The Kalman filter (KF)-based active noise control (ANC) system demonstrates superior tracking and faster convergence compared to the least mean square (LMS) method, particularly in dynamic noise cancellation scenarios. However, in environments with extremely high noise levels, the power of the control signal can exceed the system’s rated output power due to hardware limitations, leading to output saturation and subsequent non-linearity. To mitigate this issue, a modified KF with an output constraint is proposed. In this approach, the disturbance treated as an measurement is re-scaled by a constraint factor, which is determined by the system’s rated power, the secondary path gain, and the disturbance power. As a result, the output power of the system, i.e. the control signal, is indirectly constrained within the maximum output of the system, ensuring stability. Simulation results indicate that the proposed algorithm not only achieves rapid suppression of dynamic noise but also effectively prevents non-linearity due to output saturation, highlighting its practical significance.

Large language models signify a pivotal advancement in general artificial intelligence, exhibiting capabilities that exceed human performance in diverse tasks. Nevertheless, these models often lack expertise in specialized knowledge areas. To augment the performance of LLMs in downstream applications, enhancing their knowledge acquisition and comprehension is imperative. In this paper, we introduce a knowledge-enhanced large language model, named "KARLM", which integrates symbolic AI into the training of LLM through a knowledge base derived from logic extracted from datasets and external resources. By incorporating this KB in conjunction with training corpora, "KARLM" is adeptly enabled to acquire and understand domain-specific knowledge. We validated the effectiveness of this method on recommendation tasks. Extensive experiments on multiple datasets indicate that "KARLM" successfully learns item knowledge and outperforms state-of-the-art baselines.

Virtual sensing (VS) technology enables active noise control (ANC) systems to attenuate noise at virtual locations distant from the physical error microphones. Appropriate auxiliary filters (AF) can significantly enhance the effectiveness of VS approaches. The selection of appropriate AF for various types of noise can be automatically achieved using convolutional neural networks (CNNs). However, training the CNN model for different ANC systems is often labour-intensive and timeconsuming. To tackle this problem, we propose a novel method, Transferable Selective VS, by integrating metric-learning technology into CNN-based VS approaches. The Transferable Selective VS method allows a pre-trained CNN to be applied directly to new ANC systems without requiring retraining, and it can handle unseen noise types. Numerical simulations demonstrate the effectiveness of the proposed method in attenuating suddenvarying broadband noises and real-world noises.

Hybrid active noise control (ANC) is widely employed in portable commercial products to attenuate both broadband and narrowband noise. Although the adaptive hybrid ANC, updated by the filtered reference least mean square (FxLMS) algorithm, can achieve optimal noise control even with uncorrelated noise, its slow convergence speed significantly decreases dynamic noise reduction performance. To address this challenge, we propose a two-stage model-agnostic meta-learning (MAML) approach to compute the optimal initial coefficients of the control filters for the hybrid ANC, effectively reducing the convergence time of adaptive algorithms. Different from conventional variable step-size strategies, this data-driven method determines the optimal initial coefficients based on the statistical characteristics of the noise, ensuring system stability. Furthermore, numerical simulations demonstrate that two-stage MAML initialization of adaptive hybrid ANC significantly accelerates convergence speed for attenuating broadband and real-world noise.

Photoplethysmography (PPG) is commonly used to gather health-related information but is highly affected by motion artifacts from daily activities. Inspired by the strong denoising capabilities and generalization of diffusion probabilistic models, this paper proposes a novel PPG denoising method using a diffusion probabilistic model to reduce the impact of these artifacts. While typical diffusion models handle Gaussian noises, motion artifacts often involve non-Gaussian noise. To address this, the proposed method incorporates noisy PPG signals into both the diffusion and reverse processes, allowing the model to adapt better to complex and non-Gaussian noises. A dataset with clean and noisy PPG signals from 15 subjects performing various motion tasks was collected for evaluation. The results show the proposed model significantly improves PPG signal quality, reducing the Peak-Rejection-Rate (PRR) from 0.24 to 0.03. It also enhances the accuracy of heart rate (HR) estimation and various heart rate variability (HRV) measures, showing robustness and good generalization across different tasks and subjects.

Delayless noise control is achieved by our earlier generative fixed-filter active noise control (GFANC) framework through efficient coordination between the co-processor and real-time controller. However, the one-dimensional convolutional neural network (1D CNN) in the co-processor requires initial training using labelled noise datasets. Labelling noise data can be resource-intensive and may introduce some biases. In this paper, we propose an unsupervised-GFANC approach to simplify the 1D CNN training process and enhance its practicality. During training, the co-processor and real-time controller are integrated into an end-to-end differentiable ANC system. This enables us to use the accumulated squared error signal as the loss for training the 1D CNN. With this unsupervised learning paradigm, the unsupervised-GFANC method not only omits the labelling process but also exhibits better noise reduction performance compared to the supervised GFANC method in real noise experiments.

Selective Fixed-filter Active Noise Control (SFANC) is limited by its selection of a single candidate from pre-trained control filters. In contrast, Generative Fixed-filter Active Noise Control (GFANC) addresses this limitation by employing an adaptive combination of sub control filters to generate more suitable control filters for different primary noises. However, GFANC solely relies on the information from the current noise frame to generate its control filter, resulting in potential inaccuracies when dealing with dynamic noises. Therefore, we propose a GFANC-Kalman approach that integrates an efficient one-dimensional convolutional neural network (1D CNN) with a Kalman filter to further improve the performance of GFANC. Specifically, the weight vector used to combine sub control filters is predicted by the 1D CNN for each noise frame, and then processed by the Kalman filter with minimal complexity. By considering the correlation between adjacent noise frames, the Kalman filter can enhance the accuracy and robustness of weight vector prediction. Hence, GFANC-Kalman is more able to adapt to changes in noise distribution, particularly for dynamic noises. Numerical simulations validate the efficacy of the proposed GFANC-Kalman approach in dealing with real-world dynamic noises.

By assigning the massive computing tasks of the traditional multichannel active noise control (MCANC) system to several distributed control nodes, distributed multichannel active noise control (DM-CANC) techniques have become effective global noise reduction solutions with low computational costs. However, existing DMCANC algorithms simply complete the distribution of traditional centralized algorithms by combining neighbour nodes’ information but rarely consider the degraded control performance and system stability of distributed units caused by delays and interruptions in communication. Hence, this paper develops a novel DMCANC algorithm that utilizes the compensation filters and neighbour nodes’ information to counterbalance the cross-talk effect between channels while maintaining independent weight updating. Since the neighbours’ information required barely affects the local control filter updating in each node, this approach can tolerate communication delay and interruption to some extent. Numerical simulations demonstrate that the proposed algorithm can achieve satisfactory noise reduction performance and high robustness to real-world communication challenges.

Due to the slow convergence and poor tracking ability, conventional LMS-based adaptive algorithms are less capable of handling dynamic noises. Selective fixed-filter active noise control (SFANC) can significantly reduce response time by selecting appropriate pre-trained control filters for different noises. Nonetheless, the limited number of pre-trained control filters may affect noise reduction performance, especially when the incoming noise differs much from the initial noises during pre-training. Therefore, a generative fixed-filter active noise control (GFANC) method is proposed in this paper to overcome the limitation. Based on deep learning and a perfect-reconstruction filter bank, the GFANC method only requires a few prior data (one pre-trained broadband control filter) to automatically generate suitable control filters for various noises. The efficacy of the GFANC method is demonstrated by numerical simulations on real-recorded noises.

Active noise control (ANC) has been widely utilized to reduce unwanted environmental noise. The primary objective of ANC is to generate an anti-noise with the same amplitude but the opposite phase of the primary noise using the secondary source. However, the effectiveness of the ANC application is impacted by the speaker’s output saturation. This paper proposes a two-gradient direction ANC algorithm with a momentum factor to solve the saturation with faster convergence. In order to make it implemented in real-time, a computation-effective variable step size approach is applied to further reduce the steady-state error brought on by the changing gradient directions. The time constant and step size bound for the momentum two-gradient direction algorithm is analyzed. Simulation results show that the proposed algorithm performs effectively in the time-unvaried and time-varied environment.

Practical active noise control (ANC) systems, like the active noise cancellation headphone, usually adopt a control filter with preset coefficients to achieve satisfactory noise reduction performance for dynamic noise and higher robustness. In this strategy, selecting the appropriate control filter for different types of noise is critical to the noise cancellation performance, and this selection mechanism is typically determined by trial and error. Hence, this article proposes a computation-efficient one-dimensional convolutional neural network capable of selecting the most suitable pre-trained control filter for each distinct primary noise. Applying the similarity matching method allows the proposed model to have a better generalization and can even deal with zero-shot noise, whose class does not exist in the training set. The Large-margin softmax (L-softmax) is also investigated to improve the proposed model's performance. Furthermore, when dealing with the N-shot learning problem, where there are few known real-world noise samples for the ANC system, an additional fine-tuning strategy is used to improve control filter selection accuracy. Numerical simulations on measured primary and secondary paths validate the proposed method's efficacy.

Due to their computational efficiency, least mean square (LMS)–based algorithms are still widely utilized to achieve optimal control in active noise control (ANC) applications. Real-world implementation of advanced ANC functionalities, such as selective cancellation of frequencies, is nonetheless hampered by complexity trade-offs, especially with computationally-expensive frequency-domain approaches. Prevailing time-domain adaptive algorithms – proposed to alleviate complexities from transformation – continue to incur increased complexities while constraining the magnitude of frequency bins in the time-domain filters. To address existing complexities in time-domain approaches, this paper proposes a circulant convolutional penalty factor that assists the extended leaky filtered-reference LMS (FxLMS) algorithm in achieving frequency constraint without any frequency-domain transform. This circulant convolutional penalty factor is readily determined by methods for designing finite response filters, such as frequency sampling. Additionally, the coordinate descent method is adopted to further reduce the proposed algorithm's computations, significantly increasing its feasibility for implementation on conventional real-time processors. Finally, the numerical simulations performed on the measured primary and secondary paths demonstrate the effectiveness of the proposed algorithm.

Active noise control (ANC) technology is commonly used to cancel acoustic noise in daily life. The conventional ANC headphone, being one of the mature commercial products that implement this approach, utilizes microphones on its earcup to pick up the reference signal. However, in a multi-noise source situation, the reference signal mixed with uncorrelated interference usually results in poor noise reduction performance of the ANC system. Hence, we proposed a novel hybrid approach that employs wireless microphones to acquire high signal-to-noise-ratio reference signals from far-end noise sources, increasing coherence and thus improving noise reduction performance. Additionally, an error separation model is applied in the proposed structure to enhance the coherence between the error signal and each adaptive filter. As a result, the proposed hybrid approach to combine wireless and earcup microphones for ANC headphone significantly improves its noise reduction performance when dealing with multi-noise sources. Furthermore, numerical simulation and real-time experiments of the proposed structure have shown that it improves noise reduction performance by 4 − 6 dB when compared to a conventional ANC headphone.

The minimum output variance filtered reference least mean square (MOV-FxLMS) algorithm is a effective algorithm that utilizes the penalty mechanism to help the active noise control (ANC) system achieve noise cancellation with constrained output variance or power. As it can constrain output power, the MOV-FxLMS algorithm can freely determine the ANC system’s control effort, avoiding output saturation, and improving system stability. However, its performance is determined by a penalty factor, which is normally chosen by trial and error. Hence, this work proposes an optimal penalty factor and its feasible estimation that does not require any assumptions of Gaussian reference signal or input independence. This factor assists the MOV-FxLMS in achieving the optimal solution under the target output-variance constraint. Numerical simulations on measured paths demonstrate its effectiveness for various types of noise.

Feedforward active noise control (ANC) is widely utilized to attenuate the broadband noise picked up by the reference microphone. However, in some situations, it is impractical to obtain a clean reference signal when the noise source is far away from the controller. Hence, we adopt a wireless reference microphone to pick up the reference signals around the noise sources. Furthermore, a coherence-based-selection algorithm is proposed to select the reference signals with high coherence. The proposed method improves the quality of the reference signals and the noise reduction performance of the ANC system. Numerical simulations and real-time experiments are conducted to validate the effectiveness of the proposed algorithm.

Output saturation, mainly caused by the power amplifier, is a critical issue influencing the performance and stability of an adaptive system, such as in active noise control. In this paper, a quadratically constrained quadratic program (QCQP) is defined to achieve optimal control under the averaging-output-power constraint, which ensures the output of the system operates linearly and hence, avoids the output saturation. To solve this QCQP problem recursively in practice, this paper utilizes one of the leaky-based filtered-x least mean square algorithm with an optimal leak factor. However, this method only can be applied when the statistical feature of the control signal with maximum output-power is known, which is difficult to obtain in practice. Hence, by incorporating the adaptive inverse modeling technique, we can derive a practical estimation of the optimal leaky factor, which is applicable to different noise types. Furthermore, as the optimal output-constraint control forces the output to operate linearly, the nonlinear amplifier model is not required for the leak factor estimate. The simulation of the proposed algorithm is carried out on measured nonlinear paths to validate its efficacy.

The successful launch of the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) provides a new and advanced tool for sea ice thickness (SIT) estimations in the Arctic. However, the performance of ICESat-2 for SIT estimations still remains unknown. In the present study, SIT estimates derived from ICESat-2 are examined using three retrieval methods, namely, two buoyancy methods with the merged snow depth and empirical snow depth (BMA and BME, respectively) and one empirical estimation method (EEM), and these estimates are compared to near-simultaneous airborne measurements from the IceBird mission in April 2019. Overall, the ICESat-2 total freeboard registers quite well with that from the near-concurrent IceBird mission with a mean bias of 2.5 cm, which demonstrates the high reliability of ICESat-2 data for SIT estimation. However, the much more evident difference between SIT estimations than total freeboard from ICESat-2 and IceBird indicates that other parameters (e.g., snow depth and snow/ice densities) may bring increased uncertainties to the SIT estimation. Overall, BMA is the best method for SIT estimation and has the closest thickness distribution to that of IceBird data with a mean bias of 0.11 m, followed by the BME and EEM methods. The dominate error sources for SIT estimation using the buoyancy method are ice density and snow depth that require further investigation in future studies.

To study the applicability of full polarimetric synthetic aperture radar (SAR) data to identify alpine glaciers in the central Himalayas, six polarimetric decomposition methods were used to obtain 20 polarimetric characteristic parameters based on the Advanced Land Observing Satellite 2 (ALOS-2) Phased Array type L-band SAR (PALSAR) data. Object-oriented multiscale segmentation was performed on a Landsat 8 Operational Land Imager (OLI) image prior to classification, and the vector boundaries of different types of training samples were selected from the segmented results. We performed a support vector machine (SVM)-based classification on the characteristic parameters from each polarimetric decomposition. All 20 parameters were then screened and combined according to different requirements: the degree of separability of different types of training samples and the type of scattering mechanisms. The results show that the classification accuracy of the incoherent decomposition characteristics based on the covariance matrix is the best, reaching 87%, and it can exceed 91% after adding the local incidence angle to the suite of classifiers. Eventually, more than 93% accuracy was achieved using a combination of multiple polarimetric parameters, which reduced the misclassification between bare ice and rock. We also analyzed the use of controlling factors on the accuracy of alpine glacier identification and found that the polarimetric information and aspect of the glacier surface are the most important factors. The former is the main basis for identification but the latter will confuse the feature distributions of different categories and cause misclassification.

Sea ice type is one of the most sensitive variables in Arctic ice monitoring and detailed information about it is essential for ice situation evaluation, vessel navigation, and climate prediction. Many machine-learning methods including deep learning can be employed for ice-type detection, and most classifiers tend to prefer different feature combinations. In order to find the optimal classifier-feature assembly (OCF) for sea ice classification, it is necessary to assess their performance differences. The objective of this letter is to make a recommendation for the OCF for sea ice classification using Cryosat-2 (CS-2) data. Six classifiers including convolutional neural network (CNN), Bayesian, K nearest-neighbor (KNN), support vector machine (SVM), random forest (RF), and back propagation neural network (BPNN) were studied. CS-2 altimeter data of November 2015 and May 2016 in the whole Arctic were used. The overall accuracy was estimated using multivalidation to evaluate the performances of individual classifiers with different feature combinations. Overall, RF achieved a mean accuracy of 89.15%, followed by Bayesian, SVM, and BPNN (~86%), outperforming the worst (CNN and KNN) by 7%. Trailing-edge width (TeW) and leading-edge width (LeW) were the most important features, and feature combination of TeW, LeW, Sigma0, maximum of the returned power waveform (MAX), and pulse peakiness (PP) was the best choice. RF with feature combination of TeW, LeW, Sigma0, MAX, and PP was finally selected as the OCF for sea ice classification and the results that demonstrated this method achieved a mean accuracy of 91.45%, which outperformed the other state-of-art methods by 9%.