In traffic accident, an accurate and timely severity prediction method is necessary for the successful deployment of intelligent transportation system to provide corresponding levels medical aid in a manner. The existing accident's methods mainly use shallow models statistical models. To promote accuracy, novel prediction-convolutional neural network (TASP-CNN) model proposed that considers combination relationships among features. Based on weights features, feature matrix gray image (FM2GI)...

Machine learning models may not be able to effectively learn and predict from imbalanced data in the fields of machine mining. This study proposed a method for analyzing performance impact binary on models. It systematically analyzes 1. relationship between varying imbalance rate (IR); 2. stability data. In method, augmentation algorithms are first designed obtain dataset with gradually IR. Then, order more objective classification results, evaluation metric AFG, arithmetic mean area under...

Being the necessary data of city-scale seismic damage simulations, structural types buildings a city need to be collected. To this end, prediction method based on machine learning (ML) is proposed herein. Specifically, using training 230,683 in Tangshan city, China, supervised ML solution decision forest model was designed for prediction. The scale sensitivity and regional applicability are discussed, respectively, results show that can maintain high accuracy different scales; however, it...

Short-term traffic flow forecast plays an important role in intelligent transportation systems. Existing prediction model used deep layer neural network, which can't prevent over fitting, resulting performance loss and lack of generalization ability. We propose a regularized LSTM that fused recurrent dropout max-norm weight constraint. apply to the connections use constraint arrest input weights not grow very large. Simultaneously, we merge ADAM optimizer into our model. three datasets from...

Image acquisition and transmission in wireless sensor networks (WSN) are core issues for some resource-deficient multimedia sensing applications. Reducing sampling rates data lowers node costs energy, addressing communication bottlenecks. Block compressed (BCS) can meet the above requirements. For BCS, sparsity or smoothness of block signal is a crucial parameter, which determines setting range rate. side node, we cannot directly obtain complete digital signal. Therefore, it becomes...

This paper presents a non-linear inertia weight adjustment strategy of PSO algorithm (ULWPSO), and injects the SA algorithmpsilas thinking accepting worse by certain probability to ULWPSO, thatpsilas ULWPSO-SA fusion which can solve global optimal solution efficiently. Through contrasted experiment, results show increase diversity PSO, enhance ability solving overall solution, own higher convergence speed more precise accuracy.

The abundance of event data in current software configuration management systems makes it possible to discover process models automatically by using actual observed behavior. However, traditional mining algorithms cannot be applied logs recorded (SCM) systems, such as SVN, because missing activity attributes. To address this problem, a classifier is proposed build event-activity mapping relationships from development streams, revealing attributes and associating the original SVN log....

Habitat directly affects the population size and geographical distribution of wildlife species, including Mangshan pit viper (Protobothrops mangshanensis), a critically endangered snake species endemic to China. We searched for using randomly arranged transects in their area assessed habitat association plots, with goals gaining better understanding features associated P. mangshanensis detection determining if these varies across season. conducted transect surveys, found 48 individual...

This study aims at quantifying the influence on traversability of road network caused by building collapse in earthquake. To this end, an analysis method post-earthquake considering is proposed. First, time-history seismic response based multi-degree freedom (MDOF) model performed for regional groups, so that collapsed buildings could be determined. Subsequently, impact ranges are calculated a probabilistic debris distributions. Finally, algorithm designed ranges, and therefore solution to...

This paper proposed a multiobjective artificial bee colony algorithm (MOABC) to solve the reliability redundancy allocation problem (RAP) with multiple objectives, which introduces rank value and crowding distance in greedy selection strategy, applies fast non-dominated sort procedure exploitation search inserts tournament onlooker phase. The number of components is be decided so as maximize availability minimize designing cost system simultaneously. It shows that can efficiently.

In this paper, we are devoted to the construction and analysis of research network. Firstly, on basis data Erdos1, a co-author network is built two novel measures proposed analyze properties A extraction method using string matching technique developed visualized UCINET. Then, first-order second-order entropy defined depict complexity network, node’s invalidity probability load-bearing capacity robustness

This paper proposed a multiobjective artificial bee colony algorithm (MOABC) to solve the reliability redundancy allocation problem (RAP) with multiple objectives, which introduces rank value and crowding distance in greedy selection strategy, applies fast non-dominated sort procedure exploitation search inserts tournament onlooker phase. The number of components is be decided so as maximize availability minimize designing cost system simultaneously. It shows that can efficiently.