Should All Noises Be Treated Equally: Impact of Input Noise Variability on Neural Network Robustness

Salma Alsinan; Maksim Makarenko; Sixiu Liu; Ali Aldawood; Ibrahim Hoteit

cs.LGSep 13, 2026

Should All Noises Be Treated Equally: Impact of Input Noise Variability on Neural Network Robustness

Authors: Salma Alsinan, Maksim Makarenko, Sixiu Liu, Ali Aldawood, Ibrahim Hoteit

Organizations: King Abdullah University of Science and Technology, Saudi Arabia · Aramco Upstream Research Center at KAUST, Saudi Arabia · EXPEC Advanced Research Center, Aramco, Saudi Arabia

Abstract

Geophysical data collected from active field sites are often contaminated by complex and heterogeneous noise, obscuring weak seismic events, and complicating automated interpretation. Although deep learning offers promising solutions for seismic processing, its performance is highly sensitive to the nature of training noise, especially under out-of-distribution (OOD) conditions. This study investigates the influence of noise parameters, such as type, scale, and complexity on the performance, generalization and robustness of neural networks in two geophysical tasks: first break picking and denoising. We simulate seismic-while-drilling data and apply controlled input source noise augmentation using stochastic generators to vary the noise characteristics. Different neural networks are trained on fixed noise types and scales, then evaluated across both seen and unseen noise scenarios. We incrementally increase the complexity of the noise by introducing compound noise mixtures and assess the performance of the model under increasingly challenging OOD conditions. This yields a robustness matrix that captures the generalizability of each model relative to its training configuration. Results indicate that larger noise scales boost generalization, and that effective alignment between noise type, task complexity, and architecture is key for maximizing generalization gains. In addition, training with compound noises mitigate weaknesses associated with single-noise training, acting as an additional implicit regularizer to improving robustness. These findings highlight key factors influencing model resilience in noisy geophysical environments and offer guidance for developing deep learning models that generalize effectively across diverse and unpredictable noise conditions.

Figures & tables

Figure 2: Training and inference pipelines.

Figure 38: Effect of high scale ( $\beta=2.0$ ) single noise training on network predictions for the first break picking task. Compound noise (a) at low scale $\beta_{0.05}$ , (b) at medium scale $\beta_{0.5}$ , (c) at medium scale $\beta_{0.5}$ after pretraining.

Figure 43: Effect of low scale ( $\beta=0.05$ ) single noise training on network predictions for the First break picking task. (a) Single Gaussian noise at a higher scale; (b) compound noise at a higher scale.

Figure 45: Effect of medium scale ( $\beta=0.25$ ) single noise training on network predictions on the first break picking task. (a) Single Gaussian noise at the same training scale, (b) single Gaussian noise at a higher scale, (c) compound noise at a higher scale.

Figure 46: Effect of medium scale ( $\beta=0.25$ ) single noise training on network predictions for the denoising task. Single Gaussian noise at (a) a lower scale, (b) the same training scale, (c) a higher scale.

Figure 47: Effect of high scale ( $\beta=1.0$ ) single noise training on network predictions for the first break picking task. Single Gaussian noise at (a) a lower scale $\beta_{0.25}$ , (b) a lower scale $\beta_{0.5}$ , (c) a higher scale $\beta_{1.5}$ .

Figure 48: Effect of high scale ( $\beta=1.0$ ) single noise training on network predictions for the denoising task. (a) a lower scale $\beta_{0.25}$ , (b) a lower scale $\beta_{0.5}$ , (c) a higher scale $\beta_{1.5}$ . Single (a) colored noise at lower scale $\beta_{0.5}$ , (b) colored noise at a higher scale $\beta_{1.2}$ , (c) Gaussian noise at a higher scale - $\beta_{1.2}$ .

Figure 49: Effect of high-scale ( $\beta=1.0$ ) single-noise training on predictions with a structured noise component present in the compound noise. Compound (Gaussian + linear) (a) at a lower scale $\beta_{0.5}$ on the first break picking task , (b) at a higher scale $\beta_{1.2}$ on the denoising task.

Figure 50: Impact of training scale ( $\beta$ ) on the performance on higher scale linear noises for the denoising task. (a) Low training scale ( $\beta_{0.05}$ ); (b) medium training scale ( $\beta_{0.25}$ ); (c) high training scale $\beta_{1.0}$

Figure 51: Effect of training with compound noise at different scales ( $\beta$ ) on predictions under higher-scale compound noise. Tasks: (a-b) denoising, (c) first break picking. (a) Low training scale ( $\beta_{0.05}$ ) - additive compound noise; (b) medium training scale ( $\beta_{0.25}$ ) additive compound noise; (c) medium training scale ( $\beta_{0.25}$ ) additive compound noise.

Figure 52: Effect of training with compound noise at ( $\beta=0.25$ ) on predictions under higher-scale compound noise (additive + data manipulations). Tasks: (a) first break picking, (b) denoising lowpass (c) denoising missing traces.

Statistic	Raw Data	⋆ Processed Data
Size	$2121\times 241\times 1001$	$2121\times 224\times 224$
Min	$-407.97$	$-404.44$
Max	$1411.20$	$404.44$
Mean	$5.71$	$5.90\times 10^{-5}$
Std. Dev.	$1.70$	$0.920$
5th Percentile	$-1.603$	$-0.836$

Table 1: Statistical summary of the synthetic dataset before and after preprocessing.

Noise Robustness

Noise robustness studies how to preserve reliable predictions when data, labels, signals, or computing hardware are corrupted.

See all 90 papers

Jun 30, 2026

Robustness of neural networks to random noise perturbations of their inputs

We investigate the problem of the robustness of a trained neural network to the perturbation of its input values. More specifically, we examine the interplay between the accuracy of the network, as measured by the mean squared error, and robustness. Accordingly, we present a robustness measure, which, with high probability, suggests an upper bound on the mean squared error of the network, with respect to an input data set, for a given perturbation of the input values of the network. The measure we propose is both simple and efficient to compute, treating the neural network as a black box. We provide experimental results on several real-world data sets showing the efficacy of the proposed method. We also introduce the concept of robustness curves, which allows us to further analyse robustness within and between data sets.

Mark Levene, Martyn Harris

School of Computing and Mathematical Sciences, Birkbeck, University of London, Malet Street, London WC1E 7HX, U.K.

Neural Network Robustness Noise Robustness

Ask PDF

May 20, 2026

GAMR: Geometric-Aware Manifold Regularization with Virtual Outlier Synthesis for Learning with Noisy Labels

Deep neural networks (DNNs) experience significant performance degradation when processing noisy labels, primarily due to overfitting on mislabeled data. Current mainstream approaches attempt to mitigate this issue by passively filtering clean samples during training. However, simple sample filtering within feature spaces degraded by noise struggles to distinguish between challenging samples and noisy samples, creating a bottleneck for model performance. We highlight for the first time the fundamental importance of actively reshaping feature space geometry for learning from noisy data. We propose a novel Geometry-aware Manifold Regularization Paradigm whose core idea is to explicitly construct energy barriers between data manifolds by actively synthesizing virtual outlier samples. By imposing geometric constraints that promote intra-class compactness and inter-class separation, this approach enhances the discriminability between hard and noisy samples, leading to the learning of more robust representations. Our regularization mechanism exhibits high universality, with effectiveness independent of any prior assumptions about noise patterns. It can be integrated as a standalone mechanism into existing sample selection frameworks, providing stronger robustness against diverse noisy environments. Experiments demonstrate that our paradigm achieves performance surpassing current state-of-the-art (SOTA) methods on multiple benchmarks, including CIFAR-10, with particularly pronounced advantages under more challenging asymmetric noise conditions. Furthermore, this paradigm significantly enhances the model's capability in Out-of-Distribution (OOD) detection, ensuring superior reliability and safety for deployment in open-world scenarios.

Ningkang Peng, Jingyang Mao, Xiaoqian Peng +4

Nanjing Normal University Nanjing, China · Nanjing University of Chinese Medicine Nanjing, China

Geometric Representation Learning Manifold Learning Learning with Noisy Labels

Ask PDF

Jun 13, 2026

Learning Earthquake Wave Arrival Time Picking from Labels with Inaccuracies

Inaccurately labeled training data, or "label noise", poses a significant threat to the integrity of supervised machine learning models. This corruption directly degrades performance by teaching the model erroneous mappings between features and labels, which leads to poor generalization and reduced accuracy on properly labeled validation and test data. Current seismological applications mainly rely on large-scale training sets or data augmentation to reduce the label-noise impact, which can be labor-intensive and costly. Here, we introduce a Label Noise-Contrastive Robust Learning (LaNCoR) approach that can effectively handle noisy labels in seismic signal processing tasks, without requiring large-scale training datasets. In this approach, the input waveform feature and label representation distributions are aligned in the feature space to correct mislabeling and reduce its impact on the training process. We present LaNCoR's performance on the task of P-phase arrival-time picking of real microseismic data using two baseline models and training approaches. Our results indicate that LaNCoR can improve performance by up to 28.8% across performance metrics. This approach holds great promise for model training in seismology and geosciences.

Sen Li, Xu Yang, S. Mostafa Mousavi +5

Department of Earth and Planetary Sciences, Harvard University, Cambridge, MA, USA. · School of Computer Science and Technology, China University of Mining and Technology, Xuzhou, Jiangsu, China · School of Mines, China University of Mining and Technology, Xuzhou, Jiangsu, China +1

Contrastive Learning Seismology Learning with Noisy Labels

Ask PDF

Abstract

Noise Robustness

Explore similar work

Robustness of neural networks to random noise perturbations of their inputs

GAMR: Geometric-Aware Manifold Regularization with Virtual Outlier Synthesis for Learning with Noisy Labels

Learning Earthquake Wave Arrival Time Picking from Labels with Inaccuracies