Search

The issue of model explainability has gained significant attention in understanding the vulnerabilities and anonymous decision-making processes inherent in deep neural networks (DNN).While there has been considerable research on explainability for traditional DNN, there is a lack of exploration on the operation mechanism and explainability of reversible neural networks (RevNN).Additionally, the existing explanation methods for traditional DNN are not suitable for RevNN and suffer from issues such as excessive noise and gradient saturation.To address these limitations, a visual explanation method called visual explanation method for reversible neural network (VERN) was proposed for RevNN.VERN leverages the reversible property of RevNN and is based on the class-activation mapping mechanism.The correspondence between the feature map and the input image was explored by VERN, allowing for the mapping of classification weights of regional feature maps to the corresponding regions of the input image.The importance of each region for model decision-making was revealed through this process, which generates a basis for model decision-making.Experimental comparisons with other explanation methods on generalized datasets demonstrate that VERN achieves a more focused visual effect, surpassing suboptimal methods with up to 7.80% improvement in average drop (AD) metrics and up to 6.05% improvement in average increase (AI) metrics in recognition tasks.VERN also exhibits an 82.00% level of localization for the maximum point of the heat value.Furthermore, VERN can be applied to explain traditional DNN and exhibits good scalability, improving the performance of other methods in explaining RevNN.Furthermore, through adversarial attack analysis experiments, it is observed that adversarial attacks alter the decision basis of the model.This is reflected in the misalignment of the model’s attention regions, thereby aiding in the exploration of the operation mechanism of adversarial attacks.

To estimate the up-to-date prevalence of chronic kidney disease among the health check-up population in economically developed areas of China using estimated glomerular filtration rate, urinary albumin creatinine ratio, and kidney ultrasound. Healthcare data from 38,093 subjects in 10 megalopolises of China who had an annual health check-up in 2021 were used. The overall and stratified prevalence of chronic kidney disease by sex, age, region and comorbidity group was reported. The association between chronic kidney disease and covariates of demographics, and comorbidities were analyzed in the multivariable-adjusted logistic regression model. A total of 3837 CKD cases were detected meeting any of the three CKD diagnostic criteria, with a crude prevalence of 10.1% in the study population. Using one criterion of decreased glomerular filtration rate, albuminuria and kidney structural abnormalities alone detected 204 (5.3%), 3289 (85.7%) and 563 (14.7%) cases, respectively. The addition of kidney ultrasound detected 427 (11.1%) structural abnormality cases without decreased GFR and albuminuria. The most common abnormalities were renal masses, hydronephrosis due to obstruction and congenital anomalies of kidney and urinary tract. Female, older age, low city-tier, hypertension, diabetes, obesity, hypertriglyceridemia as well as early disease stages such as pre-hypertension, impaired fasting glucose and overweight were significantly associated with chronic kidney disease. Kidney ultrasound helps to amplify the detection of CKD patients, which is a supplement to kidney function and urine protein. [ABSTRACT FROM AUTHOR]