International Journal of Adaptive and Innovative Systems (7 papers in press)
Multi-Domain Intelligent System for Document Image Retrieval
by Donato Barbuzzi, Alessandro Massaro, Angelo Galiano, Leonardo Pellicani, Giuseppe Pirlo, Matteo Saggese
Abstract: This paper presents an experimental analysis on document image retrieval using a multi-domain intelligent system. More specifically, on the same document image, the combination of three different domains: layout, logo and signature is discussed. This new method analyzes every single decision provided by multi-domain system so that, in the training phase, a new sample classified with a dissimilar confidence to the previous trained samples is used to update the system. DTW, Euclidean Distance and Cosine Similarity have been used, respectively for the analysis of layout, logo and signature. Finally, the weighted combination of individual decisions was considered. The experimental results, carried out on 30 rotated forms belonging to 13 different companies, demonstrate the superiority of the proposed approach with respect to single-domain retrieval systems, based on the ANR performance index. The ANR parameter is able to evaluate the multi-domain system.
Keywords: Document Management System; Document Image Retrieval; Multi-Expert Intelligent System; Feedback-based strategy; Instance Selection.
Prediction of Instantaneous Heart Rate Using Adaptive Algorithms
by Sarita Kansal, P.P. Bansod, Abhay Kumar
Abstract: In this paper, adaptive filter based on adaptive algorithms like Least Mean Square (LMS), Normalized Least Mean Square (NLMS) and Recursive Least Square (RLS) are used for the prediction of instantaneous heart rate in ECG signal. The adaptive algorithms works on the principle of optimizing the least square error by achieving wiener solution. The weights of the filter coefficients are changing, as per the changes in the signal. There is always the issue of selecting parameters of adaptive filter for adaptation, which affects the performance of it. The parameters are like step size, filter length, minimum number of iterations etc. Therefore, the performance of adaptive filter is analyzed by Mean Square Error (MSE) with varying parameters and using different adaptive algorithms. The prediction accuracy is observed by measuring parameter Mean Absolute Error (MAE). The total ten healthy records of ECG are considered to evaluate the performance of adaptive algorithms and results are presented after averaging the results of each record. The simulation results show that the adaptive algorithms NLMS and RLS have faster convergence rate with less number of iteration than LMS but the forecasting accuracy is higher in LMS compared to NLMS and RLS algorithms. The complexity of RLS algorithm is more as it takes the average value of past error, whereas in LMS and NLMS only instantaneous error is considered.
Keywords: ECG; Instantaneous Heart Rate; Adaptive Algorithm; LMS; NLMS; RLS.
Designing DNA Code: Quantity and Quality
by Bin Wang, Kaiqiang Liu
Abstract: With the development of modern technology, traditional computers have reached the physical limit in information storage and parallel computing. In order to meet the needs of large-scale computing, the development of a high-performance computer has become the focus of the scientific community. DNA computing was born with this as one of its objectives. DNA computing is essentially implemented based on a new computational model of DNA molecules. Because of its high parallelism, massive information storage capacity, and low energy consumption, it has attracted the attention of scholars worldwide. The design of DNA code is a critical step in the realization of DNA computing. The quantity and quality of the code can directly affect the accuracy and efficiency of DNA computing. At present, the research direction of DNA code is divided into set design and quality optimization. Set design is to improve the lower bound of the code. Quality optimization makes DNA code meet more constraints. DNA code studies have made the results of DNA computing more reliable. This paper summarizes the code problems in the DNA computing process, the combinatorial constraints that the code needs to satisfy, and the intelligent optimization algorithm used in code. Finally, the problems of DNA code and the future development direction are analyzed.
Keywords: DNA computing; DNA code; Combinatorial constraints; Intelligent optimization algorithm.
A Chain Membrane Model with Application in Cluster Analysis
by Yuzhen Zhao, Xiyu Liu, Wenxing Sun
Abstract: Membrane computing is a kind of bio-inspired parallel distributed computing paradigm which can reduce computational complexity by the strategy of a space-time tradeoff. Traditionally, there are three kinds of membrane computing models (P systems) based on the tree and the graph topological structures. In this paper, a new P system with chain topological structure is proposed which is called the chain P systems. In the chain P systems, membranes, objects and rules are all in the form of chains which can store more information and therefore further improve the computational efficiency. The computational power and efficiency of the chain P systems are analysed. The graph clustering and the ROCK clustering algorithms based on the chain P systems are given as applications.
Keywords: Membrane Computing; Membrane Model; Chain P System; Computational Power; Computational Efficiency; Graph Clustering; ROCK Clustering.
Adaptive Spam Filtering System Using Complement Na
by Michael Adegoke, Olayide Abass
Keywords: Spam; Spam filtering; complement naïve bayes; adaptive filtering; prior; bias; adaptive; skewedness;filter.
An Optimal RSSI based Cluster-Head Selection for Sensor Networks
by KHUSHBOO JAIN
Abstract: The energy utilization is one of the most common challenges in Wireless Sensor Network (WSN), as frequent communication between the sensor nodes (SNs) results in huge energy drain. Moreover, optimization and load-balancing within the WSN are the significant concern to grant intellect for the extensive period of network lifetime. As a matter of fact, many WSNs are deployed and operating outdoors is exposed to varying environmental conditions, which may further set grounds for severe performance degradation of such networks. Therefore, it is necessary to take into consideration the factors like radio signal strength in order to reduce the impact and to adapt to varying environmental conditions. Since clustering is a topological control technique to reduce the activity of SNs transceivers, it extensively increases overall system scalability and energy efficiency. It selects CH to manage the entire network to achieve longevity in WSN. In this paper, we present an optimal CH selection (OCHS) algorithm which is also based on environmental conditions to achieve energy efficiency and enhanced network lifetime. The originality of this work is that we have taken into consideration the received signal strength index (RSSI) of SNs from the base-station (BS). The OCHS algorithm mainly focuses on maximizing the network lifetime based on RSSI values and residual energy levels of SNs. The OCHS algorithm is simulated on Cooja Simulator and its performance is compared with existing LEACH and HEED protocols. Simulation analysis and results proved that our OCHS algorithm can effectively enhance the network lifetime by two times and thus it is an energy-efficient way to choose a CH.
Keywords: Cluster Head Selection; Energy Efficiency; Network Lifetime; Residual Energy; RSSI; WSNs.
Predicting the Net Asset Value of Mutual Fund: An Extended Literature Review
by Shikha Singla, Gaurav Gupta
Abstract: : Mutual funds have emerged as the most dynamic segment of the Indian financial system. With its potential to provide higher return by investment in diversified securities, mutual funds emerge as one of the most promising investments in uncertain markets. With a variety of mutual funds competing in the present scenario market, it becomes a challenging decision for the investor to balance risk and return trade off on the portfolio to maximize returns. Therefore, the important aspect for portfolio manager is to predict the Net Asset value (NAV) of mutual funds. Various methods and techniques in the field of economics and computer science have been used in the quest to gain insights into NAV prediction. The aim of the paper is to provide an extended literature review of different techniques in different areas of computer science in order to explore the future possibilities. This paper explores the past research work and proposes the future road-map to predict the NAV of mutual funds by using different techniques with greater accuracy.
Keywords: Net Asset Value (NAV); Radial Basis Function (RBF); Functional Link Artificial Neural Network (FLANN); Mutual Fund.