A robust interactive estimation of the regularization parameter

We have developed a new and robust method (in the sense of it being applicable to a wide range of situations) to estimate the regularization parameter [Formula: see text] in a regularized inverse problem. For each tentative value of [Formula: see text], we perturb the observations with [Formula: see text] sequences of pseudorandom noise and we track down the instability effect on the solutions. Then, we define a quantitative measure [Formula: see text] of the solution instability consisting of the largest value among the Chebyshev norms of the vectors obtained by the differences between all pairs of the perturbed solutions. Despite being quantitative, [Formula: see text] cannot be used directly to estimate the best value of [Formula: see text] (the smallest value that stabilizes the solution) because, in practice, instability may depend on the particular and specific interests of the interpreter. Then, we determine that the interpreter, at each iteration of a bisection method, visually compares, in the ([Formula: see text], [Formula: see text], [Formula: see text]) space, the pair [Formula: see text] and [Formula: see text] of the solutions most distant from each other and associated with the current [Formula: see text]. From this comparison, the interpreter decides if the current [Formula: see text] produces stable solutions. Because the bisection method can be applied only to monotonic functions (or segments of monotonic functions) and because [Formula: see text] has a theoretical monotonic behavior that can be corrupted, in practice by a poor experiment design, the set of values of [Formula: see text] can be used as a quality control of the experiments in the proposed bisection method to estimate the best value of [Formula: see text]. Because the premises necessary to apply the proposed method are very weak, the method is robust in the sense of having broad applicability. We have determined part of this potential by applying the proposed method to gravity, seismic, and magnetotelluric synthetic data, using two different interpretation models and different types of pseudorandom noise.

Download Full-text

Combined assimilation of streamflow and snow water equivalent for mid-term ensemble streamflow forecasts in snow-dominated regions

10.5194/hess-2016-166 ◽

2016 ◽

Author(s):

Jean M. Bergeron ◽

Mélanie Trudel ◽

Robert Leconte

Keyword(s):

Data Assimilation ◽

State Vector ◽

Snow Water Equivalent ◽

Synthetic Data ◽

Hydrologic Model ◽

Open Loop ◽

The State ◽

Wide Range ◽

Snow Water ◽

The Impact

Abstract. The potential of data assimilation for hydrologic predictions has been demonstrated in many research studies. Watersheds over which multiple observation types are available can potentially further benefit from data assimilation by having multiple updated states from which hydrologic predictions can be generated. However, the magnitude and time span of the impact of the assimilation of an observation varies according not only to its type, but also to the variables included in the state vector. This study examines the impact of multivariate synthetic data assimilation using the Ensemble Kalman Filter (EnKF) into the spatially distributed hydrologic model CEQUEAU for the mountainous Nechako River located in British-Columbia, Canada. Synthetic data includes daily snow cover area (SCA), daily measurements of snow water equivalent (SWE) at three different locations and daily streamflow data at the watershed outlet. Results show a large variability of the continuous rank probability skill score over a wide range of prediction horizons (days to weeks) depending on the state vector configuration and the type of observations assimilated. Overall, the variables most closely linearly linked to the observations are the ones worth considering adding to the state vector. The performance of the assimilation of basin-wide SCA, which does not have a decent proxy among potential state variables, does not surpass the open loop for any of the simulated variables. However, the assimilation of streamflow offers major improvements steadily throughout the year, but mainly over the short-term (up to 5 days) forecast horizons, while the impact of the assimilation of SWE gains more importance during the snowmelt period over the mid-term (up to 50 days) forecast horizon compared with open loop. The combined assimilation of streamflow and SWE performs better than its individual counterparts, offering improvements over all forecast horizons considered and throughout the whole year, including the critical period of snowmelt. This highlights the potential benefit of using multivariate data assimilation for streamflow predictions in snow-dominated regions.

Download Full-text

Denoising large-scale biological data using network filters

10.21203/rs.3.rs-66071/v2 ◽

2021 ◽

Author(s):

Andrew J Kavran ◽

Aaron Clauset

Keyword(s):

Large Scale ◽

Synthetic Data ◽

Interaction Network ◽

Learning Task ◽

Biological Data ◽

Data Sets ◽

Proteomics Data ◽

Life History Variation ◽

Wide Range ◽

Underlying Processes

Abstract Background: Large-scale biological data sets are often contaminated by noise, which can impede accurate inferences about underlying processes. Such measurement noise can arise from endogenous biological factors like cell cycle and life history variation, and from exogenous technical factors like sample preparation and instrument variation.Results: We describe a general method for automatically reducing noise in large-scale biological data sets. This method uses an interaction network to identify groups of correlated or anti-correlated measurements that can be combined or “ﬁltered” to better recover an underlying biological signal. Similar to the process of denoising an image, a single network ﬁlter may be applied to an entire system, or the system may be ﬁrst decomposed into distinct modules and a diﬀerent ﬁlter applied to each. Applied to synthetic data with known network structure and signal, network ﬁlters accurately reduce noise across a wide range of noise levels and structures. Applied to a machine learning task of predicting changes in human protein expression in healthy and cancerous tissues, network ﬁltering prior to training increases accuracy up to 43% compared to using unﬁltered data.Conclusions: Network ﬁlters are a general way to denoise biological data and can account for both correlation and anti-correlation between diﬀerent measurements. Furthermore, we ﬁnd that partitioning a network prior to ﬁltering can signiﬁcantly reduce errors in networks with heterogenous data and correlation patterns, and this approach outperforms existing diﬀusion based methods. Our results on proteomics data indicate the broad potential utility of network ﬁlters to applications in systems biology.

Download Full-text

AI-driven deep CNN approach for multi-label pathology classification using chest X-Rays

PeerJ Computer Science ◽

10.7717/peerj-cs.495 ◽

2021 ◽

Vol 7 ◽

pp. e495

Author(s):

Saleh Albahli ◽

Hafiz Tayyab Rauf ◽

Abdulelah Algosaibi ◽

Valentina Emilia Balas

Keyword(s):

Neural Networks ◽

Data Augmentation ◽

State Of The Art ◽

Synthetic Data ◽

X Rays ◽

Deep Convolutional Neural Networks ◽

Current State ◽

Pathology Classification ◽

Wide Range ◽

Multi Class Classification

Artificial intelligence (AI) has played a significant role in image analysis and feature extraction, applied to detect and diagnose a wide range of chest-related diseases. Although several researchers have used current state-of-the-art approaches and have produced impressive chest-related clinical outcomes, specific techniques may not contribute many advantages if one type of disease is detected without the rest being identified. Those who tried to identify multiple chest-related diseases were ineffective due to insufficient data and the available data not being balanced. This research provides a significant contribution to the healthcare industry and the research community by proposing a synthetic data augmentation in three deep Convolutional Neural Networks (CNNs) architectures for the detection of 14 chest-related diseases. The employed models are DenseNet121, InceptionResNetV2, and ResNet152V2; after training and validation, an average ROC-AUC score of 0.80 was obtained competitive as compared to the previous models that were trained for multi-class classification to detect anomalies in x-ray images. This research illustrates how the proposed model practices state-of-the-art deep neural networks to classify 14 chest-related diseases with better accuracy.

Download Full-text

A Kriging Based Fast and Efficient Method for Defect Detection in Massive Pipelines Using Magnetic Flux Leakages

Volume 14: Safety Engineering, Risk, and Reliability Analysis ◽

10.1115/imece2020-24421 ◽

2020 ◽

Author(s):

Subrata Mukherjee ◽

Xuhui Huang ◽

Lalita Udpa ◽

Yiming Deng

Keyword(s):

Magnetic Flux ◽

Defect Detection ◽

Synthetic Data ◽

Cost Effective ◽

Error Rates ◽

Lattice Points ◽

Operating Characteristics ◽

Wide Range ◽

Similar Accuracy ◽

Fast Procedure

Abstract Systems in service continue to degrade with passage of time. Pipelines are among the most common systems that wear away with usage. For public safety it is of utmost importance to monitor pipelines and detect new defects within the pipelines. Magnetic flux leakage (MFL) testing is a widely used nondestructive evaluation (NDE) technique for defect detections within the pipelines, particularly those composed of ferromagnetic materials. Pipeline inspection gauge (PIG) procedure based on line-scans or 2D-scans can collect accurate MFL readings for defect detection. However, in real world applications involving large pipe-sectors such extensive scanning techniques are extremely time consuming and costly. In this paper, we develop a fast and cheap methodology that does not need MFL readings at all the points used in traditional PIG procedures but conducts defect detection with similar accuracy. We consider an under-sampling based scheme that collects MFL at uniformly chosen random scan-points over large lattices instead of extensive PIG scans over all lattice points. Based on readings for the chosen random scan points, we use Kriging to reconstruct MFL readings over the entire pipe-sectors. Thereafter, we use thresholding-based segmentation on the reconstructed data for detecting defective areas. We demonstrate the applicability of our methodology on synthetic data generated using popular finite element models as well as on MFL data collected via laboratory experiments. In these experiments spanning a wide range of defect types, our proposed novel MFL based NDE methodology is witnessed to have operating characteristics within the acceptable threshold of PIG based traditional methods and thus provide an extremely cost-effective, fast procedure with competing error rates that can be successfully used for scanning massive pipeline sectors.

Download Full-text

Multiple Kinetic Parameterization in a Reactive Transport Model Using the Exchange Monte Carlo Method

Minerals ◽

10.3390/min8120579 ◽

2018 ◽

Vol 8 (12) ◽

pp. 579 ◽

Cited By ~ 2

Author(s):

Ryosuke Oyanagi ◽

Atsushi Okamoto ◽

Noriyoshi Tsuchiya

Keyword(s):

Monte Carlo ◽

Monte Carlo Method ◽

Reactive Transport ◽

Transport Model ◽

Synthetic Data ◽

Solution Chemistry ◽

Rock Interaction ◽

Reactive Transport Model ◽

Reaction Rate Constants ◽

Wide Range

Water–rock interaction in surface and subsurface environments occurs in complex multicomponent systems and involves several reactions, including element transfer. Such kinetic information is obtained by fitting a forward model into the temporal evolution of solution chemistry or the spatial pattern recorded in the rock samples, although geochemical and petrological data are essentially sparse and noisy. Therefore, the optimization of kinetic parameters sometimes fails to converge toward the global minimum due to being trapped in a local minimum. In this study, we simultaneously present a novel framework to estimate multiple reaction-rate constants and the diffusivity of aqueous species from the mineral distribution pattern in a rock by using the reactive transport model coupled with the exchange Monte Carlo method. Our approach can estimate both the maximum likelihood and error of each parameter. We applied the method to the synthetic data, which were produced using a model for silica metasomatism and hydration in the olivine–quartz–H2O system. We tested the robustness and accuracy of our method over a wide range of noise intensities. This methodology can be widely applied to kinetic analyses of various kinds of water–rock interactions.

Download Full-text

Evaluation of methods for gravity wave extraction from middle-atmospheric lidar temperature measurements

Atmospheric Measurement Techniques ◽

10.5194/amt-8-4645-2015 ◽

2015 ◽

Vol 8 (11) ◽

pp. 4645-4655 ◽

Cited By ~ 30

Author(s):

B. Ehard ◽

B. Kaifler ◽

N. Kaifler ◽

M. Rapp

Keyword(s):

Gravity Wave ◽

Gravity Waves ◽

Spectral Response ◽

Synthetic Data ◽

Temperature Measurements ◽

Data Set ◽

Wide Range ◽

Lidar Measurements ◽

Background Temperature ◽

Wave Induced

Abstract. This study evaluates commonly used methods of extracting gravity-wave-induced temperature perturbations from lidar measurements. The spectral response of these methods is characterized with the help of a synthetic data set with known temperature perturbations added to a realistic background temperature profile. The simulations are carried out with the background temperature being either constant or varying in time to evaluate the sensitivity to temperature perturbations not caused by gravity waves. The different methods are applied to lidar measurements over New Zealand, and the performance of the algorithms is evaluated. We find that the Butterworth filter performs best if gravity waves over a wide range of periods are to be extracted from lidar temperature measurements. The running mean method gives good results if only gravity waves with short periods are to be analyzed.

Download Full-text

A New Method of Low Amplitude Signal Detection and Its Application in Acoustic Emission

Applied Sciences ◽

10.3390/app10010073 ◽

2019 ◽

Vol 10 (1) ◽

pp. 73 ◽

Cited By ~ 3

Author(s):

Einar Agletdinov ◽

Dmitry Merson ◽

Alexei Vinogradov

Keyword(s):

Time Series ◽

Acoustic Emission ◽

Logarithmic Derivative ◽

Random Noise ◽

Synthetic Data ◽

Spectral Density Function ◽

Integrity Assessment ◽

Wide Range ◽

Time Average ◽

Low Amplitude

A novel methodology is proposed to enhance the reliability of detection of low amplitude transients in a noisy time series. Such time series often arise in a wide range of practical situations where different sensors are used for condition monitoring of mechanical systems, integrity assessment of industrial facilities and/or microseismicity studies. In all these cases, the early and reliable detection of possible damage is of paramount importance and is practically limited by detectability of transient signals on the background of random noise. The proposed triggering algorithm is based on a logarithmic derivative of the power spectral density function. It was tested on the synthetic data, which mimics the actual ultrasonic acoustic emission signal recorded continuously with different signal-to-noise ratios (SNR). Considerable advantages of the proposed method over established fixed amplitude threshold and STA/LTA (Short Time Average / Long Time Average) techniques are demonstrated in comparative tests.

Download Full-text

Resilience indicator for ecosystems subject to high risk of irreversible degradation: a probabilistic method based on Bayesian inference

10.5194/egusphere-egu2020-6182 ◽

2020 ◽

Author(s):

Yong Liu ◽

Sifeng Wu

Keyword(s):

Bayesian Inference ◽

Spatial Heterogeneity ◽

Lake Level ◽

Probabilistic Method ◽

Quantitative Measure ◽

Lake Eutrophication ◽

Ecosystem Degradation ◽

Parameter Estimations ◽

Wide Range

Ecosystem degradation is usually abrupt and unexpected shifts in ecosystem states that cannot be easily reversed. Some ecosystems might be subject to high risks of irreversible degradation (RID) because of strong undesirable resilience. In this study, we propose a probabilistic method to quantify RID by measuring the probability of the recovering threshold being unattainable under real world scenarios. Bayesian inference was used for parameter estimations and the posteriors were used to calculate the threshold for recovery and thereby the probability of it being unattainable, i.e., RID. We applied this method to lake eutrophication as an example. Our case study supported our hypothesis that ecosystems could be subject to high RID, as shown by the lake having a RID of 72% at the whole lake level. Spatial heterogeneity of RID was significant and certain regions were more susceptible to irreversible degradation, whereas others had higher chances of recovery. This spatial heterogeneity provides opportunities for mitigation because targeting regions with lower RID is more effective. We also found that pulse disturbances and ecosystem-based solutions had positive influences on lowering the RID. Pulse disturbances had the most significant influence on regions with higher RID, while ecosystem-based solutions performed best for regions with moderate RID, reducing RID to almost 0. Our method provides a practical framework to identify sensitive regions for conservation as well as opportunities for mitigation, which is applicable to a wide range of ecosystems. Our findings highlighted the worst scenario of irreversible degradation by providing a quantitative measure of the risk, thus raising further requirements and challenges for sustainability.

Download Full-text

Ambiguities in AVO inversion of reflections from a gas‐sand

Geophysics ◽

10.1190/1.1443740 ◽

1995 ◽

Vol 60 (1) ◽

pp. 134-141 ◽

Cited By ~ 21

Author(s):

Giuseppe Drufuca ◽

Alfredo Mazzotti

Keyword(s):

Unique Solution ◽

Incidence Angle ◽

Synthetic Data ◽

Real Data ◽

Shear Velocity ◽

Sand Layer ◽

Borehole Data ◽

Starting Point ◽

Wide Range ◽

Compressional Velocity

We examine the reflections from a thick sand layer embedded in shales deposited in an open marine environment of Miocene age. Borehole data indicate that the sand bed is gas saturated. Making the assumptions of single interface reflections, plane‐wave propagation in elastic and isotropic media, and correct amplitude recovery of the actual seismic data, we try to invert the amplitude variation with offset (AVO) response for the compressional velocity [Formula: see text], shear velocity [Formula: see text], and density [Formula: see text] of the gas‐sand layer, knowing the parameters of the upper layer and the calibration constant. The actual reflections reach incidence angles up to 54 degrees at the farthest offset. Notwithstanding the large range of incidence angles, the outcomes of the inversion are ambiguous for we find many solutions that fit equally well, in a least‐squares sense, the observed AVO response. We present the locus of the solutions as curves in compressional velocity [Formula: see text], shear velocity [Formula: see text], and density [Formula: see text] space. To gain a better understanding of the results, we also perform the same inversion experiment on synthetic AVO data derived from the borehole information. We find that when inverting the AVO response in the same range of incidence angles as in the real data case, the exact solution is found whichever starting point we choose; that is, we have no ambiguity. However, if we limit the incidence angle range, e.g., to 15 degrees, the invention is no longer able to find a unique solution and the set of admissible solutions defines regular curves in [Formula: see text], [Formula: see text], [Formula: see text] space. We infer that residual noise in the recorded data is responsible for the ambiguities of the solutions, and that because of numerical noise, a wide range of incidence angle is required to obtain a unique solution even in noise‐free synthetic data.

Download Full-text

Simultaneous multifrequency inversion of full-waveform seismic data

Geophysics ◽

10.1190/1.3073002 ◽

2009 ◽

Vol 74 (2) ◽

pp. R1-R14 ◽

Cited By ~ 91

Author(s):

Wenyi Hu ◽

Aria Abubakar ◽

Tarek M. Habashy

Keyword(s):

Seismic Data ◽

Noise Suppression ◽

Regularization Parameter ◽

Synthetic Data ◽

Frequency Data ◽

Perfectly Matched Layers ◽

Inversion Algorithm ◽

Inversion Process ◽

High Convergence Rate ◽

Velocity Image

We present a simultaneous multifrequency inversion approach for seismic data interpretation. This algorithm inverts all frequency data components simultaneously. A data-weighting scheme balances the contributions from different frequency data components so the inversion process does not become dominated by high-frequency data components, which produce a velocity image with many artifacts. A Gauss-Newton minimization approach achieves a high convergence rate and an accurate reconstructed velocity image. By introducing a modified adjoint formulation, we can calculate the Jacobian matrix efficiently, allowing the material properties in the perfectly matched layers (PMLs) to be updated automatically during the inversion process. This feature ensures the correct behavior of the inversion and implies that the algorithm is appropriate for realistic applications where a priori information of the background medium is unavailable. Two different regularization schemes, an [Formula: see text]-norm and a weighted [Formula: see text]-norm function, are used in this algorithm for smooth profiles and profiles with sharp boundaries, respectively. The regularization parameter is determined automatically and adaptively by the so-called multiplicative regularization technique. To test the algorithm, we implement the inversion to reconstruct the Marmousi velocity model using synthetic data generated by the finite-difference time-domain code. These numerical simulation results indicate that this inversion algorithm is robust in terms of starting model and noise suppression. Under some circumstances, it is more robust than a traditional sequential inversion approach.

Download Full-text