Papers by Dr. Radha Senthilkumar
… and Networking, 2008
... [II] Radha Senthilkumar, A. Kannan, V. Jasher Jenitha, R. Kalpana, P. Karthikeyan, "... more ... [II] Radha Senthilkumar, A. Kannan, V. Jasher Jenitha, R. Kalpana, P. Karthikeyan, "Querying and Maintaining a compact Redundancy Free XML Storage", International conference on Advanced computing, March 2008. [12] S. Harrusi, A. Averbuch, A. Yehudai. ...
Proceedings of the International Conference on Informatics and Analytics
Sequential rule mining is an important data mining issue which has numerous applications. They ar... more Sequential rule mining is an important data mining issue which has numerous applications. They are profoundly used in predicting the behaviour of learners in Educational data, predicting the web traversal patterns, finding the consecutive connections between gene expressions of different patients in Bio Informatics, determining the purchase pattern of customers in shop etc. Mining for sequential rules common to multiple sequences has some drawbacks such as strict ordering between items, because of which several rules may represent the same situation, similar rules are rated very differently and, rules may be too specific and less likely to be useful, sometimes none of the rules would match the new sequence. Thus, a more broad type of sequential rules common to multiple sequences, such that items in the forerunner and in the resulting of every rule are unordered, is required. These are called partially ordered sequential rules. (POSR). Rule Growth Algorithm and T-Rule Growth algorithm are used for mining the POSR. Both Rule Growth and TRule Growth Algorithm gives rise to lesser number of rules with greater prediction accuracy compared to mining sequential rules common to multiple sequences. The proposed work focuses on making these partially ordered sequential rules adaptive to the changes that occur over course of time. Two approaches are used to bring out adaptive rules. The first approach uses rating as the key parameter to eliminate the weakest rules and strengthen the stronger rules. The second approach classifies the rules based on their rule scores into different categories of strength and uses fuzzy inference system to infer the incremented rule scores. The performance of Rule Rating algorithm (Incremental approach) seems to have a better execution time with respect to Recomputation (i.e Rule Growth / T-Rule Growth algorithm applied for the entire dataset).
Lecture Notes on Data Engineering and Communications Technologies, 2019
Big data applications are very large in size in terms of map and reduce tasks and it takes large ... more Big data applications are very large in size in terms of map and reduce tasks and it takes large time to execute those map reduce based big data applications. Map Reduce is a familiar bulk data processing concept for extensive data computing in cloud environment. The existing Energy Conscious Arrangement is implemented with static map and reduce slot allocation, where there is no room for effective resource utilization. In case of static slot allocation, the pre-determination of map and reduce slots does not employ efficient run time performance and the slots can be strictly under-utilized. To deal with it, the proposed work discovers and enhances the resource utilization and execution time. In the proposed work, dynamic slot allocation technique is accomplished with the existing Energy Conscious Arrangement concept. The technique called Energy Efficient Dynamic Slot Allocation (EEDSA) is accomplished by using dynamic slot allocation technique, which ensures efficient resource utili...
Classification of label-specific users’ diversified interests is the most formidable task in pers... more Classification of label-specific users’ diversified interests is the most formidable task in personalized news recommendation systems (PNRS). To bring personalization to PNRS, many remarkable features have to be considered from their user profile to classify their interest. In this paper, 13, 346 features are considered per user to classify their interest for 15 labels using Multi-label Convolution Neural Network (MLCNN). The efficiency of MLCNN highly depends on its architecture through the tuning of its hyper parameters. Generally, researchers have manually designed a constant CNN architecture for each label and every label and verified the effectiveness, but this leads to additional complexity as well as large computational resources were consumed. Moreover, Designing the structure for all 15 labels leads to an increase in network structure exponentially with an increase in labels. Hence, in this paper, MLCNN architectures are optimized by implementing a novel approach Modified G...
2018 Tenth International Conference on Advanced Computing (ICoAC)
Face recognition is still a challenging problem as there is a high possibility that the differenc... more Face recognition is still a challenging problem as there is a high possibility that the differences existing within a person or subject can exceed the differences present between different persons. Most of the current research work in the field of biometric face recognition has dealt with (Red-GreenBlue) RGB images. However, Hyper Spectral Imaging (HSI), introduces the spectral dimension for improved discrimination and leads to building a more efficient face recognition system. The spectral dimension adds more intricate details to the image. As a result hyper spectral imaging helps in improving face recognition accuracy. The faces are captured at varying spectral wavelengths of the electromagnetic spectra. Hyper spectral imaging often increases facial discrimination by capturing more biometric measurements and thus revealing information that is not revealed by the commonly used RGB images. The proposed methodology for face recognition using hyper spectral imaging includes performing band selection and band fusion on the hyper spectral face cubes and then classifying them using 3 Dimensional Convolution Neural Networks (3D-CNN). The classification using 3D-CNN gave a promising accuracy of 97.3%.
2017 Ninth International Conference on Advanced Computing (ICoAC), 2017
Maintaining network security is very important and tedious in today's world. Since web applic... more Maintaining network security is very important and tedious in today's world. Since web applications are not built on sound security methodology, they are the major target for the attackers. Analyzing access logs for detecting anomalous activities is a form of defense achieved in this paper. Anomaly detection is important because if the anomalies are not detected apriori, it may lead to hacking of the entire system. This paper is based on analyzing the stored access logs and detecting the anomalous events. Our experiment evaluates both static and dynamic logs. In dynamic implementation, the pattern matching approach is used to detect the anomalies from access logs. In Weka, the supervised neural network approach gives better anomaly prediction than unsupervised neural network approach for static logs. Maximum prediction accuracy is achieved in supervised neural networks by using Naive Bayes Multinomial Text Algorithm. Since the input attributes (logs) are strings, the use of Baye...
New Trends in Computational Vision and Bio-inspired Computing
Remote sensing is the acquisition of physical characteristics (reflecting radiation) of the remot... more Remote sensing is the acquisition of physical characteristics (reflecting radiation) of the remote object. It can be collected via special cameras or sensors in the satellite or aircraft or weather balloons. Each remote sensing images has multiple spectral bands. The remote sensing images analysis is used by multiple applications like metrological prediction, Land Cover and Land Usage prediction (LCLU), vegetation change detection. Missing image in the remote sensing time series produces a lot of glitches, causing serious upshot in the multi-temporal analysis, when the images at various time stamps are missing over a period of time. The existing work reconstructs missing image in remote sensing time series via spatial and temporal data. The proposed method Tensor-Deep Stacking Network Spatial-Temporal-Spectral (TDSN-STS) helps to reconstructs the missing image in remote sensing time series using spatial, temporal and spectral data. Thus the accuracy of the reconstructed image in TDS...
Due to the inherent flexibility in both structure and semantics, XML documents are massive in nat... more Due to the inherent flexibility in both structure and semantics, XML documents are massive in nature. The ratio of the size of the XML document to the size of the text data in it is usually large. Apart from data values, the huge size of the XML document is contributed by its tree structure. The structure of the XML document tightly bounded with the data renders the original form of XML less efficient in terms of both time and space. The problem of designing a compressor for XML documents which facilitates both update and query operations has turned the attention of many. In this paper, we propose an efficient storage scheme for XML documents called XQUICK. XQUICK exploits the high regularity of XML documents to compress the tree structure. It also handles updates in an efficient manner with minimum space and time overhead. This paper also describes a novel path-based querying approach that supports fast querying. Additional mechanisms such as indexing are provided to elicit faster ...
Remote sensing provides timely and reliable information about urban areas and their changing patt... more Remote sensing provides timely and reliable information about urban areas and their changing patterns. Detection of urban change is an important factor that affects the supply of food supply, water supply, etc. Detection of land use and land change is essential in order to understand the current situation and plan for the future to avoid scarcity. In this work, we proposed two new framework called Tensor - Deep Stacking Network (TDSN) with back propagation and Deep Stacking Convolution Neural Networks (DS - CNN) for land cover change detection. Changes are detected between the year 2005 and 2015 in the month of September over the Hutong area, Beijing, China. The work has been validated by using Sentinel 2 (4 bands cirrus, water vapor, coastal aerosol, red edge band) and Landsat 8(11 bands) satellite image. All the bands from Landsat −8 image are pan-sharpen to 15m resolution to get more accurate classification result. Bands in the Sentinel −2 satellite image are used to remove cloud...
Big Data Analytics is the process of examining a large volume of data which is collected from var... more Big Data Analytics is the process of examining a large volume of data which is collected from various sources. It plays a vital role in Intelligent Transportation Systems. Taxi is absolutely the most prevailing type of on-demand transportation service in citified areas because they offer more and better services and also it provides a comfortable travel to the passengers. But it makes the metropolitan areas to suffer from inefficiencies of taxis due to the uncoordinated management of the dispatch systems. Many transport organizations stumble to provide the proper dispatching of the vehicles. Hence an effectual taxi dispatching system is provided using Hadoop map reduce framework. The main goal of this system is to produce an optimized dispatch for anticipated future request for taxis thereby minimizing the total idle driving distance. This is achieved by making predictions in the historical data. The predictions helps the taxi dispatching system to locate more taxis in the predicted...
Advances in Intelligent Systems and Computing
Big data deals with the prodigiously and sizably voluminous volume of data engendered at high spe... more Big data deals with the prodigiously and sizably voluminous volume of data engendered at high speed and it is arduous to process and manage with the subsisting database management tools. Query processing in astronomically immense data is a challenging task and frequently encounters problem. To overcome the complexity involved in processing the larger dataset, query optimization is the promising solution. Performance is a bottleneck, when complicated queries access an unbounded amount of data, resulting in high response time using the existing query optimization technique. In proposed work, to surmount this issue, scale independence is identified with access schema and query execution is optimized with invariant data. With astronomically immense precomputation and incremental computation dataset is used for querying. In precomputation approach, the invariant data is computed afore execution and thus resulting in lesser computation time during query processing. Incremental computation technique is applied to optimize the query for the streaming data. Thus, the invariant data is computed incrementally with the incipiently inserted data along with the precomputed data and then utilized for the query processing. By applying these approaches for optimizing scale independent queries, the performance can be ameliorated with tolerable response time.
Journal of Intelligent & Fuzzy Systems
Face recognition is one of the best applications of computer recognition and recent smart house a... more Face recognition is one of the best applications of computer recognition and recent smart house applications. Therefore, it draws considerable attention from researchers. Several face recognition algorithms have been proposed in the last decade, but these methods did not give the efficient outcome. Therefore, this work introduces a novel constructive training algorithm for smart face recognition in door locking applications. The proposed Framed Recurrent Neural Network with Mutated Dragonfly Search Optimization (FRNN-MDSO) Strategy is applied to face recognition application. The steady preparing system has been utilized where the training designs are adapted steadily and are divided into completely different modules. The facial feature process works on global and local features. After the feature extraction and selection process, employ the improved classifier followed by the Framed Recurrent Neural Network classification technique. Finally, the face image based on the feature libra...
Concurrency and Computation: Practice and Experience
2016 Eighth International Conference on Advanced Computing (ICoAC)
The advent of social and collaboration networks has resulted in different methods of forming larg... more The advent of social and collaboration networks has resulted in different methods of forming large groups to deal with complex tasks. Team Formation (TF) in online social networks is crucial in several applications, such as collaborative software development and community based question and answer forums. The problem involves the formation of expert teams from a social network who can collaborate effectively under multiple constraints. In a practical scenario, the problem involves a minimization of the following major objectives: communication cost, expert cost and the size of the team. The minimization is performed by finding Pareto-Optimal teams, in which no team dominates the solution teams in terms of the three chosen objectives. Existing approaches use approximation algorithms and cannot be easily extended to incorporate additional objectives. Therefore, an optimization framework the Non-dominated Sorting Genetic Algorithm for Team Formation (NSGA-II TF) is proposed which is robust and extensible. A mapping scheme is defined for representing the chromosome in the NSGA-II algorithm to satisfy the constraints of the TF problem. The scalability, precision and recall for NSGA-II TFare evaluated and it is observed that it results in teams with minimized cardinality, communication and expert cost.
Learning and Analytics in Intelligent Systems
Today cloud computing is an evolved form of utility computing which is widely used for commercial... more Today cloud computing is an evolved form of utility computing which is widely used for commercial computing needs. The Cloud service provider’s success, profit, and efficiency lie in optimally allocating the computing resources to users from a vast pool of resources. The ability to allocate resources in a ubiquitous, seamless and on-demand connection involves serious challenges. Task scheduling is a variant of job-shop scheduling problem which is categorized as NP-COMPLETE. In this paper, a novel meta-heuristic algorithm of hybrid Firefly-Genetic combination is propounded for scheduling tasks. The proposed algorithm blends benefits of a mathematical optimization algorithm like Firefly with an evolutionary algorithm like Genetic algorithm to form a powerful metaheuristic search algorithm. The proposed hybrid Firefly-Genetic algorithm was able to schedule the tasks with the objective of minimal execution time for all tasks and a swift convergence to the near optimal solution. The proposed algorithm was tested in CloudSim which is a simulator toolkit for testing cloud-based algorithms. The experimental results showed that the proposed algorithm outweighed the performances of traditional First In First Out (FIFO) and Genetic algorithms.
Uploads
Papers by Dr. Radha Senthilkumar