Recent studies show that adversarial attacks on neural network weights, aka, Bit-Flip Attack (BFA), can degrade Deep Neural Network's (DNN) prediction accuracy severely. In this work, we propose a novel weight reconstruction method as countermeasure to such BFAs. Specifically, during inference, the weights are reconstructed perturbation due BFA is minimized or diffused neighboring weights. We have successfully demonstrated our significantly improve DNN robustness against random and...

The dam displacement can effectively reflect the security status. To improve accuracy of prediction, a combination prediction model is presented based on extreme learning machine (ELM). In this model, predictive values grey GM(1,1) and regression analysis, combined with average results two methods, are used as input vectors ELM, actual selected output vectors, then nonlinear built. simulation show that mean relative error, absolute error 3.04% 4.14% method machine, respectively, which less...

This article presents an improved cuckoo search (ICS) algorithm based on the parametric adaptation mechanism and non-uniform mutation. In ICS algorithm, biased random walk operator is modified to take advantage of neighbourhood information current solution, then control parameters step size, discovery probability scaling factor are directly integrated into optimized problems. Meanwhile, mutation operation used adaptively tune optimal solution. To evaluate feasibility efficiency presented...

The sine cosine algorithm (SCA) is a simple and effective optimization method. However, it may exhibit stagnation behaviour in addressing complex problems. Therefore, novel improved version called the with peer learning (PLSCA) proposed this study. According to roulette wheel selection method, strategy designed achieve information exchange between individuals. Then, control parameter introduced combination of original position updating equations, can be adaptively adjusted by means...

Hemodialysis is an important means to maintain life in patients with end-stage renal disease (ESRD). Approximately 76.8% of who begin hemodialysis do so through catheters, which play vital roles the delivery patients. During past decade, materials, structures, and surface-coating technologies catheters have constantly been evolving ameliorate catheter-related problems, such as recirculation, thrombosis, infections, malfunction. In this study, based on commercial GDK catheter, six catheter...

The Cr, Mn, Cu, Se, Ni, As, Hg, Cd, Pb, Zn, and Fe levels in chicken, duck, quail eggs from Jiangxi Province were determined by inductively coupled plasma mass spectrometry (ICP-MS), the potential risks to human health evaluated. of trace elements not above Chinese national standard. target hazard quotient (THQ) values for total THQ a combination <1, suggesting absence risk elements. carcinogenic (CRs) Pb all egg samples negligible or within acceptable range. These results demonstrate that...

This paper describes how to use XML Web services and public key cryptography on small devices in consumer settings achieve a high level of interoperation security. is done while maintaining the strict performance requirements that are expected from low-cost operating with limited energy other resources.

Data mining is becoming an increasingly important technology for the information society. Due to limited computational resources of data owners and prevalence cloud computing, outsourced more attractive. The privacy security issues are outstanding recently. Though existing model computing consists multiple owners, there little consideration collaboration between them. However, such necessary with trend partition among different entities nowadays. Besides, most work based on semi-honest...

This paper presents a new variant of cuckoo search (CS) algorithm named neighborhood learning-based CS (NLCS) to address global optimization problems. Specifically, in this modified version, each individual learns from the personal best solution rather than found so far entire population discourage premature convergence. To further enhance performance on complex multimode problems, is allowed learn different learning exemplars dimensions. Moreover, exemplar chosen predefined maintain...

An online gradient method is presented and discussed for Pi-Sigma neural networks with stochastic inputs. The error function proved to be monotone in the training process, of tends zero if weights sequence uniformly bounded. Furthermore, after adding a moderate condition, itself also convergent

Basic (t, n)-Secret sharing (SS) schemes share a secret among n shareholders by allocating each share. The can be reconstructed only if at least t shares are available. An adversary without valid may obtain the when more than participate in reconstruction. To address this problem, paper introduces notion and gives formal definition of m, n)-Group oriented (GOSS); proposes n)-GOSS scheme based on Chinese remainder theorem. Without any verification or user authentication, uses Randomized...

As is well known, earthquake often inflicts severe casualties and property losses. The occurrence of earthquakes can not be reliably predicted by current technology, therefore the emergency rescue an important part protecting against mitigating disasters. Earthquake Emergency Command System (EECS) helps relief work carry through well. An EECS prototype based on integration Geographic Information (GIS) Remote Sensing (RS) introduced in this paper. In order to enhance stability, compatibility,...

We investigated the current status and main factors of loneliness among rural left-behind children. conducted a survey 812 children in China using Sociodemographic Questionnaire, Children’s Loneliness Scale, Social Anxiety Scale for Children, Rosenberg’s Self-Esteem Piers-Harris Child Self-Concept Scale. Results showed that (a) 66.26% had sense loneliness; (b) there were correlations between 7 sociodemographic variables, social anxiety, self-esteem, self-concept; (c) grade, age, parental...