Search

The random Fourier feature as an efficient kernel approximation method can effectively suppress the network growth of the traditional kernel-based adaptive filtering algorithm. Polarized random Fourier feature kernel least-mean-square(PRFFKLMS) remarkably improved the accuracy performance of random Fourier feature-based kernel least-mean-square algorithm and become the most representative random Fourier feature-based least-mean-square algorithm. In this paper, we studied the method that can improve the convergence speed of random Fourier feature-based least-mean-square algorithm. Based on the variable forgetting factor and variable step size strategy, three algorithm are proposed. The computational complexity of proposed algorithms are also given. The simulation results show that compared with PRFFKLMS algorithm, the convergence speed of the proposed algorithm is significantly improved.

Deep learning is widely used in vision tasks, but feature extraction of IR small targets is difficult due to the inconspicuous contours and lack of color information. This paper proposes a new convolutional neural network–based (CNN-based) method for IR small target detection called DRUNet. The algorithm is divided into two parts: the feature extraction network and the prediction head. For the small IR targets, which are difficult to accurately label, Gaussian soft labels are added to supervise the training process and make the network converge faster. We use a simplified object keypoint similarity to evaluate the network accuracy by the ratio of the distance to the radius of the inner tangent circle of the target box and a fair method for evaluating the model inference speed after GPU preheating. The experimental results show that our proposed algorithm performs better when compared with commonly used algorithms in the field of small target detection. The model size is 10.5 M, and the test speed reaches 133 FPS under the RTX3090 experimental platform.

Testing of real-time embedded systems (RTESs) under input timing constraints is a critical issue. Models which can specify timing constraints have respective merits and demerits and test suites which can cover more input possibilities and detect more faults under input timing constraints are worthy of study. In this paper, clocked computation tree logic which is used to specify input timing constraints is presented. Neighbor covering arrays and parallel input time correlation test suites are introduced to test RTESs under serial and parallel input timing constraints. Three algorithms are described in generating test suites, respectively, and corresponding random testing-based algorithms which are used as baselines for comparison are introduced. Benchmarks with different configurations are conducted to evaluate the algorithms' performance. Three real-world RTESs are tested with the test suites described in this paper, respectively. The test results show that random testing may omit some neighbor input time point combinations as the randomness and increase test suite scales. This fact may lead to the omission of some faults and heavy costs. Therefore, the proposed test suites are more effective and efficient for testing RTESs under input timing constraints.

This paper presents a polarized random Fourier feature kernel least-mean-square algorithm that aims to overcome the dimension curve of the random Fourier feature kernel least-mean-square (RFFKLMS) algorithm. RFFKLMS is an effective nonlinear adaptive filtering algorithm based on the kernel approximation technique. However, random samples drawn from the distribution need more dimensions to achieve better-generalized performance because they are independent of the training data. To overcome this weakness, a kernel polarization method is adopted to optimize the random samples. Polarized random Fourier features demonstrate a clear advantage over a method without using the polarization method. The experimental results in the context of Lorenz time series prediction and channel equalization verify the effectiveness of the proposed method.

In actual engineering applications, inevitable noise seriously affects the accuracy of fault diagnosis for rotating machinery. To effectively identify the fault classes of rotating machinery under noise interference, an efficient fault diagnosis method without additional denoising procedures is proposed. First, a one-dimensional deep residual shrinkage network, which directly takes the raw vibration signals contaminated by noise as input, is developed to realize end-to-end fault diagnosis. Then, to further enhance the noise immunity of the diagnosis model, the first layer of the model is set to a wide convolution layer to extract short time features. Moreover, an adaptive batch normalization algorithm (AdaBN) is introduced into the diagnosis model to enhance the adaptability to noise. Experimental results illustrate that the fault diagnosis model for rotating machinery based on one-dimensional deep residual shrinkage network with a wide convolution layer (1D-WDRSN) can accurately identify the fault classes even under noise interference.

A parallel search strategy based on sparse representation (PS-L1 tracker) is proposed in the particle filter framework. To obtain the weights of state particles, target templates are represented linearly with the dictionary of target candidates. Sparse constraints on the coefficient guarantee that only true target candidates can be selected, and the nonnegative entries denote the associate weights of efficient target states. Then the optimal target state can be estimated by the linear combination of above weighted states. In this way, efficient target states are selected simultaneously from all the particles, which we call a parallel search strategy. Experimental results demonstrate excellent performance of the proposed method on challenging infrared images.

Robust small target detection of low signal-to-noise ratio (SNR) is very important in infrared search and track applications for self-defense or attacks. Due to the complex background, current algorithms have some unsolved issues with false alarm rate. In order to reduce the false alarm rate, an infrared small target detection algorithm based on saliency detection and support vector machine was proposed. Firstly, we detect salient regions that may contain targets with phase spectrum Fourier transform (PFT) approach. Then, target recognition was performed in the salient regions. Experimental results show the proposed algorithm has ideal robustness and efficiency for real infrared small target detection applications.

Although combinatorial testing has been widely studied and used, there are still some situations and requirements that combinatorial testing does not apply to well, such as a system under test whose test cases need to be performed contiguously. For thorough testing, the testing requirements are not only to cover all the interactions among factors but also to cover all the value sequences of every factor. Generally, systems under test always involve constraints and dependencies in or among test cases. The constraints among test cases have not been effectively specified. First, we introduce extended covering arrays that can achieve both t-way combinatorial coverage and t-wise sequence coverage, and propose a clocked computation tree logic-based formal specification method for specifying constraints. Then, Particle Swarm Optimization (PSO) based Extended covering array Generator (PEG) is elaborated. To evaluate the constructed test suites, a method for verifying the constraints’ validity is presented, and kernel functions for measuring the coverage are also introduced. Finally, the performance of the proposed PEG is evaluated using several sets of benchmark experiments for some common constraints, and the feasibility and usefulness of PEG is validated.

To construct an online kernel adaptive filter in a non-stationary environment, we propose a randomized feature networks-based kernel least mean square (KLMS-RFN) algorithm. In contrast to the Gaussian kernel, which implicitly maps the input to an infinite dimensional space in theory, the randomized feature mapping transform inputs samples into a relatively low-dimensional feature space, where the transformed samples are approximately equivalent to those in the feature space using a shift-invariant kernel. The mean square convergence process of the proposed algorithm is investigated under the uniform convergence analysis method of a nonlinear adaptive filter. The computational complexity is also evaluated. In Lorenz time series prediction and nonstationary channel equalization scenarios, the simulation results demonstrate the effectiveness of the proposed algorithm.

In recent years, the online secondary path modeling (SPM) method with an auxiliary noise power scheduling strategy has been crucial for active noise control (ANC) systems and has also become a popular research topic. However, most scheduling strategies have been designed for broadband active noise control (BANC), and these ideas may not be directly applied to narrowband active noise control (NANC) systems due to the difference in structure between BANC and NANC systems. In this paper, a new online SPM method with auxiliary noise power scheduling is proposed. Here, the controlled system is adapted using the filtered-x weighted accumulated least mean square (FXWALMS) algorithm proposed by the authors. Moreover, the auxiliary noise power is scheduled based on the convergence status of the forgetting factor to accurately track the change in the NANC system. As a result, the proposed method not only achieves good modeling accuracy and fast convergence but also considerably increases noise attenuation. Extensive simulations are conducted to prove the superior performance of the proposed method in various scenarios.