In this paper, we propose a novel implicit semantic data augmentation (ISDA) approach to complement traditional techniques like flipping, translation or rotation. Our work is motivated by the intriguing property that deep networks are surprisingly good at linearizing features, such certain directions in feature space correspond meaningful transformations, e.g., adding sunglasses changing backgrounds. As consequence, translating training samples along many can effectively augment dataset...

Industrial wireless sensor network (IWSN) has changed the information transmission way for existing industrial control system. In mobile sink-based networks, energy consumption optimization data collection always been a hot research issue. To meet delay requirements and minimize consumption, strategy based on ant colony with sink is proposed networks. Firstly, in order to reduce number of nodes directly accessed by shorten traversed path, selection rendezvous entropy weight method introduced...

In order to mitigate the hot spot problem and prolong network lifetime, data gathering with mobile sink is an effective measure enhance system performance. However, movement strategy of node can be regarded as traveling salesman problem, which hardly obtain solution polynomial running time. To address above energy-aware mechanism for in wireless sensor networks using particle swarm optimization introduced. Firstly, mathematical model established according total energy consumption delay...

Using a forward-lockstep build-test process that combines theoretic and pragmatic approaches to ontology building.

Urban sensor networks often consist of a large number low-cost nodes. Due to the constrained resource devices and hazardous deployment, urban sensing is vulnerable interference destruction external factors or impact environmental emergencies. Abnormal data, outliers, anomalies have affected utility in various domains seriously. Timely accurate detection unexpected events, monitoring network performance, anomaly data flow are great significance improve decision-making ability system. In this...

As the scale of data on Internet continues to increase, management and monitoring time series are facing significant challenges. Efficient stable time-series anomaly detection methods necessary for fields such as traffic detection, power grid operation maintenance, financial stock market, industry. However, there fewer abnormal labels in data, labeling cost is high. Traditional expert knowledge-based supervised have been difficult adapt large-scale metric timely alarms. At same time, way...

Aiming at the waste of computing resources resulting from sequential control running mechanism MapReduce model on Hadoop platform,Fork/Join framework has been introduced into this to make full use CPU resource each node. From perspective fine-grained parallel data processing, combined with Fork/Join framework,a and multi-thread model,this paper optimizes puts forward a MapReduce+Fork/Join programming which is distributed architecture coarse-grained platform Support two-tier levels...

This paper describes the realization of a kind wireless data acquisition system based on nRF24E1. In this system, acquired from sensor modules, through transceiver terminal, which adopts mode ShockBurst <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">¿</sup> , transmit to host server and then send computer. The can be analyzed by developing computer software. Through establishing communication protocols star-network, conflict avoided when no...

Existing trust evaluation models for wireless sensor networks can accurately and objectively evaluate value of nodes, but the nodes' energy saving problem was ignored.Especially when there are a few malicious nodes in network, overall calculation all would waste lots energy.Beside that, network failure caused by death also not considered.In this paper, we proposed method avoiding hole which applied based on information entropy, so as to achieve purpose improving utilization.Simulation...

DOI: 10.1049/cp:20061590 ISBN: 0 86341 644 6 Location: Hangzhou, China Conference date: 6-9 Nov. 2006 Format: PDF By deep analysis of J2EE platform and WebGIS systematic structure, this section settles the key technology coordinated operation WebGIS, expatiates cooperation principle, proposes a developing architecture based on platform, then gives working mechanism directed by architecture. The avoids illogicalities problems which lying in nowadays correlative research application about...

In order to make full use of multiple types data and improve the accuracy recommendation algorithm, a scoring prediction model combining multi-source heterogeneous such as product attributes, user reviews friend relationships is proposed for recommendation. this attributes comment information are combined, combination method LDA Word2vec adopted extract user-commodity joint features, community division algorithm used divide users into communities, features ratings integrated establish rating...

Abnormal event detection in crowded scenes is one of the most challenging tasks video surveillance for public security control. Different from previous work based on learning. We proposed an unsupervised Interaction Power model with adaptive threshold strategy to detect abnormal group activity by analyzing steady state individuals' behaviors crowed scene. Firstly, optical flow field potential pedestrians only calculated within extracted foreground reduce computational cost. Secondly, each...

The term software architecture (SA) intuitively denotes the high level structures of a system.It can be defined as set needed to reason about system, which comprise components and connectors, their relations properties.Nowadays SA has become an important factor in process development researches mainly focus on languages, modeling, dynamic evolution process, etc.In this article, complex system brittleness theory are applied into field SA, concept graph collapse path is introduced analytical...

Aiming at the problems of insufficient labeled samples and high missed detection rate in common textured surface anomaly detection, paper designs a self-supervised learning model based on masked Autoencoder, which can realize accurate location anomalies without providing mass samples. Autoencoder is widely used, but it difficult to detect locate by reconstruction error due its strong generalization ability reconstructed with small errors. Then, method proposed reduce performance. First, each...

In this letter, we point out that key substitution attacks should be taken into account for multisignature schemes, which implies the existing security notions schemes are not sufficient. As an example, show scheme proposed by Boldyreva at PKC'03 is susceptible to attacks.