The effectiveness of white noise for sleep quality in patients admitted to the intensive care unit: a randomized controlled trial.

Repetitive contrastive learning enhances Mamba's selectivity in time series prediction.

Concurrent Serial Feature Negative and Simple Discrimination Training and Testing in the Mouse.

NLMap-ATVR: A novel combination of nonlinear mapping network and adaptive total variation regularization for MRI denoising.

Well-posedness and regularity analysis for fractional stochastic diffusion-wave equation driven by multiplicative fractional Gaussian noise with Hurst index H ∈ (0, 1)

Pedestrian behavior exhibits inherent multi-modality, necessitating predictions that balance accuracy and diversity to adapt effectively to various complex scenarios. However, conventional noise addition in diffusion models is often aimless and unguided, leading to redundant noise reduction steps and the generation of uncontrollable samples. To address these issues, we propose a Prior Condition-Guided Diffusion Model (CGD-TraP) for multi-modal pedestrian trajectory prediction. Instead of directly adding Gaussian noise to trajectories at each timestep during the forward process, our approach leverages internal intention and external interaction to guide noise estimation. Specifically, we design two specialized modules to extract and aggregate intention and interaction features. These features are then adaptively fused through a spatial-temporal fusion based on selective state space, which estimates a controllable noisy trajectory distribution. By optimizing the noise addition process in a more controlled and efficient manner, our method ensures that the denoising process is effectively guided, resulting in predictions that are both accurate and diverse. Extensive experiments on the ETH-UCY, SDD, and NBA datasets demonstrate that CGD-TraP surpasses state-of-the-art diffusion-based and other generative methods, achieving superior efficiency, accuracy, and diversity.

Multi-scale spatial diffusion under frequency information-guidance For low-light image enhancement.

Empirical evaluation of variability and multi-institutional generalizability of deep learning survival models: application to renal cancer CT scans.

We present a continuous-variable (CV), quantum-key-distribution (QKD)-inspired, keyed physical-layer masking method for classical M-level pulse-amplitude-modulation (M-PAM) links. A transmitter adds a per-symbol Gaussian dither [Formula: see text], generated by a seeded pseudorandom number generator (PRNG), directly to the analog waveform, and an authorized receiver that shares the PRNG, its seed, and [Formula: see text] regenerates and subtracts the same sequence prior to slicing. Because masking and demasking act on the data-carrying optical signal, the scheme operates over conventional, amplifier-compatible fiber without reserving a separate quantum channel. First, we analyze an idealized continuous-valued baseband model in which the PAM symbols pass through an additive-white-Gaussian-noise (AWGN) channel and no extra amplitude quantization or link impairments are present. In this setting we show that, when the seed and [Formula: see text] are matched, subtractive cancellation is essentially ideal and the Gray-coded 4-PAM bit-error-rate (BER) versus signal-to-noise ratio (SNR) curve coincides with the standard AWGN benchmark, whereas seed or parameter mismatches act as additional Gaussian noise and produce several-decibel SNR penalties or high, SNR-independent BER floors. We then implement the same masking mechanism in a system-level OptiSystem model of an intensity-modulation/direct-detection (IM/DD) link and enable a [Formula: see text] double-quantization mapping, in which dithered 4-PAM symbols are passed through a mid-rise 8-level quantizer at the transmitter and a mid-tread 4-level quantizer at the receiver. In this OptiSystem realization the resulting pseudo-constellation exhibits amplitude statistics reminiscent of probabilistic shaping and an intrinsic, SNR-independent algorithmic BER floor, typically [Formula: see text] in back-to-back simulations. By choosing forward-error-correction (FEC) codes whose waterfall threshold lies near this floor and by tuning the dither variance at the transmitter and receiver, the link can be operated in a deliberately fragile edge-of-FEC regime in which small excess disturbances (for example, seed or variance mismatch) push the BER out of the decodable region and strongly obfuscate the payload. The proposed masking is modulation-agnostic and is intended as a classical protection layer rather than a full QKD protocol; keying material can be supplied over an authenticated supervisory channel or by an independent QKD system.

Although MRI reconstruction requires a dealiasing transformation from undersampled to fully-sampled data, task-agnostic diffusion priors sample images via a denoising-based generative trajectory from an asymptotic start-point of Gaussian noise onto fully-sampled data. Since aliasing artifacts in MR images carry spatial structure deviating from Gaussian noise, this noise-governed trajectory can cause suboptimal artifact suppression. To address this limitation, we introduce the first Fourier-constrained diffusion bridge (FDB) for MRI reconstruction in the literature. Unlike task-agnostic diffusion priors, FDB does not rely on noise in its forward process and instead learns a dealiasing transformation between a start-point of undersampled data and the end-point of fully-sampled data. The start-point is derived via a stochastic Fourier-constrained degradation operator that removes a progressively growing set of spatial frequencies. Unlike cold/soft diffusion priors that use an asymptotic start-point of severely degraded measurements, FDB uses a realistically undersampled start-point to ensure closer alignment of model input between training and test distributions. Unlike existing diffusion bridges that use degradations based on weighted linear averages and noise addition, FDB implements degradations based on binary removal of compact k-space sets to conform to the physics of accelerated MRI. To further improve image quality, FDB leverages a novel sampling algorithm based on progressive dealiasing by continually correcting recovered k-space data across reverse diffusion steps. Demonstrations on brain MRI show that FDB outperforms competing methods by 4.5dB PSNR and 8.3% SSIM in within-domain and by 4.7dB PSNR and 16.4% SSIM in cross-domain reconstructions.

Stochastic partial differential equations (SPDEs) have become key tools for modeling randomly disturbed systems in a broad spectrum of scientific and engineering disciplines. Among them, the optical soliton propagation in birefringent fibers has been of great interest because of the natural stochastic effects present in real optical communication systems. Although some research has been conducted on soliton dynamics in deterministic and integer-order models, the literature has few studies focusing on fractional stochastic frameworks coupled with M-truncated fractional operators. This paper fills this gap by proposing and studying the fractional stochastic Biswas–Arshed equation (FSBAE) that includes multiplicative white noise to represent random perturbations during optical signal propagation. We employ the [Formula: see text] method with Itô calculus and M-truncated fractional derivatives: in order to arrive at three different classes of solutions: stochastic optical breather solitons, M-shaped solitons, and singular solitons. The first contribution of this work consists in the combination of M-truncated fractional calculus — a not extensively developed operator in stochastic nonlinear optics — within the FSBAE, for the first time applied to this model. Comparative graphical analysis under conditions of different levels of white noise and fractional orders also identifies the robustness and stability of the solutions in close proximity to the zero-noise limit. This research adds to the development of soliton theory in that it has proposed a new mathematical model that can capture both nonlocal memory effect and stochastic variability, which are essential in real optical systems. The efficiency, simplicity, and flexibility of the approach make it an effective tool in solving a wider class of nonlinear stochastic issues in optical engineering and other fields dealing with stochastic partial differential equations.

Action modulates visual detection and metacognitive sensitivity.

Accurate deflection prediction is vital for structural health monitoring of large-span bridges yet remains challenging due to complex nonlinear environmental couplings. This paper proposes a hybrid deep learning framework, BGCO-PIC-DA-LSTM, for precise bridge deflection prediction. First, a Prior-Informed Correlation (PIC) strategy incorporating temperature lag terms is introduced to enhance the statistical consistency of input features. Second, a dual-stream residual Bi-LSTM network integrating adaptive temporal attention is developed to simultaneously capture long-term evolutionary trends and instantaneous dynamic fluctuations. Furthermore, a Bayesian-Gradient Cooperative Optimization (BGCO) strategy is employed to automatically configure optimal hyperparameters. Validation using in situ data from a large-span cable-stayed bridge demonstrates that the proposed method significantly outperforms baseline algorithms in prediction accuracy and robustness. Additionally, the prediction residuals exhibit characteristics approximating zero-mean Gaussian white noise, establishing a reference baseline for structural state evolution and providing a certain basis for identifying potential performance shifts.

. In this article, we will develop linear and nonlinear filtering methods for a large class of nonlinear wave equations that arise in applications such as quantum dynamics and laser generation and propagation in a unified framework. We consider both stochastic calculus and white noise filtering methods and derive measure-valued evolution equations for the nonlinear filter and prove existence and uniqueness theorems for the solutions. We will also study first-order approximations to these measure-valued evolutions by linearizing the wave equations and characterize the filter dynamics in terms of infinite-dimensional operator Riccati equations and establish solvability theorems.

The signature of stochastic resonance is that additional noise surprisingly enhances the signal-to-noise ratio (SNR). A noise-adaptive system that learns to add an optimal amount of noise to trigger stochastic resonance and improve SNR is known as adaptive stochastic resonance. However, the current stochastic resonance mechanism fails when environmental noise exceeds the optimal noise level, as any additional noise merely worsens the SNR. In this case, instead of adding noise, stochastic resonance can be facilitated by adapting the potential energy landscape of the bistable system. Here, we propose a novel approach to enhance SNR in noisy environments, involving a potential adjustable microelectromechanical systems resonator. A periodic signal with an amplitude of 0.28 Vrms is buried in ambient noise, emulated by a white noise signal with amplitude ranging from 0.7 Vrms to 4 Vrms. Experimental results show that when the ambient noise exceeds 1 Vrms, adding additional noise leads to a decline in SNR. However, SNR enhancement induced by stochastic resonance is experimentally demonstrated by tuning the potential well of the resonator. This advancement highlights the feasibility of potential adjustable systems to overcome the limitations of conventional noise adjustable stochastic resonance methods in noisy environments. The proposed mechanism is further applied to detect the frequency of 2.7 nN periodic forces with various waveforms.

Background/Objectives: Brain tumors are among the most severe neurological disorders, and their variability in size, morphology, and anatomical location complicates early and accurate diagnosis. Although magnetic resonance imaging (MRI) is the most reliable non-invasive modality for tumor detection, manual interpretation remains time-consuming, subjective, and susceptible to human error. This study aims to develop an optimization-driven hybrid machine learning framework for accurate and computationally efficient automatic brain tumor classification. Methods: The dataset includes 834 MRI images (583-training, 123-validation, 128-independent test). Because YOLOv11 detects tumor and non-tumor regions separately, the sample size doubled during region-based analysis, and all subsequent stages were conducted at the regions of interest (ROI) level. On the independent test set, YOLOv11 achieved 98.87% mAP@50, 98.54% precision, and 98.21% recall. The proposed framework combines automated tumor localization with image standardization using Gaussian noise reduction and bilinear interpolation. From the processed MR images, 39 entropy-based features were extracted. To enhance diagnostic performance and eliminate redundant information, the superb fairy-wren optimization algorithm (SFOA) was applied for feature selection and compared with particle swarm optimization (PSO), Harris hawk optimization (HHO), and puma optimization (PO). Final classification was primarily performed using k-nearest neighbors (kNN), while support vector machines (SVM) were used for comparative evaluation. Results: SFOA reduced the feature dimensionality from 39 to 5 features while achieving 99.20% classification accuracy on the independent test set. In comparison, PSO selected 10 features, HHO selected 6 features and PO selected 10 features, all achieving 98.45% accuracy. The best performance obtained with SVM was 98.45% accuracy (HHO-SVM), which remained lower than the 99.20% achieved by the proposed SFOA-kNN model. Conclusions: The results indicate that combining entropy-based feature extraction with SFOA-driven feature selection and kNN classification significantly enhances diagnostic accuracy while reducing computational complexity, highlighting the strong potential of the proposed framework for integration into computer-aided diagnosis systems to support clinical decision-making.

RIFoL: A robust image forgery localization network for noisy images.

As critical damping components of vehicles, suspension systems play an essential role in maintaining vehicle stability and enhancing ride comfort. This paper studies the dynamic behaviors and reliability of the suspension system. First, based on Newton’s second law, a single-degree-of-freedom suspension system model is established through simulating the rough road fluctuations as a combination of typical cosinusoidal road excitation and Gaussian white noise. Then, considering the linear damping and nonlinear damping, respectively, the dynamic evolution and first-passage failure of the system under primary resonance and 1/3 subharmonic resonance conditions are examined by the path integral method. The influence mechanisms of dampings, road surface excitation amplitude and noise intensity on the dynamics of suspension systems are explored. The results demonstrate that reduced damping, increased road excitation amplitude and higher noise intensity collectively impair system stability. Crucially, the system’s response to these parameters is governed by the resonance type. Within a certain range, under primary resonance, road amplitude predominantly affects displacement, whereas under 1/3 subharmonic resonance, it significantly alters both displacement and velocity distributions, even inducing stochastic P-bifurcation. These findings provide valuable insights into the design and optimization of vehicle suspension systems for improving performance and reliability.

Abstract. Chaotic modulation is an effective communication technique that exploits deterministic chaos to produce pseudo-random signals. A widely adopted approach involves modulation of the chaotic bifurcation parameter. This paper introduces a deep learning–based demodulation method for keying of the bifurcation parameter. It describes the architecture of the convolutional neural network and evaluates performance metrics for signals generated using the chaotic logistic map. The study assesses the bit error rate for binary signals and reports a bit error rate of 0.0819 for a bifurcation parameter deviation of 1.34% under additive white Gaussian noise at a signal-to-noise ratio of -13 dB (corresponding to a normalized signal-to-noise ratio of +20 dB). The results demonstrate the capability to detect chaotic patterns even when the specific patterns were not included in the training dataset.

Deep neural networks perform well on clean image classification tasks but often fail under common corruptions and distribution shifts. This paper introduces DistortionMix, a lightweight hybrid distortion-based augmentation technique designed to improve model robustness. It randomly applies contrast variation, Gaussian noise, or impulse noise to training images, enhancing data diversity and encouraging resilient feature learning. We evaluate DistortionMix on CIFAR-10 (clean) and CIFAR-10-C (corrupted), which includes 19 corruption types at five severity levels. A variety of architectures e.g ResNet, DenseNet, EfficientNet, MobileNet, VGG, AlexNet, GoogleNet, and ViT are fine-tuned with and without DistortionMix. Experimental results show that DistortionMix improves corrupted accuracy by up to 13.8%, while maintaining or slightly improving clean accuracy. Among all models, ViT-Base (timm) achieves the highest robustness, reaching 89.4% on severe corruptions and 97.43% on clean data. These findings highlight DistortionMix as a simple yet effective strategy for enhancing out-of-distribution generalization. Future work includes extending distortion types, developing adaptive augmentation policies, and evaluating performance on real-world corrupted datasets. Source code: github.com/HusniFadhilah/DistortionMix.