SKEW DETECTION OF PRINTED DOCUMENT IMAGES USING CHARACTER ORIENTATION ESTIMATION
Yangli Wang and Rui Song, School of Telecommunications Engineering, Xidian University, Xi’an China
In the capturing process of document images, skew is usually unavoidable. This skew may seriously degrade the performance of subsequent processing steps such as layout analysis and segmentation. As such, skew detection and correction is a necessary step in the document image processing pipeline. In this paper we propose a new skew detection method using character orientation estimation. The first step of the approach is classifying the connected components in a binarized document image into text/non-text. This is accomplished by a convolutional neural network (CNN). The next step of the proposed approach
is estimating character orientation using the minimum area rect surrounding a character. Initial document skew is then estimated with the first ?? largest characters and their orientation. This initial skew is then refined by clustering deskewed nearest characters and minimizing sum of the area of their surrounding rects in a small search range. As text/non-text are separated first, and only text is involved in document skew estimation, the proposed method overcomes the drawbacks of projection profile based methods, which can not deal well with documents containing other elements, and are weak in handling multicolumn documents when the lines of different columns are not well aligned. The proposed approach also has the advantage over those methods estimating text line orientation, such as the Hough transform
based ones, they have difficulties in choosing representing points for characters as characters are of different height. Simulation results on test document images show that, the proposed approach has achieved a mean estimation error of as low as 0.1169°. It also has the advantage of potentially saving the geometric layout analysis step in the conventional document image processing pipeline.
Character Orientation, Convolutional Neural Network, Minimum Area Rect, Character
Cluster, Document Skew Detection
PERFORMANCE EVALUATION OF FEATURE EXTRACTION FROM BIOELECTRICAL SIGNAL USING SMARTWATCH
Timibloudi S Enamamu1,2, Abayomi M Otebolaku1 and Joy Dany2, 1Sheffield Hallam University, Sheffield, UK and 2Communications and Network Research (CSCAN) Plymouth University, United Kingdom
This work provides a process for selecting suitable features using biorthogonal wavelet decomposition of signal from a non-intrusive extraction. A smartwatch is used for extracting bioelectrical signal before decomposing the signal into sub-bands of Detail and Approximation Coefficient. A detail experiment is conducted extracting suitable statistical features from the bioelectrical signal from 30 subjects using different biorthogonal wavelet family. Ten features are extracted using Biorthogonal wavelet to decompose the signal into three levels of sub-band Detail and Approximation Coefficient and features extracted from each levels the decomposed Detail and Approximation Coefficients. Comparison analysis is done after the classification of the extracted features based on the Equal Error Rate (EER). Using Natural Network (NN) classifier, Biorthogonal Wavelet Detail Coefficient Sub-band level 3 of bior1.1 achieved the best result of EER 13.80% with the fusion of the best sub-band three levels of bior1.1 achieving a better result of 12.42% EER.
Bioelectrical signals, biorthogonal wavelet, approximation coefficients, detail coefficient, wavelet transform.
CONSTRUCTION OF SEIZURE DETECTION USING ENTROPY ANALYSIS IN MULTI-SCALE
Xiaoyan Wei1, Ziyi Chen2, and Yi Zhou1, 1Department of Biomedical Engineering, Zhongshan School of Medicine, Sun Yat-sen University, Guangzhou, China and 2Department of Neurology, First Affiliated Hospital, Sun Yat-sen University, Guangzhou, China
The diagnosis of epilepsy mainly relies on the experts’ visual inspection for electroencephalogram (EEG), which is the gold standard of epilepsy diagnosis. In a wearable clinical application, the automatic classifier of epileptic EEG signal requires to be fast and simple. Although the nonlinear dynamics index permutation entropy (PE) could identify the differences among normal, inter-ictal and ictal states, multi-scale entropy analysis on how to choose suitable scaling factors to calculate regarding patients with specific EEG data patterns is lacking. In this paper, seizure detection was constructed using entropy analysis in multi-scale. Multi-scale permutation entropy (MPE) was calculated considering the differences in the patients’ specific characteristics. kernel extreme learning machine (K-ELM) was constructed to classify the different epileptic stages from EEG data, namely normal, inter-ictal and ictal stages. The system performance was evaluated in the multi-center dataset, which was compared with classical predictors such as support vector machine (SVM) and extreme learning machine (ELM). It showed that the value of MPE changes in normal and the epileptic subjects with the change in scale factors, multi-scale entropy analysis could increase the performance of seizure detection. The overall algorithm not only has a high classification accuracy but also has a fast calculation speed that gives a great significance for real-time detection of epileptic seizures. The results addressed the multi-scale nonlinear dynamics changes in seizure procedure and provided a fast and exquisite classifier for clinical transformation.
epilepsy electroencephalogram multi-scale permutation entropy Kernel extreme learning machine.
SATELLITE-BASED DATA COLLECTION ARCHITECTURE FOR VIRTUAL POWER PLANT MANAGEMENT IN RURAL AREAS
Atm S. Alam1, Prashant Pillai2, Y.Fun Hu3 and Haile-Selassie Rajamani4, 1Queen Mary University of London, UK, 2University of Wolverhampton, UK, 3University of Bradford, UK and 4University of Wollongong Dubai,UAE
Smart grid is envisaged to be the next-generation electrical power grids, and this is founded based on successfully building up smart grid communication networks that can support all identified smart grid functionalities. Despite a range of communication choices available, utilities still struggle with how to affordably and reliably extend their networks to 100% of their service territories, especially to remote locations that are beyond the reach of primary networks. In all smart grid models, it is often emphasized that consumers play a vital role in electricity supply and demand management and they are expected to be co-producers of electricity, so-called prosumers. As such, virtual power plants (VPPs) by interconnecting hundreds of prosumers are expected to be a new paradigm shift in smart grid systems to better utilize the distributed energy sources. However, an efficient VPP management is of great challenge in rural areas that are beyond the reach of primary networks and they require enormous data exchange. To provide a connectivity solution in rural areas, this paper proposes a satellite-based smart grid communication architecture for the VPP management that requires collecting data from prosumers forming the VPP. In addition, a priority-based scheduling algorithm for different smart grid data types is proposed to improve the performance of delay-sensitive applications. Simulation results demonstrate that the satellite-based communications can be a viable solution as a mean of smart grid communications for VPPs.
Smart Grid Communications, Virtual Power Plant (VPP), Smart Meter, Satellite Architecture, Scheduling
INTRANET SECURITY USING A LAN PACKET SNIFFER TO MONITOR TRAFFIC
Ogbu N. Henry1 and Moses Adah Agana2, 1Department of Computer Science, Ebonyi State University, Abakaliki, Nigeria and 2Department of Computer Science, University of Calabar, Nigeria
This paper was designed to provide Intranet traffic monitoring by sniffing the packets at the local Area Network (LAN) server end to provide security and control. It was implemented using five computer systems configured with static Internet Protocol (IP) addresses used in monitoring the IP traffic on the network by capturing and analyzing live packets from various sources and destinations in the network. The LAN was deployed on windows 8 with a D-link 16-port switch, category 6 Ethernet cable and other LAN devices. The IP traffics were captured and analyzed using Wireshark Version 2.0.3. Four network instructions were used in the analysis of the IP traffic and the results displayed the IP and Media Access Control (MAC) address sources and destinations of the frames, Ethernet, IP addresses, User Datagram Protocol (UDP) and Hypertext Transfer Protocol (HTTP). The outcome can aid network administrators to control Intranet access and provide security.
Packet, Sniffer, Protocol, Address, Network, Frame
ANALYTICAL STUDY AND SIMULATION FOR PROPAGATION OF ALERTS, CASE OF EMERGENCY VEHICLES IN SMART CITIES
Éloi B. KEITA1, Pierre-Yves LUCAS1, Babacar DIOP2 and Bernard POTTIER1, 1Université de Bretagne Occidentale, Lab-STICC, UMR CNRS 6285 – Brest, France and 2Université Gaston-Berger – Saint Louis, Sénégal
This paper proposes methods and tools to support analysis and simulation of propagation of alerts in a city. Environmental modeling and monitoring is now a major framework for application of wireless sensor networks. Sound and visual alerts remain a major way to warn and prevent accidents in social life. This work combines a cellular segmentation of the city, representation of buildings and roads, representation of vehicle paths, and cell behaviour that compute sound wave propagation with respect to space and time. As a first result, there is the possibility to evaluate nuisance from vehicles repetitively travelling their sirens along avenues. Coupling to smart cities sensing systems will allow a better control on traffic lights and management of autonomous intersections. This can help to prevent and reduce noise and accidents. In the context of smarts cities, cooperative sound detection can be associated to the development of new smart cars, and better rescue or police vehicles. High speed simulations with real time opportunities are obtained thanks to code generation for graphic accelerators.
Wireless sensor networks, cellular automata, siren alert, sound propagation simulation, parallel computing
A Parallel Bit-Map Based Framework for Classification Algorithms
Amila De Silva, Shehan Perera, Department of Computer Science & Engineering, University of Moratuwa, Katubedda, Sri Lanka
Bitmap representations have been abundantly used in data analytic queries for their ability to represent data concisely and for being able to simplify processing. For the same reasons, bitmaps are gaining popularity in Data Mining domain, with the arrival of GPUs, since Memory organisation and the design of a GPU demands for regular & simple structures. However, due to the nature of processing, use of bitmaps have largely been restricted to FIM based algorithms. We in this paper, present a framework based on bitmap techniques, which speeds up classification algorithms on GPUs. The proposed framework uses both CPU and GPU for the algorithm execution, where the core computing is delegated to GPU. We implement two classification algorithms Naive Bayes and Decision Trees, using the framework, both of which outperform CPU counterparts by several orders of magnitude
Data Mining, Classification, Bitmaps, Bit-Slices, GPU.
The Role of Opinions and Ideas in TACIT Knowledge Externalization: Tacit Knowledge Categorization
Jamal El-Den and Narumon Sriratanaviriyakul, Charles Darwin University, Australia
Purpose:The paper identifies the difficulties associated with managing tacit knowledge in its entirety among distributed individuals and proposes its categorization into types/kinds as a solution for its effective externalization and measurement. The categorization process implies the identification of those types or kinds of tacit knowledge which could be externalized and measured easier than others. The paper posits that such categorization is a step in the right direction forbetter tacit-to-explicit transformation.
OPTIONS TRADING AND HEDGING STRATEGIES BASED ON MARKET DATA ANALYTICS
Huang-Ming Chen, Hao-Hsuan Chang, Shen-Wei Fang and Wei-Guang Teng, Department of Engineering Science, National Cheng Kung University, Tainan, Taiwan
Based on mathematics and engineering point of view, we aim to explore the establishment of model structure, and thus calculate the benefits and risks of futures options. In this work, we exploit a large amount of market data of futures options to address two issues. The first is to use the spread strategies for risk control. The second is to discover the appropriate timing for profitable trading. The reason for using the spread strategy is that being an option seller requires to take the risk if unexpected events occur. Additionally, it is crucial to determine the appropriate timing for trading. We thus investigate the effectiveness of several technical indicators by scrutinizing the market data. Our goal in this work is to develop some simple but effective strategies for being option sellers. Experimental results show that our strategies have yielded good profit in the TAIFEX market over the past ten years.
Data Analytics, Financial Engineering, Futures Options.
APPLIANCE RECOGNITION USING A DENSITY-BASED CLUSTERING APPROACH WITH MULTIPLE GRANULARITIES
Chun-Wei Yen, Yu-Lin Ke, Sheng-Ta Chen, Yi-Chieh Pai, Hung-Chieh Wei, and Wei-Guang Teng*, Dept. of Engineering Science, National Cheng Kung University, Tainan, Taiwan
Electricity may not be economically stored as other forms of energy such that it would be in short supply during the peak time. Most power suppliers use Time-of Use method to avoid this situation, so requiring the information of power consumption is essential. With the progress of Internet of Things, smart sockets are great tools for users to manage power consumption in families. However, smart sockets on the market nowadays can only present the information of the total electricity consumption. If users want to require the information of each appliance, installing multiple smart sockets may be high cost. In this paper, we implement appliance recognition into smart sockets so that we can distinguish each appliance respectively.The proposed recursive DBSCAN method realizes the recognition task without prior knowledge of new appliances and shows an effective result.
Data Mining, Clustering, Appliances Recognition, Smart Socket, DBSCAN
Adapting an Agile Approach in Mobile E – commerce
Beshair Khalid Alsiddiq and Nor Shahida Mohd Jamail, Collage of Computer and Information Science, Prince Sultan University, Riyadh, Saudi Arabia
E-commerce nowadays has expanded in the past years, and it is substituting the traditional brick and mortar stores. And most of the brands now have their own web or mobile application so the users can shop online, it is easier and faster. Building an e-commerce application is like any other software it goes through phases in the SDLC. This paper is going to be a guide to adapt the Agile mythology in the development of the mobile e-commerce application.
E – commerce, Agile, E – commerce application
An Irregular Spatial Cluster Detection Combining the Genetic Algorithm
Shan Mei1, Yitong Zhao2, Yonglin Lei1 and Mei Yang1, 1College of System Engineering, National University of Defence Technology, Changsha, China
266136 Troop of PLA, Beijing, China
Spatial cluster detection is widely used for disease surveillance, prevention and containment. However, the commonly used clustering methods cannot resolve the conflicts between the accuracy and efficiency of the detection. This paper proposes an improved method for flexibly-shaped spatial scanning, which can identify irregular spatial clusters more accurately and efficiently. By using a genetic algorithm, we also accelerate the detection process. We convert geographic information to a network structure, in which nodes represent the regions and edges represent the adjacency relationship between regions. According to Kulldorff’s spatial scan statistics, we set the objective function. A constraint condition based on the spectral graph theory is employed to avoid disconnectedness or excessive irregularity of clusters. The algorithm is tested by analysing the simulation data of H1N1 influenza in Beijing. The results show that compared with the previous spatial scan statistic algorithms, our algorithm performs better with shorter time and higher accuracy.
Spatial cluster detection, flexibly-shaped spatial scanning, H1N1 influenza in Beijing
Fake Check Scams: A Blockchain Based Detection Solution
Badis HAMMI1 and Yves Christian Elloh Adja2, 1PSB School, Paris, France
2Telecom ParisTech, Paris, France
Fake checks are one of the most common instruments used to commit fraud against consumers. This fraud is particularly costly for victims, since they generally loose thousands of dollars as well as being exposed to judicial proceedings. Currently, there is no existing solution to authenticate checks and detect fake ones instantly. Instead, banks must wait for a period of more than 48 hours to detect the scam. In this context, we propose a blockchain based scheme to authenticate checks. More precisely, our approach helps the
banks to share information about provided checks without exposing the banks’ customers’ personal data.A proof of concept of our scheme was realized using Python language and relying on Namecoin blockchain.
Blockchain, Security, Fake check scam, Spam, Fraud detection
Ques-Chain: an Ethereum Based E-Voting System
Qixuan Zhang, Bowen Xu, Haotian Jing Sicheng Zhang and Zeyu Zheng, School of Information Science and Technology, ShanghaiTech University, China, School of Information and Electronic Engineering, Zhejiang Gongshang University, China
Ethereum is an open-source, public, blockchain-based distributed computing platform and operating system featuring smart contract functionality.In this paper, we proposed an Ethereum based eletronic voting (e-voting) protocol, Ques-Chain, which can ensure the authentication can be done without hurting confidentiality and the anonymity can be protected without problems of scams at the same time.Furthermore, the authors considered the wider usages Ques-Chain can be applied on, pointing out that it is able to process all kinds of messages and can be used in all fields with similar needs.
Electronic voting, Ethereum, Smart contracts, Blind signature
Weakly-Supervised Network Alignment with Adversarial Learning
Nguyen Thanh Toan1, Nguyen Quoc Viet Hung1, Phan Thanh Cong1 and Quan Thanh Tho2, 1School of Information Communication Technology, Griffith University, Queensland, Australia and 2Department of Computer Science and Engineering, Ho Chi Minh City University of Technology, HCMC, Vietnam
Network alignment, the task of seeking the hidden underlying correspondence between nodes across networks, has become increasingly studied as an important task to multiple network analysis. A few of themany recent applications of network alignment include protein network alignment, social network reconciliation, and computer vision. However, traditional methods which are based on matrix factorization directly work on networks themselves rather than exploit their intrinsic structural consistency, and thus their performance is sensitive to structural variations of networks. Recently, many supervised approaches which leverage latent representation have been proposed. Although they can handle large-scale datasets, most of them rely on a large number of parallel anchor links which are unavailable or expensive to obtain for many domains. Therefore, in this paper, we propose the WENA Framework, a representation learningbased network alignment, in which we study how to design weakly-supervised methods to align large-scale networks with a limit of ground truth available. Empirical results show that, with only two anchor links, WENA significantly outperforms existing unsupervised aligners and even outperforms state-of-the-art supervised methods that use richer resources in terms of both noise robustness and accuracy.
Network embedding, Graph mining, Network alignment, Graph matching, Knowledge representation
ISLA An Algorithmic Approach to Assisted Narrative Planning and Assembly
Djyron Sarroza, Institute of Computer Science, University of the Philippines Los Baños, Philippines
Intelligent Story Layout Assistant (ISLA) is a forward-chaining narrative planner based on Stephen Ware’s GLAIVE. It constructs story layouts that achieve the author’s goals while making sure that most steps in the plan have clear motivations. These layouts, or solution plans, are based on a handcrafted knowledge-base of story universe elements. ISLA provides the data structures needed to potentially further assist the author in fleshing out the produced story layout.
Artificial Intelligence, Narrative Planner
A PERSONALIZED PRODUCT RECOMMENDATION SYSTEM BASED ON VECTOR SPACE MODELS
Kiruparan Balachandran, Malankandage Ganeesha Sandeepani, and Thilina Randeniya, Innovation Quotient (Pvt) Ltd, Colombo, Sri Lanka
Recommendation engines are integrated with e-commerce platforms to provide better customer experience and improve their sales. Most e-commerce platforms struggle to achieve these milestones because of their weakly-built recommendation models. This research focuses on improving the process of building a recommendation engine. Available approaches for constructing recommendation engines are limited in various aspects. Such limitations include: (1) considering customers’ demographic characteristic to create models - what customers shared were mostly incomplete and imbalance data; (2) Customer behavior is extracted from social media - extracting data from social media at present is not easy (3) Content-based, and collaborative-filtering uses customer ratings on products to recommend products - content-based engine loops into the same product zone, and not all e-commerce platforms have a function to rate products numerically; (4) Almost no one discussed time directionality on their recommendation engines. Our study uses customer purchase history which is always complete and has balanced data. This study considers the frequency of each product bought by customers on collaborative-filtering, and vectors representing customers are defined in a way to consider time directionality. Evaluations indicate our recommendation engine is better at recommending products with 69.93% accuracy level.
Recommendation Engine, Product Similarity, User Similarity, Taste Communities
A FUSION METHOD FOR WORD VECTOR BASED ON FASTTEXT-KDTREE
DAI Yu1, Hua Hongcui1, Zhang Huixue2, Ma Chenyan1, You Jiaqi2, 4, YANG Lei2,3, 1College of Software,Northeastern University, Shenyang, China
2College of Computer Science and Technology, Northeastern University, Shenyang,China, 3Department of Computer Science, Australian National University, Canberra, Australia and 4Northeastern Network Technology Co., Ltd
Text categorization is an important part of the field of natural language processing, and it is also one of the current research hot issues. However, at present, text categorization technology still faces the problem of losing some semantic information caused by new words. For this reason, this paper proposes a fusion method for word vector based on fastText-kdTree. Firstly, the method trains word vectors by using fastText model and fills in unknown word vectors by combining n-gram model. Secondly, it uses the idea of kdTree nearest neighbor to find multiple word vectors similar to unknown words. Finally, it fuses the multiple word vectors to form a new representation of word vectors by a gate mechanism. The experimental results show that the proposed method can achieve 91.08% classification accuracy.
Text categorization, OOV, word vector, deep learning