Amount Of Computational Overhead Research Articles

Sequential Feature Filtering Classifier

The Ensemble and mixture of expertise method is the most intuitive and simple way to improve performance in the field of recognition using convolutional neural networks (CNNs). However, Ensemble is difficult to apply in real-time operation applications because the amount of computational overhead and parameters increase in proportion to the number of models. In another, a mixture of expertise that extracts various expertise and combines it is cumbersome to apply because it requires a large change in the network. In this study, we propose Sequential Feature Filtering Classifier (FFC), a simple but effective classifier for CNNs. Using sequential LayerNorm and ReLU, FFC zeroes out low-activation units and preserves high-activation units. The sequential feature filtering process generates multiple features, which are transmitted to a shared classifier, yielding multiple outputs (multiple expertise). FFC can be applied to any CNN with a classifier and it significantly improves the performance with negligible overhead. In this study, the efficacy of FFC is validated extensively on various tasks—ImageNet-1K classification, MS COCO detection, Cityscapes segmentation, and HMDB51 action recognition. Moreover, it is empirically established that FFC can further improve performances using additional techniques, including attention modules. Code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/seominseok0429/Sequential-Feature-Filtering-Classifier</uri> .

A Distributed Access Control with Outsourced Computation in Fog Computing

With the rapid development of information technology and the Internet of Things Technology (IOT), data security and healthy privacy are getting a lot of attention. In order to store, access, and share electronic health records, storage of this data is transferred to a third-party-cloud server. The security and privacy of electronic health records stored at date center or cloud server are not guaranteed. Before being sent to date center or cloud server, this data should be encrypted. Designing an efficient and secure fine-grained access control strategy for personal health records is facing enormous challenges. Security and privacy for electronic health records are very important because the electronic health data which plays an important role in medical server and treatment is directly associated with a particular patient. Attribute-based encryption (ABE) can effectively achieve fine-grained access control. However, the computation of bilinear pairings requires a large amount of computation overhead in ABE scheme. In order to decrease the computational overhead and ensure the confidentiality of electronic health records, a distributed fine-grained access control scheme with outsourced computation for IOT is proposed in this paper. Little calculation is executed by the receiver and sender in our proposed scheme. Outsourcing computing reduces the computing burden. The analyses of safety and performance show that our proposed scheme is safe and effective compared with previous schemes.

The comparison of two reliable methods for the accurate solution of fractional Fisher type equation

PurposeThe purpose of this paper is the comparative analysis of Haar Wavelet Method and Optimal Homotopy Asymptotic Method for fractional Fisher type equation. In this paper, two reliable techniques, Haar wavelet method and optimal homotopy asymptotic method (OHAM), have been presented. The Haar wavelet method is an efficient numerical method for the numerical solution of fractional order partial differential equation like the Fisher type. The approximate solutions of the fractional Fisher-type equation are compared with those of OHAM and with the exact solutions. Comparisons between the obtained solutions with the exact solutions exhibit that both the featured methods are effective and efficient in solving nonlinear problems. However, the results indicate that OHAM provides more accurate value than the Haar wavelet method.Design/methodology/approachComparisons between the solutions obtained by the Haar wavelet method and OHAM with the exact solutions exhibit that both featured methods are effective and efficient in solving nonlinear problems.FindingsThe comparative results indicate that OHAM provides a more accurate value than the Haar wavelet method.Originality/valueIn this paper, two reliable techniques, the Haar wavelet method and OHAM, have been proposed for solving nonlinear fractional partial differential equation, i.e. fractional Fisher-type equation. The proposed novel methods are well suited for only nonlinear fractional partial differential equations. It also exhibits that the proposed method is a very efficient and powerful technique in finding the solutions for the nonlinear time fractional differential equations. The main significance of the proposed method is that it requires less amount of computational overhead in comparison to other numerical and analytical approximate methods. The application of the proposed methods for the solutions of time fractional Fisher-type equations satisfactorily justifies its simplicity and efficiency.

A relationship-based approach for energy aware secure routing in MANETs

Secure routing is one of the major issues which is affecting the widespread use of MANETs. To address this issue several methodologies have been proposed and some of them were able to achieve a considerable amount of success but with a considerable amount of computational overhead. In this paper, we propose a new secure routing mechanism for MANETS. This mechanism uses trust-based method to establish secure routes. It is a family relationship-based approach in which security is established by considering certain parameters which can be used to determine the ingenuity of the nodes. It also considers the parameters like energy consumption and signal strength of the node. The proposed method does not use any protocol specific parameter and thus can be used over any on demand routing protocol like AODV or DSR. As components of MANETs have several resource constraints, this paper aims at providing an improved AODV protocol called secure family-based routing protocol (SFRP) with reduced resource consumption and security enabled.

An Approach Based on Chain Key Predistribution against Sybil Attack in Wireless Sensor Networks

In wireless sensor networks (WSN), Sybil attack can destroy the routing and data distributed storage mechanisms through fabricating identity information of legitimate nodes. This paper presents a chain key predistribution based approach against Sybil attack. To enhance the security of common keys between neighboring nodes, during the chain key predistribution phase, our approach uses a lightweight hash function to generate several chain keys by hashing the unique identity information of every node sequentially in the trusted base station. These chain keys construct a pool of chain keys. During the pairwise key authentication establishment phase, a node-to-node chain key based authentication and exchange (CK-AE) protocol is proposed, by which each node can share the unique pairwise key with its neighboring node. The CK-AE protocol is provably secure in the universally composable security model (UCSM). Finally, we analyze our approach from the resilience of network and the performance overhead, and the results show that our approach can not only enhance the ability of resilience to Sybil attack, but also reduce the communication overhead significantly at the cost of a certain amount of computational overhead.

APP: An Ultralightweight Scheme to Authenticate ONS and Protect EPC Privacy without Cryptography in EPCglobal Networks

EPCglobal network is used to share product data between trading partners, which was proposed by EPCglobal. Object Name Service (ONS) in EPCglobal framework raises two critical security risks: the authenticity of IP addresses for Physical Markup Language (PML) servers and the privacy of Electronic Product Codes (EPCs). Existing work considers either the IP address authentication or the EPC privacy. In addition, that work mainly relies on cryptographic tools, in which key distribution is not a trivial task and also causes a large amount of computation overhead. In this paper, we make the first attempt to solve those two security risks together without relying cryptography. We propose a scheme, namely, APP (authenticate ONS and protect EPC privacy), to guarantee the authenticity of IP addresses for PML servers as well as EPC privacy and to maintain ultralightweight computation cost. Moreover, we give formal definition of the authenticity and the privacy in ONS context. The security achievements are strictly analyzed and proved. The extensive analysis results justify the applicability of the proposed scheme.

Design of trellis coded vector quantizers using Kohonen maps

Although the performance of trellis coded vector quantizers (TCVQs) is much better than that of conventional memoryless vector quantizers, designing the trellis in TCVQs has a certain amount of computational overhead. In this paper, we apply Kohonen maps to simplify the design process of the TCVQs by exploiting the ordering property of Kohonen maps. This new implementation of the TCVQs is named as trellis coded Kohonen maps (TCKMs). Simulation results show that, with such a simple design process, the performance of the TCKMs is still comparable to that of the conventional TCVQs.