Kongu Engineering College

About: Kongu Engineering College is a based out in . It is known for research contribution in the topics: Computer science & Cluster analysis. The organization has 2001 authors who have published 1978 publications receiving 16923 citations.

Clustering based optimal summary generation using Genetic Algorithm

Abstract: This paper presents Genetic Algorithm based sentence extraction strategy and threshold based document clustering algorithm to produce cluster wise optimal summary. Related documents are grouped into same cluster using threshold based document clustering algorithm. From each cluster important sentences are selected using feature profile which is generated by considering sentence specific features like word weight, sentence position, sentence length, sentence centrality, proper nouns in the sentence and numerical data in the sentence. Based on the feature profile sentence score is calculated for each sentence. To produce optimal summary fitness function is employed which is based on summary quality criteria like maximizing length, coverage and informativeness while minimizing the redundancy. Machine generated summaries are compared against human summaries using Precision, Recall, F-measure and ROUGE-1 measure. The experimental results shows that the proposed approach is efficient and outperforms than the existing multi-document summarization system based on genetic algorithm (MSBGA) approach.

Multiple regression analysis for the prediction of extraction efficiency in mining industry with industrial IoT

Abstract: Manufacturers and industrialists have a significant opportunity at hand in automation for the complex processes involved in manufacturing rather than labor-intensive based system analysis and control Especially industrial IoT (IIoT) technology provides far more intricate details to the industrial automation for prompt decisions automatically through a web server Hence in this present work, online data monitoring of important attributes associated with the mining industry during the extraction of zinc and lead are analyzed using IIoT The real-time analysis of the extraction efficiency rate of zinc and lead concerning temperature, pH and a particle size parameter in mining sectors is carried out by storing data in the cloud It is accomplished by using an integrated IoT module holding Revolution-Pi IIoT (IIoT) gateway with AC500 PLC to afford enhanced data communication from the mining field to the cloud server to increase the performance of the processes Based on the retrieval of historical data from the cloud, a multivariate regression model for extraction efficiency of zinc and lead is formulated by using pH, temperature and particle size as influencing parameters to estimate the predictions

Decision Making Based on Machine Learning Algorithm for Identifying Failure Rates in the Oil Transportation Pipeline

Abstract: In Oil Industry Safety Directorate (OISD), it has been testified that nearly 33% of pipeline defects are due to improper pigging and improper precast-forecasting of the existence of a crack in the long run pipelines. Therefore, pipeline engineers are requisite to exploit effective and proficient intelligent approach to identify and pinpoint these pipeline imperfections. To sort out these issues, an unsupervised machine learning technique with partition clustering algorithm is implemented to figure out the occurrence of crack or sedimentation inside the pipelines in the premature stage during the long run passage of oil through pipelines. As a result, partition clustering best fits for the observation of performance by organizing clusters as in spherical shapes which affords similarity within the cluster is higher and the similarity between the clusters is minimum. In the proposed work, for the prediction on the occurrence of anomaly in oil pipeline system the well-suit partitioning cluster approach is combined with K-means clustering.

Hyperparameter Optimization for Transfer Learning of VGG16 for Disease Identification in Corn Leaves Using Bayesian Optimization.

Abstract: One of the world's most widely grown crops is corn. Crop loss due to diseases has a major economic effect, putting the food supply in jeopardy. In many parts of the world, lack of infrastructure still slows disease diagnosis. In this context, effective detection of corn leaf diseases is necessary to limit any unfavorable impacts on the yield. This research has been carried out on the corn leaf images, having three classes of diseases and one healthy class, collected from web resources by using the densely connected convolutional neural networks (CNNs). In this work, VGG16, a variant of CNN, is investigated to classify the infected and healthy leaves. We conduct four different sets of experiments using pretrained VGG16 as a classifier, feature extractor, and fine-tuner. To improve our results, Bayesian optimization is used to choose optimal values for hyperparameters, and transfer learning is explored to fine-tune and reduce the training time of the proposed models. In comparison with earlier proven methods, transfer learning on VGG16 produced better results by leveraging a test accuracy of more than 97% while requiring less training time.