Self-relabeling for noise-tolerant retina vessel segmentation through label reliability estimation

Background Retinal vessel segmentation benefits significantly from deep learning. Its performance relies on sufficient training images with accurate ground-truth segmentation, which are usually manually annotated in the form of binary pixel-wise label maps. Manually annotated ground-truth label maps, more or less, contain errors for part of the pixels. Due to the thin structure of retina vessels, such errors are more frequent and serious in manual annotations, which negatively affect deep learning performance. Methods In this paper, we develop a new method to automatically and iteratively identify and correct such noisy segmentation labels in the process of network training. We consider historical predicted label maps of network-in-training from different epochs and jointly use them to self-supervise the predicted labels during training and dynamically correct the supervised labels with noises. Results We conducted experiments on the three datasets of DRIVE, STARE and CHASE-DB1 with synthetic noises, pseudo-labeled noises, and manually labeled noises. For synthetic noise, the proposed method corrects the original noisy label maps to a more accurate label map by 4.0–$$9.8\%$$ 9.8 % on $$F_1$$ F 1 and 10.7–$$16.8\%$$ 16.8 % on PR on three testing datasets. For the other two types of noise, the method could also improve the label map quality. Conclusions Experiment results verified that the proposed method could achieve better retinal image segmentation performance than many existing methods by simultaneously correcting the noise in the initial label map.

Breadth-First Search Multi-Dimensional Binary Search Tree based Algorithms for Structural

This research proposes a set of novel algorithms for structural reliability estimation based on muti-dimensional binary search tree and breadth-first search, namely the reliability accuracy supervised searching algorithm, the limit-state surface resolution supervised searching algorithm and the reliability index precision supervised fast searching algorithm. The proposed algorithms have the following strengths: 1, all the proposed algorithms have satisfactory computational efficiency by reducing redundant samplings; 2, their computational costs are stable and computable; 3, performance functions of high non-linearity can be will handled; 4, the reliability accuracy supervised searching algorithm can adapt its computational cost according to a prescribed accuracy; 5, the limit-state surface resolution supervised searching algorithm is able to probe sharp changes on limit-state surfaces; 6, the reliability index precision supervised fast searching algorithm computes the reliability index with sufficient precision in a fast way.

Locomotive Reliability Assessment Method

The authors describe the analysis of the current state of the problem under consideration. A definition of "averaged failure flow parameter" is given. The periods of traction rolling stock life cycle are considered. The assumption of event distribution laws exponentiality is introduced, which makes it possible to obtain expressions of the main reliability indices in the analytical form. The work of depot service locomotives to ensure the required reliability and readiness of the rolling stock during their normal operation has been assessed. The introduction of the term "readiness" into the modern practice of traction rolling stock reliability estimation is considered. The initial data for calculating the indexes of locomotive uptime and readiness are presented. Calculated values of readiness and no-failure indices of electric locomotives in operation are obtained. The calculated values of internal and technical availability coefficients are compared with similar indicators established by technical specifications. Control procedures were performed to determine the compliance of each set of locomotives (EP1, 2ES4K) with the uptime requirements. As a result of comparing the calculated values of internal and technical availability factors (for electric locomotives EP1 and 2ES4K with analogous values set by specifications (EP1 and 2ES4K) it was determined that the surveyed locomotives comply with the established availability requirements. As a result of control procedures to determine the compliance of each set of EP1 and 2ES4K locomotives with the uptime requirements, it was determined that the set of 2ES4K electric locomotives for the run in question does not fully comply with the uptime requirement. And the set of EP1 electric locomotives meets the reliability requirements, but the error value is higher than 20%. To clarify both events, it is necessary to increase the mileage interval of the locomotives and repeat the procedure for determining compliance with the uptime requirements. The method of assessing the uptime and readiness of locomotives during their normal operation makes it possible to identify existing shortcomings in the operation of rolling stock and to form measures to improve the quality of rolling stock operation.

SRAM-Based PUF Reliability Prediction Using Cell-Imbalance Characterization in the State Space Diagram

This work proposes a methodology to estimate the statistical distribution of the probability that a 6T bit-cell starts up to a given logic value in SRAM memories for PUF applications. First, the distribution is obtained experimentally in a 65-nm CMOS device. As this distribution cannot be reproduced by electrical simulation, we explore the use of an alternative parameter defined as the distance between the origin and the separatrix in the bit-cell state space to quantify the mismatch of the cell. The resulting distribution of this parameter obtained from Monte Carlo simulations is then related to the start-up probability distribution using a two-component logistic function. The reported results show that the proposed imbalance factor is a good predictor for PUF-related reliability estimation with the advantage that can be applied at the early design stages.

Gompertz Fréchet stress-strength Reliability Estimation

In this paper, the reliability of the stress-strength model is derived for probability P(Y<X) of a component having its strength X exposed to one independent stress Y, when X and Y are following Gompertz Fréchet distribution with unknown shape parameters and known parameters . Different methods were used to estimate reliability R and Gompertz Fréchet distribution parameters, which are maximum likelihood, least square, weighted least square, regression, and ranked set sampling. Also, a comparison of these estimators was made by a simulation study based on mean square error (MSE) criteria. The comparison confirms that the performance of the maximum likelihood estimator is better than that of the other estimators.

Reliability Estimation of Inverse Lomax Distribution Using Extreme Ranked Set Sampling

In survival analysis, the two-parameter inverse Lomax distribution is an important lifetime distribution. In this study, the estimation of R = P Y < X is investigated when the stress and strength random variables are independent inverse Lomax distribution. Using the maximum likelihood approach, we obtain the R estimator via simple random sample (SRS), ranked set sampling (RSS), and extreme ranked set sampling (ERSS) methods. Four different estimators are developed under the ERSS framework. Two estimators are obtained when both strength and stress populations have the same set size. The two other estimators are obtained when both strength and stress distributions have dissimilar set sizes. Through a simulation experiment, the suggested estimates are compared to the corresponding under SRS. Also, the reliability estimates via ERSS method are compared to those under RSS scheme. It is found that the reliability estimate based on RSS and ERSS schemes is more efficient than the equivalent using SRS based on the same number of measured units. The reliability estimates based on RSS scheme are more appropriate than the others in most situations. For small even set size, the reliability estimate via ERSS scheme is more efficient than those under RSS and SRS. However, in a few cases, reliability estimates via ERSS method are more accurate than using RSS and SRS schemes.

Reliability Estimation for the Remained Stress-Strength Model under the Generalized Exponential Lifetime Distribution

A stress-strength reliability model compares the strength and stresses on a certain system; it is used not only primarily in reliability engineering and quality control but also in economics, psychology, and medicine. In this paper, a novel extension of stress-strength models is presented. The mew model is applied under the generalized exponential distribution. The maximum likelihood estimator, asymptotic distribution, and Bayesian estimation are obtained. A comprehensive simulation study along with real data analysis is performed for illustrating the importance of the new stress-strength model.

Efficient approximation of reliabilities for single-step genomic BLUP models with the Algorithm for Proven and Young

The objectives of this study were to develop an efficient algorithm for calculating prediction error variances (PEV) for GBLUP models using the Algorithm for Proven and Young (APY), extend it to single-step GBLUP (ssGBLUP), and to apply this algorithm for approximating the theoretical reliabilities for single and multiple trait models in ssGBLUP. The PEV with APY was calculated by block-sparse inversion, efficiently exploiting the sparse structure of the inverse of the genomic relationship matrix with APY. Single-step GBLUP reliabilities were approximated by combining reliabilities with and without genomic information in terms of effective record contributions. Multi-trait reliabilities relied on single-trait results adjusted using the genetic and residual covariance matrices among traits. Tests involved two datasets provided by the American Angus Association. A small dataset (Data1) was used for comparing the approximated reliabilities with the reliabilities obtained by the inversion of the left-hand side of the mixed model equations. The large dataset (Data2) was used for evaluating the computational performance of the algorithm. Analyses with both datasets used single-trait and three-trait models. The number of animals in the pedigree ranged from 167,951 in Data1 to 10,213,401 in Data2, with 50,000 and 20,000 genotyped animals for single-trait and multiple trait-analysis, respectively, in Data1 and 335,325 in Data2. Correlations between estimated and exact reliabilities obtained by inversion ranged from 0.97 to 0.99, whereas the intercept and slope of the regression of the exact on the approximated reliabilities ranged from 0.00 to 0.04 and from 0.93 to 1.05, respectively. For the three-trait model with the largest dataset (Data2), the elapsed time for the reliability estimation was eleven minutes. The computational complexity of the proposed algorithm increased linearly with the number of genotyped animals and with the number of traits in the model. This algorithm can efficiently approximate the theoretical reliability of genomic estimated breeding values in ssGBLUP with APY for large numbers of genotyped animals at a low cost.