Real-world Noise Research Articles

Detection of ATTR cardiac amyloidosis using a novel artificial intelligence algorithm for wearable-adapted noisy single-lead electrocardiograms

Abstract Background Amyloid cardiomyopathy (ATTR-CM) is often underdiagnosed, with very few diagnosed before the onset of symptoms, leading to delayed use of disease-modifying therapies that could have substantial prognostic implications early in the disease course. In the US, the disease has a substantially higher prevalence among Black adults, who are further challenged by lower access to healthcare services and lower preventative care, further reducing the ability for clinician-led diagnosis. We sought to develop an approach for identifying ATTR-CM on noisy single-lead ECGs obtainable on portable and wearable ECG devices, allowing broad use in community screening. Purpose To develop a portable/wearable device-adapted AI-ECG algorithm capable of detecting ATTR-CM on noisy single-lead ECGs. Methods We identified patients with ATTR-CM from 2015 through June 2023 across 5 hospitals of a large U.S.-based hospital system using positive bone scintigraphy scans or pharmacotherapy with an approved transthyretin stabilizer. In the development cohort, we used lead I of 12 lead ECGs, commonly captured by portable and wearable ECG devices, to train a novel noise-adapted model explicitly designed to infer information against simulated real-world noises. For this, ECG signals were augmented using random Gaussian real-world noise within four frequency ranges at multiple signal-to-noise ratios. We tested the model in an independent set of cases and matched controls drawn from July through December 2023. Results The development cohort consisted of 1,011 ECGs from 234 patients (median age 79 years [IQR:70-85], 17% women), with 10:1 age- and sex-matched 10,110 controls with 10,110 ECGs (age 79 [IQR:70-85] years 17.7% female). In the independent test set of 139 ECGs from 47 patients (age 80 [75-86] years, 32% women) and matched controls (1390 ECGs and patients) from July through December 2023, the AUROC (area under the receiver operating characteristic curve) of the AI-ECG model for ATTR-CM was 0.90 [95%CI: 0.88-0.92] (A). The AUPRC was 0.44 [0.36-0.53]. The sensitivity and specificity of the model were 0.85 and 0.80. The positive predictive value ranged from 0.12 at a 3% prevalence level to 0.59 at a 25% prevalence (B). Conclusions A novel portable and wearable-devices adapted model demonstrates promising performance for detecting ATTR-CM on single-lead ECGs, representing a more accessible method for screening in community-dwelling adults.

MWIR image deep denoising reconstruction based on single-pixel imaging

Single-pixel imaging enables cost-effective and highly sensitive mid-wave infrared imaging, with the added convenience facilitated by deep learning algorithms. However, most networks train separate models depending on compression rates, frequently neglecting real-world infrared noise, leading to inconveniences and suboptimal reconstruction performance. To address these issues, we propose to integrate intrinsic and external deep denoising priors for the reconstruction of compressive mid-wave infrared images from single-pixel measurements. The generalized alternating projection is unfolded into a designed network to harness the benefits of robustness and internal priors. Furthermore, to mitigate real-world noise in measurements, a Swin Transformer U-shaped network that embeds noise variance is introduced to learn external denoise prior. A comparative analysis of simulated and real infrared single-pixel imaging in experiments highlights the promising performance of the proposed method.

Attention-based convolutional neural network for multi-channel active noise control

Multi-channel active noise control systems utilize multiple control units to cancel unwanted noise across different channels simultaneously. By employing adaptive filtering techniques, these systems can effectively reduce noise levels in complex environments, offering improved noise mitigation compared to single-channel active noise control approaches. The disadvantage of adaptive filtering in active noise control is its limited ability to accurately model nonlinearities present in real-world noise environments. In this work, we propose an attention-based convolutional neural network comprising convolutional, deconvolutional, and attention mechanism to approach multi-channel active noise control. The system simultaneously calculates multiple canceling signals to eliminate primary noises captured by error microphones. To evaluate the effectiveness of our proposed model, we conducted testing utilizing filtered white noise spanning various frequency bands. Experimental evaluations across various setups, including different loudspeaker and microphone configurations, as well as variations in secondary source positioning, are used to assess the performance of the proposed approach. Results demonstrate that the proposed attention-based convolutional neural network for multi-channel active noise control yields a noise reduction of 22.65 dB on wideband noise, while the conventional Filtered-x least mean square approach yields a noise reduction of 4.23 dB, indicating the effectiveness of the proposed approach.

Learning real-world heterogeneous noise models with a benchmark dataset

Noise modeling is an important research field in computer vision; the design of an accurate model for imaging sensor noise depends on not only a comprehensive benchmark dataset of the real world, but also a precise design of the noise modeling algorithm. However, due to the inaccurate estimation method of noise-free images and limited shooting scenes, the current realistic datasets could not describe the diverse noise properties sufficiently. Moreover, popular parametric noise models are not sophisticated enough to characterize the real-world noise exactly. In this work, we first construct a more comprehensive dataset of the real world by capturing more indoor and outdoor scenes under different lighting conditions using diverse smartphones, then we propose a non-parametric noise estimation method capable of modeling the spatial heterogeneity of real-world noise patterns. Specifically, in order to characterize the spatial heterogeneity of real-world noise, we assume a non-i.i.d Gaussian distribution and propose a deep convolutional neural network (DCNN)-based approach for learning pixel-wise noise variance maps. To learn the pixel-wise variance map, we have constructed a variance estimation network mapping from the conditional signals (clean image, ISO, and camera model) to surrogate labels obtained from the nonlocal search of similar patches from the clean-noisy image pair. Finally, we conducted denoising and classification experiments using different kinds of simulated noisy images, compared to the Poisson-Gaussian and Noise Flow noise models, the proposed method achieves denoising performance improvements (PSNR) of 1.13 dB and 2.51 dB respectively on the proposed real-world test dataset, denoising and classification results on the real noisy data captured by mobile phones have verified that our approach is more accurate than current noise modeling methods.

Simple Debiased Contrastive Learning for Sequential Recommendation

Recently, contrastive learning has been shown to be effective in improving sequential recommendation, alleviating noise and long-tail issues caused by real-world noise and data sparsity, and enabling models to learn high-quality user representations. The data augmentation approach of contrastive learning mainly includes randomly changing the sequence data and adding randomly perturbations, and by pulling the distance between the original sample and the positive sample and pushing the distance between the original sample and the negative sample to enhance alignment and to optimize the uniformity of representation space. However, these approaches may use false positive/negative samples, learn biased user representations, and compromise the uniformity of the representation space, leading to second-best recommendation results. To solve these issues, we propose a generic Simple Debiased Contrastive Learning for Sequential Recommendation framework (SimDCL), to alleviate these biased samples, optimize the uniformity of the representation space, and enhance anti-noise ability. Specifically, we augment the samples based on noise, and optimize the uniformity of the representation space through gradient algorithms, and design an object filtering approach to punish false positive and false negative samples in user interaction sequences. Finally, we combine contrastive learning and sequential recommendation for multi-task collaborative training to improve the model’s anti-noise capability and the quality of recommendation. The empirical studies on four benchmark datasets prove the superiority of our approach, and the code and datasets are available at URL.

From lab to life: assessing the impact of real-world interactions on the operation of rapid serial visual presentation-based brain-computer interfaces

Objective. Brain-computer interfaces (BCI) have been extensively researched in controlled lab settings where the P300 event-related potential (ERP), elicited in the rapid serial visual presentation (RSVP) paradigm, has shown promising potential. However, deploying BCIs outside of laboratory settings is challenging due to the presence of contaminating artifacts that often occur as a result of activities such as talking, head movements, and body movements. These artifacts can severely contaminate the measured EEG signals and consequently impede detection of the P300 ERP. Our goal is to assess the impact of these real-world noise factors on the performance of a RSVP-BCI, specifically focusing on single-trial P300 detection. Approach. In this study, we examine the impact of movement activity on the performance of a P300-based RSVP-BCI application designed to allow users to search images at high speed. Using machine learning, we assessed P300 detection performance using both EEG data captured in optimal recording conditions (e.g. where participants were instructed to refrain from moving) and a variety of conditions where the participant intentionally produced movements to contaminate the EEG recording. Main results. The results, presented as area under the receiver operating characteristic curve (ROC-AUC) scores, provide insight into the significant impact of noise on single-trial P300 detection. Notably, there is a reduction in classifier detection accuracy when intentionally contaminated RSVP trials are used for training and testing, when compared to using non-intentionally contaminated RSVP trials. Significance. Our findings underscore the necessity of addressing and mitigating noise in EEG recordings to facilitate the use of BCIs in real-world settings, thus extending the reach of EEG technology beyond the confines of the laboratory.

Noise4Denoise: Leveraging noise for unsupervised point cloud denoising

Existing deep learning-based point cloud denoising methods are generally trained in a supervised manner that requires clean data as ground-truth labels. However, in practice, it is not always feasible to obtain clean point clouds. In this paper, we introduce a novel unsupervised point cloud denoising method that eliminates the need to use clean point clouds as groundtruth labels during training. We demonstrate that it is feasible for neural networks to only take noisy point clouds as input, and learn to approximate and restore their clean versions. In particular, we generate two noise levels for the original point clouds, requiring the second noise level to be twice the amount of the first noise level. With this, we can deduce the relationship between the displacement information that recovers the clean surfaces across the two levels of noise, and thus learn the displacement of each noisy point in order to recover the corresponding clean point. Comprehensive experiments demonstrate that our method achieves outstanding denoising results across various datasets with synthetic and real-world noise, obtaining better performance than previous unsupervised methods and competitive performance to current supervised methods.

The attentive reconstruction of objects facilitates robust object recognition.

Humans are extremely robust in our ability to perceive and recognize objects-we see faces in tea stains and can recognize friends on dark streets. Yet, neurocomputational models of primate object recognition have focused on the initial feed-forward pass of processing through the ventral stream and less on the top-down feedback that likely underlies robust object perception and recognition. Aligned with the generative approach, we propose that the visual system actively facilitates recognition by reconstructing the object hypothesized to be in the image. Top-down attention then uses this reconstruction as a template to bias feedforward processing to align with the most plausible object hypothesis. Building on auto-encoder neural networks, our model makes detailed hypotheses about the appearance and location of the candidate objects in the image by reconstructing a complete object representation from potentially incomplete visual input due to noise and occlusion. The model then leverages the best object reconstruction, measured by reconstruction error, to direct the bottom-up process of selectively routing low-level features, a top-down biasing that captures a core function of attention. We evaluated our model using the MNIST-C (handwritten digits under corruptions) and ImageNet-C (real-world objects under corruptions) datasets. Not only did our model achieve superior performance on these challenging tasks designed to approximate real-world noise and occlusion viewing conditions, but also better accounted for human behavioral reaction times and error patterns than a standard feedforward Convolutional Neural Network. Our model suggests that a complete understanding of object perception and recognition requires integrating top-down and attention feedback, which we propose is an object reconstruction.

Typicality- and instance-dependent label noise-combating: a novel framework for simulating and combating real-world noisy labels for endoscopic polyp classification

Learning with noisy labels aims to train neural networks with noisy labels. Current models handle instance-independent label noise (IIN) well; however, they fall short with real-world noise. In medical image classification, atypical samples frequently receive incorrect labels, rendering instance-dependent label noise (IDN) an accurate representation of real-world scenarios. However, the current IDN approaches fail to consider the typicality of samples, which hampers their ability to address real-world label noise effectively. To alleviate the issues, we introduce typicality- and instance-dependent label noise (TIDN) to simulate real-world noise and establish a TIDN-combating framework to combat label noise. Specifically, we use the sample’s distance to decision boundaries in the feature space to represent typicality. The TIDN is then generated according to typicality. We establish a TIDN-attention module to combat label noise and learn the transition matrix from latent ground truth to the observed noisy labels. A recursive algorithm that enables the network to make correct predictions with corrections from the learned transition matrix is proposed. Our experiments demonstrate that the TIDN simulates real-world noise more closely than the existing IIN and IDN. Furthermore, the TIDN-combating framework demonstrates superior classification performance when training with simulated TIDN and actual real-world noise.

Comparative Study on Noisy Speech Preprocessing Algorithms

Speech Enhancement and noise reduction have wide applications in speech processing. They are often employed as pre-processing stage in various applications. The work to be presented in this paper is denoising a single-channel speech signal at the presence of a highly non-stationary background noise in order to improve the perceptible quality and intelligibility of the speech. Real world noise is mostly highly non-stationary and does not affect the speech signal uniformly over the spectrum. This paper investigates various Discrete Fourier Transform-based algorithms as single-channel pre-processing techniques consisting of: Spectral Subtraction using over-subtraction and spectral floor, Multi-Band Spectral Subtraction (MBSS), Wiener Filter, MMSE of Short-Time Spectral Amplitude (MMSE-STSA) estimator with, and without using SPU modifier, MMSE Log-Spectral Amplitude Estimator with, and without using SPU modifier, Optimally-Modified Log-Spectral Amplitude estimator (OM-LSA). The processed speeches from these algorithms are compared at the same set of conditions using visual examinations of signals in the time domain and the spectrograms, and also the objective and subjective tests for quality and perceptual evaluation. All the implemented algorithms provide considerable, different degrees of flexibility and control on noise elimination levels that reduces artifacts in the enhanced speech, resulting in the improved quality, and intelligibility.

A Smartphone-Based M-Health Monitoring System for Arrhythmia Diagnosis.

Deep learning technology has been widely adopted in the research of automatic arrhythmia detection. However, there are several limitations in existing diagnostic models, e.g., difficulties in extracting temporal information from long-term ECG signals, a plethora of parameters, and sluggish operation speed. Additionally, the diagnosis performance of arrhythmia is prone to mistakes from signal noise. This paper proposes a smartphone-based m-health system for arrhythmia diagnosis. First, we design a cycle-GAN-based ECG denoising model which takes real-world noise signals as input and aims to produce clean ECG signals. In order to train its two generators and two discriminators simultaneously, we explore an unsupervised pre-training strategy to initialize the generator and accelerate the convergence speed during training. Second, we propose an arrhythmia diagnosis model based on the time convolution network (TCN). This model can identify 34 common arrhythmia events using eight-lead ECG signals, and we deploy such a model on the Android platform to develop an at-home ECG monitoring system. Experimental results have demonstrated that our approach outperforms the existing noise reduction methods and arrhythmia diagnosis models in terms of denoising effect, recognition accuracy, model size, and operation speed, making it more suitable for deployment on mobile devices for m-health monitoring services.

LaplaceConfidence: A graph-based approach for learning with noisy labels

Real-world machine learning applications seldom provide perfect labeled data, posing a challenge in developing models robust to noisy labels. Recent methods prioritize noise filtering based on the discrepancies between model predictions and the provided noisy labels, assuming samples with minimal classification losses to be clean. In this work, we capitalize on the consistency between the learned model and the complete noisy dataset, employing the data’s rich representational and topological information. We introduce LaplaceConfidence, a method that to obtain label confidence (i.e., clean probabilities) utilizing the Laplacian energy. Specifically, it first constructs graphs based on the feature representations of all noisy samples and minimizes the Laplacian energy to produce a low-energy graph. Clean labels should fit well into the low-energy graph while noisy ones should not, allowing our method to determine data’s clean probabilities. Furthermore, LaplaceConfidence is embedded into a holistic method for robust training, where co-training technique generates unbiased label confidence and label refurbishment technique better utilizes it. We also explore the dimensionality reduction technique to accommodate our method on large-scale noisy datasets. Our experiments demonstrate that LaplaceConfidence outperforms state-of-the-art methods on benchmark datasets under both synthetic and real-world noise. Code available at https://github.com/chenmc1996/LaplaceConfidence.

TrapCog: An Anti-Noise, Transferable, and Privacy-Preserving Real-Time Mobile User Authentication System With High Accuracy

The authentication technology of mobile device users has been studied for decades. To balance security, privacy, and usability, motion sensors-based user authentication methods are widely investigated in recent years. However, existing studies meet the problems such as scarcity of training samples, underutilization of data, poor de-noising ability, insufficient transferability, privacy leakage, and low accuracy. To overcome these difficulties, we propose a system, called for Transferable, Real-time, Anti-noise, and Privacy-preserving mobile user COGnition., with the following capabilities: 1) In the phase of data collection, can eliminate man-made noise (mislabeling) through differential training based on down-sampling. 2) In the model training stage, the siamese neural network with Long Short-Term Memory (LSTM) as the sub-network is used to achieve sufficient coverage of sample patterns and the transferability of the model. 3) In the phase of real-world authentication, the privacy of the user is tremendously protected through end-side model deployment and local authentication. Experimental results on a dataset composed of 1,513 users with real-world noise show that TRAP COG has high accuracy and strong transferability, which is much better than state-of-the-art studies.

Robust QRS detection based on simulated degenerate optical parametric oscillator-assisted neural network

Accurately detecting the depolarization QRS complex in the ventricles is a fundamental requirement for cardiovascular disease detection using electrocardiography (ECG). In contrast to traditional signal enhancement algorithms, emerging neural network approaches have shown promise for QRS detection because of their generalizability on complex data. However, the inevitable noise present during ECG recording leads to a decrease in the performance of neural networks. To enhance the robustness and performance of neural network-based QRS detectors, we propose a simulated degeneration unit (SDU)-assisted convolutional neural network (CNN). An SDU simulates the physical degeneration process of interfering optical pulses, which can effectively suppress in-band noise. Through comprehensive performance evaluations on three open-source databases, the SDU-enhanced CNN-based approach demonstrated better performance in detecting QRS complexes than other recently reported QRS detectors. Furthermore, real-world noise injection tests indicate that the optimal noise robustness boundary for the CNN equipped with SDU is 167–300% higher than that for the CNN without SDU.

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.

NegVSR: Augmenting Negatives for Generalized Noise Modeling in Real-world Video Super-Resolution

The capability of video super-resolution (VSR) to synthesize high-resolution (HR) video from ideal datasets has been demonstrated in many works. However, applying the VSR model to real-world video with unknown and complex degradation remains a challenging task. First, existing degradation metrics in most VSR methods are not able to effectively simulate real-world noise and blur. On the contrary, simple combinations of classical degradation are used for real-world noise modeling, which led to the VSR model often being violated by out-of-distribution noise. Second, many SR models focus on noise simulation and transfer. Nevertheless, the sampled noise is monotonous and limited. To address the aforementioned problems, we propose a Negatives augmentation strategy for generalized noise modeling in Video Super-Resolution (NegVSR) task. Specifically, we first propose sequential noise generation toward real-world data to extract practical noise sequences. Then, the degeneration domain is widely expanded by negative augmentation to build up various yet challenging real-world noise sets. We further propose the augmented negative guidance loss to learn robust features among augmented negatives effectively. Extensive experiments on real-world datasets (e.g., VideoLQ and FLIR) show that our method outperforms state-of-the-art methods with clear margins, especially in visual quality. Project page is available at: https://negvsr.github.io/.

Mitigating Label Noise through Data Ambiguation

Label noise poses an important challenge in machine learning, especially in deep learning, in which large models with high expressive power dominate the field. Models of that kind are prone to memorizing incorrect labels, thereby harming generalization performance. Many methods have been proposed to address this problem, including robust loss functions and more complex label correction approaches. Robust loss functions are appealing due to their simplicity, but typically lack flexibility, while label correction usually adds substantial complexity to the training setup. In this paper, we suggest to address the shortcomings of both methodologies by "ambiguating" the target information, adding additional, complementary candidate labels in case the learner is not sufficiently convinced of the observed training label. More precisely, we leverage the framework of so-called superset learning to construct set-valued targets based on a confidence threshold, which deliver imprecise yet more reliable beliefs about the ground-truth, effectively helping the learner to suppress the memorization effect. In an extensive empirical evaluation, our method demonstrates favorable learning behavior on synthetic and real-world noise, confirming the effectiveness in detecting and correcting erroneous training labels.

Improving the Robustness of Knowledge-Grounded Dialogue via Contrastive Learning

Knowledge-grounded dialogue (KGD) learns to generate an informative response based on a given dialogue context and external knowledge (e.g., knowledge graphs; KGs). Recently, the emergence of large language models (LLMs) and pre-training techniques has brought great success to knowledge-grounded dialogue. However, when building KGD systems in real applications, there are various real-world noises that are inevitable to face. For example, the dialogue context might involve perturbations such as misspellings and abbreviations. In addition, KGs typically suffer from incompletion and also might contain erroneous and outdated facts. Such real-world noises pose a challenge to the robustness of KGD systems and hinder their applications in the real world. In this paper, we propose an entity-based contrastive learning framework for improving the robustness of KGD. Specifically, we make use of the entity information in a KGD sample to create both its positive and negative samples which involve semantic-irrelevant and semantic-relevant perturbations, respectively. The contrastive learning framework ensures the KGD model is aware of these two types of perturbations, thus could generate informative responses with the potentially noisy inputs in real applications. Experimental results on three widely-used benchmark datasets show that our method achieves new state-of-the-art performance in terms of automatic evaluation scores, verifying its effectiveness and potentiality. Furthermore, we show that our method is able to generate better responses than comparison models in both the noisy and the few-shot settings.

Speaker identification under noisy conditions using hybrid convolutional neural network and gated recurrent unit

<p><span>Speaker identification is biometrics that classifies or identifies a person from other speakers based on speech characteristics. Recently, deep learning models outperformed conventional machine learning models in speaker identification. Spectrograms of the speech have been used as input in deep learning-based speaker identification using clean speech. However, the performance of speaker identification systems gets degraded under noisy conditions. Cochleograms have shown better results than spectrograms in deep learning-based speaker recognition under noisy and mismatched conditions. Moreover, hybrid convolutional neural network (CNN) and recurrent neural network (RNN) variants have shown better performance than CNN or RNN variants in recent studies. However, there is no attempt conducted to use a hybrid CNN and enhanced RNN variants in speaker identification using cochleogram input to enhance the performance under noisy and mismatched conditions. In this study, a speaker identification using hybrid CNN and the gated recurrent unit (GRU) is proposed for noisy conditions using cochleogram input. VoxCeleb1 audio dataset with real-world noises, white Gaussian noises (WGN) and without additive noises were employed for experiments. The experiment results and the comparison with existing works show that the proposed model performs better than other models in this study and existing works</span><span>.</span></p>

Quantifying robustness: 3D tree point cloud skeletonization with smart-tree in noisy domains

Extracting tree skeletons from 3D tree point clouds is challenged by noise and incomplete data. While our prior work (Dobbs et al., in: Iberian conference on pattern recognition and image analysis, Springer, Berlin, pp. 351–362, 2023) introduced a deep learning approach for approximating tree branch medial axes, its robustness against various types of noise has not been thoroughly evaluated. This paper addresses this gap. Specifically, we simulate real-world noise challenges by introducing 3D Perlin noise (to represent subtractive noise) and Gaussian noise (to mimic additive noise). To facilitate this evaluation, we introduce a new synthetic tree point cloud dataset, available at https://github.com/uc-vision/synthetic-trees-II. Our results indicate that our deep learning-based skeletonization method is tolerant to both additive and subtractive noise.