Types Of Noise Research Articles

Regularly Truncated M-Estimators for Learning With Noisy Labels.

The sample selection approach is very popular in learning with noisy labels. As deep networks "learn pattern first", prior methods built on sample selection share a similar training procedure: the small-loss examples can be regarded as clean examples and used for helping generalization, while the large-loss examples are treated as mislabeled ones and excluded from network parameter updates. However, such a procedure is arguably debatable from two folds: (a) it does not consider the bad influence of noisy labels in selected small-loss examples; (b) it does not make good use of the discarded large-loss examples, which may be clean or have meaningful information for generalization. In this paper, we propose regularly truncated M-estimators (RTME) to address the above two issues simultaneously. Specifically, RTME can alternately switch modes between truncated M-estimators and original M-estimators. The former can adaptively select small-losses examples without knowing the noise rate and reduce the side-effects of noisy labels in them. The latter makes the possibly clean examples but with large losses involved to help generalization. Theoretically, we demonstrate that our strategies are label-noise-tolerant. Empirically, comprehensive experimental results show that our method can outperform multiple baselines and is robust to broad noise types and levels.

Effect of Modulated Masking on Cortical Auditory Evoked Potential in Normal Hearing Individuals: A Systematic Review and Meta-Analysis

Abstract Introduction The study of electrophysiological auditory measures with different types of masking makes it possible to understand temporal processing skills and the processes involved in speech recognition in noise situations. The use of modulated masking in cortical measures of hearing enables the obtainment of analysis parameters of the masking release and its impact on neural auditory processing. Objective To investigate the behavior of cortical auditory evoked potentials (CAEPs) with modulated masking in the normal hearing population. Data synthesis A total of 2,159 articles were identified in the initial search; of these, 12 were selected for full reading. After excluding studies that did not meet the eligibility criteria, six articles were included in the present systematic review.The results show that the type of masking has an influence on cortical auditory behavior, indicating a different effect on neural posture rergarding CAEP responses. Modulated noise as masking in the CAEP record generated statistically higher and earlier responses compared with non-modulated/steady noise, evidenced by the results obtained in the meta-analysis with subgroup analysis. These responses may indicate an influence of the type of noise in the neural auditory coding. Conclusion Better responses were observed in modulated masking in terms of the behavior of CAEPs. Decreased latency and increased amplitude of cortical measurements with the use of modulated noise indicate a lower masking effect of this noise in cortical auditory processing, evidencing the masking release phenomenon.

Efficient and robust phase unwrapping method based on SFNet

Phase unwrapping is a crucial step in obtaining the final physical information in the field of optical metrology. Although good at dealing with phase with discontinuity and noise, most deep learning-based spatial phase unwrapping methods suffer from the complex model and unsatisfactory performance, partially due to simple noise type for training datasets and limited interpretability. This paper proposes a highly efficient and robust spatial phase unwrapping method based on an improved SegFormer network, SFNet. The SFNet structure uses a hierarchical encoder without positional encoding and a decoder based on a lightweight fully connected multilayer perceptron. The proposed method utilizes the self-attention mechanism of the Transformer to better capture the global relationship of phase changes and reduce errors in the phase unwrapping process. It has a lower parameter count, speeding up the phase unwrapping. The network is trained on a simulated dataset containing various types of noise and phase discontinuity. This paper compares the proposed method with several state-of-the-art deep learning-based and traditional methods in terms of important evaluation indices, such as RMSE and PFS, highlighting its structural stability, robustness to noise, and generalization.

Investigation of the Influence of Machining Parameters and Surface Roughness on the Wettability of the Al6082 Surfaces Produced with WEDM.

Electrical Discharge Machining (EDM) is a non-conventional machining technique, capable of processing any kind of conductive material. Recently, it has been successfully utilized for producing hydrophobic characteristics in inherently hydrophilic metallic materials. In this work, Wire Electrical Discharge Machining (WEDM) was utilized for producing hydrophobic characteristics on the surface of the aluminum alloy 6082, and various parameters that can affect wettability were investigated. Adopting an orthogonal Taguchi approach, the effects of the process parameter values of peak current, pulse-on time, and gap voltage on the contact angles of the machined surfaces were investigated. After machining, all samples were observed to have obtained hydrophobic properties, reaching contact angles up to 132°. The peak current was identified as the most influential parameter regarding the contact angle, while the gap voltage was the less influential parameter. A contact angle variation of 30° was observed throughout different combinations of machining parameters. Each combination of the machining parameters resulted in a distinct surface morphology. The samples with moderate roughness values (3.4 μm > Sa > 5.7 μm) were found to be more hydrophobic than the samples with high or low values, where the contact angle was measured under 115°. In addition, the finite element modeling of the experimental setup, with parametric surfaces of uniform random and Perlin noise types of roughness, was implemented. Time dependent simulations coupling phase field and laminar flow for the modelingof the wetting of surfaces with different surface roughness characteristics showed that an increase in the Sa roughness and total wetted area can lead to an increase in the contact angle. The combination of experimental and computational results suggests that the complexity of the wettability outcomes of aluminum alloy surfaces processed with WEDM lies in the interplay between variations of the surface chemical composition, roughness, micro/nano morphology, and the surface capability of forming a composite air/water interface.

A Simplified Frequency-Domain Feedback Active Noise Control Algorithm

When the adaptive filter length is increased, the calculation complexity increases rapidly because the relationship between the calculation and the adaptive filter length N contains a power function with no secondary path identification algorithm. Under the basic premise of unreduced noise reduction, herein, a simplified frequency-domain feedback active noise control algorithm is proposed. To reduce the computation complexity, the total delay is adopted as the estimated secondary path; the filtered reference signal is produced in the frequency domain by using multiplication to replace convolution calculation in the time domain and then updating the adaptive filter coefficients in the frequency domain. Therefore, the computational complexity has a logarithmic function with the increased adaptive filter length in the proposed algorithm. If the adaptive filter length is 512, the existing WSMANC algorithm’s calculation is 271,360 real number multiplications, while that of the proposed algorithm is only 38,912 real number multiplications. To verify the proposed algorithm’s stability, convergence speed, and noise reduction, the single-frequency noise, narrowband white noise, and narrowband pink noise, respectively, are used as the primary noise types in the simulations. The results show that (1) the proposed SFDFBANC algorithm can obtain similar noise reduction performance to existing algorithm, (2) the convergence rate is faster than existing algorithm, and (3) if the adaptive filter length is more than 64, the proposed algorithm exhibits a lower computational complexity.

Plant-Denoising-Net (PDN): A plant point cloud denoising network based on density gradient field learning

Effective point cloud denoising is critical in 3D plant phenotyping applications, which reduces interference in subsequent algorithms and improves the accuracy of plant phenotypes measurement. Deep learning-based point cloud denoising algorithms have shown excellent denoising performance on simple CAD models. However, these algorithms suffer from issues including over-smoothing or shrinkage and low efficiency when applied on density uneven, incomplete, various types of noise and complex plant point clouds. We proposed a plant point cloud denoising network (PDN) based on point cloud density gradient field learning, which can effectively address the challenges posed by plant point clouds. PDN consists of three main modules: point density feature (PDF) exception module, umbrella operator feature (UOF) computation module, and point density gradient (DG) estimation module. The performance of PDN was evaluated in experiments using point clouds of multiple plant species with noise of different types. Under different levels of Gaussian noise, our method achieved a relative performance improvement of 7.6%-19.3% compared to the state-of-the-art baseline methods, reaching state-of-the-art denoising performance. For noise of different types, the majority of our denoising results outperformed the baseline methods. In addition, our method was 0.5 and 8.6 times faster than the baseline methods when processing point clouds with low and high noise level, respectively. The good robustness, generalization, and computational efficacy of PDN are expected to facilitate the acquisition of high-precision 3D point clouds for various plant species, enhance the versatility of 3D phenotyping methods, improve the accuracy of the measurement of structural phenotypes, and increase the throughput of data processing, therefore facilitate the development of modern breeding research. The source code and the datasets used in this study is available on GitHub at https://github.com/suetme/PDN-plant-denoising-net.

A novel approach for denoising electrocardiogram signals to detect cardiovascular diseases using an efficient hybrid scheme

BackgroundElectrocardiogram (ECG) signals are inevitably contaminated with various kinds of noises during acquisition and transmission. The presence of noises may produce the inappropriate information on cardiac health, thereby preventing specialists from making correct analysis.MethodsIn this paper, an efficient strategy is proposed to denoise ECG signals, which employs a time-frequency framework based on S-transform (ST) and combines bi-dimensional empirical mode decomposition (BEMD) and non-local means (NLM). In the method, the ST maps an ECG signal into a subspace in the time frequency domain, then the BEMD decomposes the ST-based time-frequency representation (TFR) into a series of sub-TFRs at different scales, finally the NLM removes noise and restores ECG signal characteristics based on structural self-similarity.ResultsThe proposed method is validated using numerous ECG signals from the MIT-BIH arrhythmia database, and several different types of noises with varying signal-to-noise (SNR) are taken into account. The experimental results show that the proposed technique is superior to the existing wavelet based approach and NLM filtering, with the higher SNR and structure similarity index measure (SSIM), the lower root mean squared error (RMSE) and percent root mean square difference (PRD).ConclusionsThe proposed method not only significantly suppresses the noise presented in ECG signals, but also preserves the characteristics of ECG signals better, thus, it is more suitable for ECG signals processing.

Optical-coherence-tomography-based deep-learning scatterer-density estimator using physically accurate noise model

We demonstrate a deep-learning-based scatterer density estimator (SDE) that processes local speckle patterns of optical coherence tomography (OCT) images and estimates the scatterer density behind each speckle pattern. The SDE is trained using large quantities of numerically simulated OCT images and their associated scatterer densities. The numerical simulation uses a noise model that incorporates the spatial properties of three types of noise, i.e., shot noise, relative-intensity noise, and non-optical noise. The SDE’s performance was evaluated numerically and experimentally using two types of scattering phantom and in vitro tumor spheroids. The results confirmed that the SDE estimates scatterer densities accurately. The estimation accuracy improved significantly when compared with our previous deep-learning-based SDE, which was trained using numerical speckle patterns generated from a noise model that did not account for the spatial properties of noise.

Generating a maximally entangled state via a pure global noise environment

We studied the entangling power of two pure global noises, i.e. amplitude damping noise and classic noise. The entangling power of the two-qubit amplitude damping global noise periodically oscillates with time. Additionally, the entangling power of two-qubit global classical noise increases exponentially with time. The maximum entangling power of both of them exceeds that of the perfect entanglers. Based on this, we propose the conditions for generating a maximally entangled state with global noise acting on a two-qubit separable state. Only if the two-qubit composite system, which is initially in one of those product states: 12(|0⟩±|1⟩)⊗12(|0⟩±|1⟩) , |10⟩ and |01⟩ suffers an amplitude damping global noise, can we prepare this system in the maximally entangled state by appropriately controlling the evolution time of amplitude damping. Finally, we investigate the disentanglement of the maximum entangled Bell state using these two types of global noise. The two global noises cannot completely disentangle the Bell states Φ± . Under the influence of amplitude damping global noise, the entanglement of Bell state Ψ+ undergoes a cyclical variation, alternating between disappearance and recovery to reach maximum entanglement within one period. The entanglement of either Bell state Ψ+ or Ψ− is completely independent of global classical noise. The entanglement of Bell state Ψ− is also robust against amplitude damping global noise.

A Picometre-Level Resolution Test Method without Nonlinearity for Heterodyne Interferometer Measurement Electronics

The wide application of displacement measurement in high-precision equipment production and high-precision metrology is placing increasing pressure on the resolution of heterodyne interferometers. However, as the core component of an interferometer, since measurement electronics includes the cross-physical process of photoelectric conversion, its resolution is rarely evaluated, either on an individual level or as a whole. Therefore, in this paper, we propose a picometer resolution test method for measurement electronics, that uses intensity modulation signals based on an AOM to replace the beat frequency interference signals, and an ordinary commercial guide rail to equivalently generate the pm-level displacement of the heterodyne interferometer under laboratory conditions. Based on the detailed analysis of the type of noise in the test device, the correlation between the light intensity and the nonlinear error was established, and nonlinearity was suppressed to 10% of the original level. Furthermore, this test method allows one to perform a 0.1 mrad phase step test at 1 MHz signal frequency, equivalent to a 2.5 pm resolution test in a double-pass heterodyne interferometer. Simultaneously, it can be directly applied to the resolution test for measurement electronics with a center frequency in the range of 1 MHz to 20 MHz.

Adaptive Local Flaw Detection Based on Magnetic Flux Leakage Images With a Noise Distortion Effect for Steel Wire Ropes

Steel wire rope (SWR) is widely used in industrial scenarios because of its high strength and toughness. To avoid accidents that local flaws (LFs) can cause, SWRs should be inspected regularly. Magnetic flux leakage (MFL) image detection, which is a common SWR inspection method, is inevitably influenced by shaking noise and strand noise. Although noise-reduction methods for these two types of noise have been proposed, the phenomenon of noise distortion has not received enough attention. This paper investigates the noise distortion phenomenon on the MFL image and finds that the distorted noise severely affects the performance of existing noise-feature-oriented (NFO) denoising methods. To solve this problem, we adopt a target-feature-oriented (TFO) denoising method. Specifically, the removal of noise is avoided, instead, an LF-enhancement-based denoising process is proposed. Moreover, to localize LFs in denoised images, a three-stage adaptive localization method mainly based on disjoint region analysis is proposed. The experiment results show that the proposed TFO denoising method significantly improves the denoising performance for images that distorted noise influences. In addition, the proposed adaptive localization method improves the intelligence of the localization process and has better localization performance for images with distorted noise.

An Algorithm to Estimate the Power Spectral Density From Allan Deviation.

Complex architectures for wireless communications, digital electronics, and space-based navigation interlink several oscillator-based devices such as clocks, transponders, and synthesizers. Estimators characterizing their stability are critical for addressing the impact of random fluctuations (noise) on the overall system performance. Manufacturers typically specify this as an Allan/Hadamard Variance (AVAR/HVAR) profile in the time domain. However, stochastic processes constituting the noise are more thoroughly described in the frequency domain by the power spectral density function (PSD). Both are second-moment measures of the time series, but it is only possible to translate unambiguously from the PSD to the AVAR/HVAR, not vice versa, except in the case of a single noise type, a rather unrealistic case. This note presents an analytical method to generate an approximated PSD expressed as a set of power-laws defined in specific intervals in the frequency domain, starting from an AVAR/HVAR expressed as a set of power-laws in the time domain. The proposed algorithm is straightforward to implement, applicable to all noise types (and combinations thereof), and can be self-validated by reconstructing the corresponding AVAR/HVAR by direct computation. Coupling with well-established algorithms relying on the PSD for power-law noise generation, the ensuing method encompasses the capability for generating multicolored noise in end-to-end simulations, as demonstrated hereby for NASA's deep space atomic clock. We also report on the limitations of the algorithm and analytical expressions of the continuous version of the algorithm.

Hierarchical remote preparation with multiple agents under the non-Markovian and Markovian noises

To meet the practical needs that the capabilities of agents in a quantum communication network are often different, we propose two deterministic schemes to achieve hierarchical remote preparation of an arbitrary single-qubit state by using cluster state as the entangled resource. One requires splitting the coefficients of the prepared state and the other does not due to the elaborately constructed measurement basis. The high-grade agent requires the help of all the other high-grade agents and any one of the low-grade agents to reconstruct the target state, while the low-grade agent needs the assistance of all the remaining agents. Comparing with the original scheme, ours are applicable to any number of agents and derive general expressions of recovery operators which clearly reveal the relationship with the measurement results. Moreover, the influences of diverse types of non-Markovian and Markovian noises are considered through calculating the fidelity of the output state.

Robustness of the emergence of synchronized clusters in branching hierarchical systems under parametric noise.

The dynamical robustness of networks in the presence of noise is of utmost fundamental and applied interest. In this work, we explore the effect of parametric noise on the emergence of synchronized clusters in diffusively coupled Chaté-Manneville maps on a branching hierarchical structure. We consider both quenched and dynamically varying parametric noise. We find that the transition to a synchronized fixed point on the maximal cluster is robust in the presence of both types of noise. We see that the small sub-maximal clusters of the system, which coexist with the maximal cluster, exhibit a power-law cluster size distribution. This power-law scaling of synchronized cluster sizes is robust against noise in a broad range of coupling strengths. However, interestingly, we find a window of coupling strength where the system displays markedly different sensitivities to noise for the maximal cluster and the small clusters, with the scaling exponent for the cluster distribution for small clusters exhibiting clear dependence on noise strength, while the cluster size of the maximal cluster of the system displays no significant change in the presence of noise. Our results have implications for the observability of synchronized cluster distributions in real-world hierarchical networks, such as neural networks, power grids, and communication networks, that necessarily have parametric fluctuations.

COMPARISON AND EVALUATION OF LMS-DERIVED ALGORITHMS APPLIED ON ECG SIGNALS CONTAMINATED WITH MOTION ARTIFACT DURING PHYSICAL ACTIVITIES

The acquisition of ECG signals offers physicians and specialists a very important tool in the diagnosis of cardiovascular diseases. However, very often these signals are affected by noise from various sources, including noise generated by movement during physical activity. This type of noise is known as Motion Artifact (MA) which changes the waveform of the signal, leading to erroneous readings. The elimination of this noise is performed by different filtering techniques, where the adaptive filtering using the LMS (least mean squares) algorithm stands out. The objective of this article is to determine which algorithms best deal with motion artifacts, taking into account the use of instruments or wearable equipment, in different conditions of physical activity. A comparison between different algorithms derived from LMS (NLMS, PNLMS and IPNLM) used in adaptive filtering is carried out using indicators such as: Pearson's Correlation Coefficient, Signal to Noise Ratio (SNR) and Mean Squared Error (MSE) as metrics to evaluate them. For this purpose, the mHealth database was used, which contains ECG signals taken during moderate to medium intensity physical activities. The results show that filtering by IPNLMS as well as PNLMS offers an improvement both visually and in terms of SNR, Pearson, and MSE indicators.

Real-time implementation and explainable AI analysis of delayless CNN-based selective fixed-filter active noise control

The selective fixed-filter active noise control (SFANC) approach can select suitable pre-trained control filters for different types of noise. With the learning ability of convolutional neural network (CNN), the CNN-based SFANC method can automatically learn its parameters from noise data. Combining practical experience, this paper abstracts ANC as a Markov progress and provides a detailed theoretical analysis to verify the reasonableness of the CNN-based SFANC method. To validate its effectiveness, we implement the method in a multichannel ANC window, where the CNN operating in the co-processor collaborates with the real-time controller to realize delayless noise control. Additionally, an explainable AI technique is used to analyze the underlying principle of the CNN-based SFANC method, enhancing its interpretability in acoustic applications. Numerical simulations and real-time experiments demonstrate that the CNN-based SFANC method achieves not only satisfactory noise reduction performance for broadband and real-world noises but also excellent transferability.11The code will be accessible at https://github.com/Luo-Zhengding/SFANC-Window.

A stochastic Susceptible Vaccinees Infected Recovered epidemic model with three types of noises

Epidemics pose a serious threat to public health, and effective disease control measures are essential. To address this challenge, we propose a novel stochastic Susceptible Vaccinees Infected Recovered (SVIR) epidemic model that incorporates temporary immunities, media coverage, and three types of noises: white noise, telegraph noise, and L vy noise. Our model is more realistic than traditional epidemic models, as it accounts for the random fluctuations in disease transmission and the impact of media coverage on public behavior. We use Lyapunov functions and stopping times to prove the existence of a global positive solution to our model. We also derive sufficient conditions for the extinction and persistence of the infection. Furthermore, we develop a numerical scheme to validate our theoretical findings and to analyze the influence of various parameters on the model's dynamics. Our model can be used to guide policymakers in developing and implementing effective disease control measures.

Learning Accurate Pseudo-Labels via Feature Similarity in the Presence of Label Noise

Due to the exceptional learning capabilities of deep neural networks (DNNs), they continue to struggle to handle label noise. To address this challenge, the pseudo-label approach has emerged as a preferred solution. Recent works have achieved significant improvements by exploring the information involved in DNN predictions and designing a straightforward method to incorporate model predictions into the training process by using a convex combination of original labels and predictions as the training targets. However, these methods overlook the feature-level information contained in the sample, which significantly impacts the accuracy of the pseudo label. This study introduces a straightforward yet potent technique named FPL (feature pseudo-label), which leverages information from model predictions as well as feature similarity. Additionally, we utilize an exponential moving average scheme to bolster the stability of corrected labels while upholding the stability of pseudo-labels. Extensive experiments were carried out on synthetic and real datasets across different noise types. The CIFAR10 dataset yielded the highest accuracy of 94.13% (Top1), while Clothing1M achieved 73.54%. The impressive outcomes showcased the efficacy and robustness of learning amid label noise.

$$C^{\infty }$$-Regularization by Noise of Singular ODE’s

AbstractIn this paper we construct a new type of noise of fractional nature that has a strong regularizing effect on differential equations. We consider an equation driven by a highly irregular vector field and study the effect of this noise on such dynamical systems. We employ a new method to prove existence and uniqueness of global strong solutions, where classical methods fail because of the “roughness” and non-Markovianity of the driving process. In addition, we prove the rather remarkable property that such solutions are infinitely many times classically differentiable with respect to the initial condition in spite of the vector field being discontinuous. The technique used in this article corresponds, in a certain sense, to the Nash–Moser iterative scheme in combination with a new concept of “higher order averaging operators along highly fractal stochastic curves”. This approach may provide a general principle for the study of regularization by noise effects in connection with important classes of partial differential equations.

Unlocking the Power of Open Set: A New Perspective for Open-Set Noisy Label Learning

Learning from noisy data has attracted much attention, where most methods focus on closed-set label noise. However, a more common scenario in the real world is the presence of both open-set and closed-set noise. Existing methods typically identify and handle these two types of label noise separately by designing a specific strategy for each type. However, in many real-world scenarios, it would be challenging to identify open-set examples, especially when the dataset has been severely corrupted. Unlike the previous works, we explore how models behave when faced with open-set examples, and find that a part of open-set examples gradually get integrated into certain known classes, which is beneficial for the separation among known classes. Motivated by the phenomenon, we propose a novel two-step contrastive learning method CECL (Class Expansion Contrastive Learning) which aims to deal with both types of label noise by exploiting the useful information of open-set examples. Specifically, we incorporate some open-set examples into closed-set classes to enhance performance while treating others as delimiters to improve representative ability. Extensive experiments on synthetic and real-world datasets with diverse label noise demonstrate the effectiveness of CECL.