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)...

Recent years, the successful application of blockchain in cryptocurrency has attracted a lot attention, but it also led to rapid growth illegal and criminal activities. Phishing scams have become most serious type crime Ethereum. Some existing methods for phishing detection limitations, such as high complexity, poor scalability, latency. In this article, we propose novel framework named on Ethereum via augmentation ego-graph based graph neural network (PEAE-GNN). First, obtain account labels...

Angiotensin-converting enzyme 2 (ACE2) is a major cell entry receptor for severe acute respiratory syndrome coronavirus (SARS-CoV-2). The induction of ACE2 expression may serve as strategy by SARS-CoV-2 to facilitate its propagation. However, the regulatory mechanisms after viral infection remain largely unknown. Using 45 different luciferase reporters, transcription factors SP1 and HNF4α were found positively negatively regulate expression, respectively, at transcriptional level in human...

In the digital age, it is imperative that Higher Education Institutions (HEIs) should cultivate talent to adapt social development and occupational requirements. To achieve this, HEIs need effectively implement transformation by understanding how leadership teacher competencies contribute technology use in educational settings. This study explored universities’ its relationship with behavior, competency serving as a mediating variable universities from Jilin Province, China. Data were...

An electronic toll collection (ETC) system is one of the most important parts an intelligent transportation (ITS), but existing ETC systems are not efficient and have vehicle fee evasion complications. Owing to traceability tamper resistance a blockchain, combination blockchain feasible way solve above problems. However, traditional (e.g., Bitcoin Ethereum) has high power consumption low efficiency. Hence, it cannot be used in with throughput suitable for power-constrained Internet Things...

The parking lot is one of the important components intelligent transportation system (ITS). current lots mainly use instant parking, which has low efficiency, during peak hours, leads to traffic congestion. To guarantee stable operation lots, we propose a blockchain-enabled reservation framework, called BPR. Traditional systems may exist condition malicious reservations, and resulting in wasted spaces. Therefore, design reputation mechanism manage behavior vehicles reduce number nodes. In...

In order to improve the accuracy of electricity load forecasting in integrated energy systems (IES), an improved method based on NAR dynamic neural network and deep belief (DBN) is proposed. Firstly, calculation net determined for typical topology IES. Based historical data, model established forecast time series components load. Then results DBN pre-training are used initialize (DNN) fine-tune parameters with back propagation (BP) algorithm volatility Finally, predicted value output...

Due to the mobility of intelligent vehicles, limited performance communication devices, and distributed deployment, integrity authenticity accessible information cannot be ensured during traditional digital vehicle forensics in event a collision. For this reason, we propose new attribute-based access control model. Meanwhile, reduce insurance disputes after collision complicate investigation criminal activities by law enforcement, classify different user behaviors according definition...

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...

Most existing methods to detect backdoored machine learning (ML) models take one of the two approaches: trigger inversion (aka.reverse engineer) and weight analysis (aka.model diagnosis).In particular, gradient-based is considered be among most effective backdoor detection techniques, as evidenced by TrojAI competition [14], Trojan Detection Challenge [1] backdoorBench [24].However, little has been done understand why this technique works so well and, more importantly, whether it raises bar...

Electric power transmission lines are integral components of the system, and their digitalization intelligent construction important directions for promoting transformation development energy power. This paper presents a three-dimensional digital management platform tower based on We3D software, which can comprehensively display model surrounding geographical information. By integrating with data interfaces, it enables simple rapid access to information sensor data. facilitates comprehensive...

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....

Unified modeling language (UML) activity diagrams depict the internal behavior of different program operations with help nodes and edges, describing business logic in user requirements. Traditionally, requirements engineers practitioners refer to process documents analyze them build UML manually, which makes labor time consuming. Recently, deep learning technology has been utilized various fields achieved excellent results. We propose a novel pipeline, named TAG, for automatically generating...

The problem of secure distributed storage systems (DSS) with regenerating codes is concerned in this paper. We consider an eavesdropper model where wiretaps a subset nodes, and either their repairing data or stored can be wiretapped. focus on two typical special cases, the Minimum Bandwidth Repair (MBR) Storage (MSR). Our main contribution to draw connection between network coding theory introduced by Cai Yeung, secrecy capacity determined method. prove that for both MBR MSR if maximal...