range of sampling rates.

In extended target tracking, Gaussian Process (GP) is utilized to model unknown contour functions based on the model-predicted target center and contour measurements. However, model prediction relies on accurate prior knowledge. When the model-predicted target center is inaccurate, it will affect the modelling of the measurement model. To address issue, this letter introduces a hybrid-driven approach that combines extended Kalman filter using GP with neural network; proposes an extended target tracking algorithm using neural network and GP. The algorithm predicts the target center according to the neural network and the target's kinematic model, and takes the prediction center and the contour measurements at the current moment as the input of the neural network, which in turn provides real-time estimates for the predicted center compensation. The simulation results show that the algorithm has a significant improvement in tracking performance and better accuracy in estimating the center position and extent state of the target.