Browsing by Author "Subbiah Karthikeyan"

Now showing 1 - 12 of 12

An Econometric Time Series Forecasting Framework for Web Services Recommendation
(Elsevier B.V., 2020) Vijendra Pratap Singh; Manish Kumar Pandey; Pangambam Sendash Singh; Subbiah Karthikeyan
The convergence of SMAC technologies resulted in an unexpected upsurge of web services on the internet. The flexibility and rental approach of the cloud makes it an attractive option for the deployment of web services-based applications. Once a number of web services are available to gratify the similar functionalities, then the choice of the web service based on personalized quality of service (QoS) parameters plays an important role in deciding the selection of the web service. The role of time is rarely being discussed in deciding the QoS of web services. The delivery of QoS is not made as declared due to the correlated behavior of non-functional performance of web services with the invocation time. This happens because service status usually changes over time. These limitations have affected the performance of neighborhood-based collaborative filtering. Hence, the design of the time aware web service recommendation system based on the personalized QoS parameters is very crucial and turn out to be a challenging research issue. In the current work, various econometric models are used for experimentation with the input time series to find the best fit model for the prediction of personalized QoS based web services recommendation. The Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) value is used as an evaluation metric and their value for the prediction of Response time is found to be 1.7392e+04 and 1.7416e+04 respectively with the ARMA model. The AIC and BIC value for throughput (TP) is found to be 1.5249e+04 and 1.5334e+04 respectively with the ARIMA. The value of variance is 13.2437 and 6.8131 for RT and TP respectively which is also the lowest among other models with t-statistic greater than the p-value. Thus, the experimental results show that the ARMA and ARIMA model of Time Series Forecasting for Web Services Recommendation Framework is performing better in case of Response time and throughput respectively. © 2020 The Authors. Published by Elsevier B.V.
An empirical mode decomposition (EMD) enabled long sort term memory (LSTM) based time series forecasting framework for web services recommendation
(IOS Press BV, 2019) Vijendra Pratap Singh; Manish Kumar Pandey; Pangambam Sendash Singh; Subbiah Karthikeyan
The convergence of SMAC technologies resulted in an unexpected upsurge of web services on the internet. The flexibility and rental approach of the cloud makes it an attractive option for the deployment of web services based applications. Once a number of web services are available to gratify the similar functionalities, then the choice of the web service based on personalized quality of service (QoS) parameters plays an important role in deciding the selection of the web service. The role of time is rarely being discussed in deciding the QoS of web services. The delivery of QoS is not made as declared due to the correlated behavior of non-functional performance of web services with the invocation time. This happens because service status usually changes over time. These limitations have affected the performance of neighborhood based collaborative filtering. Hence, the design of the time aware web service recommendation system based on the personalized QoS parameters is very crucial and turn out to be a challenging research issue. In the current work, empirical mode decomposition (EMD) enabled deep learning model long short term memory (LSTM) is used for the prediction of these time aware QoS parameters and the results are compared with the previous approaches. The experimental results show that the EMD-LSTM based Time Series Forecasting Framework is performing better. The RMSE, MAE and MAPE are used as an evaluation metric and their value for the prediction of Response time (RT) is found to be 0.085661, 0.049031 and 1.46208 respectively. The RMSE, MAE and MAPE are used as an evaluation metric and their value for the prediction of throughput (TP) is found to be 0.043878, 0.030688 and 1.485613 respectively. Thus, the experimental results show that the EMD-LSTM model of Time Series Forecasting for Web Services Recommendation Framework is performing better as compared to previous methods. © 2019 The authors and IOS Press. All rights reserved.
An LSTM Based Time Series Forecasting Framework for Web Services Recommendation
(Instituto Politecnico Nacional, 2020) Vijendra Pratap Singh; Manish Kumar Pandey; Pangambam Sendash Singh; Subbiah Karthikeyan
The convergence of Social Mobility Analytics and Cloud (SMAC) technologies gives rise to an unforeseen aggrandization of the web services on the internet. The resilience and payment-based approach of the cloud makes it an obvious choice for the deployment of web services-based applications. Out of available web services, to gratify the similar functionalities, the choice of the web service based on the personalized quality of service (QoS) parameters plays an important role in determining the selection of the web service. The role of time is rarely being discussed in deciding the QoS of web services. The delivery of QoS is not made as declared due to the non-functional performance of web services correlated behavior with the invocation time. This happens because service status usually changes over time. Hence, the design of the time aware web service recommendation system based on the personalized QoS parameters is very crucial and becomes a challenging research issue. In this study, LSTM based deep learning models were used for the prediction of these time aware QoS parameters and the results are compared with the previous approaches. The experimental results show that the LSTM based Time Series Forecasting Framework is performing better. The RMSE, MAE, and MAPE are used as an evaluation metric and their value for the prediction of Response time (RT) is found to be 0.030269, 0.02382 and 0.59773 respectively with adaptive moment estimation as the training option and is found to be 0.66988, 0.66465 and 27.9934 respectively with root mean square propagation as the training option. The RMSE, MAE, and MAPE value for the prediction of throughput (TP) is found to be 0.77787, 0.4792 and 1.61 respectively with adaptive moment estimation as the training option and is found to be 0.2.7087, 1.4076and 7.1559 respectively with root mean square propagation as the training option respectively. Thus, the experimental results show that the LSTM model of Time Series Forecasting for Web Services Recommendation Framework is performing better as compared to previous methods. © 2020 Instituto Politecnico Nacional. All rights reserved.
Discrimination of psychrophilic and mesophilic proteins using random forest algorithm
(2012) Abhigyan Nath; Radha Chaube; Subbiah Karthikeyan
Psychrophilic organisms are those organisms which thrive at very low temperatures. In order to carry out the normal physiological and biochemical functions, these organisms produces psychrophilic proteins that have evolved through a vast amount of physicochemical adaptations at the sequence and structural levels. Our study is focussed on selecting suitable classification algorithm and appropriate input features for better discrimination of psychrophilic protein sequences from mesophilic protein sequences. We have used amino acid composition and hydrophobic residue patterns as input features and found Random Forest algorithm, a recently developed ensemble machine learning technique for better discriminating between mesophilic and psychrophilic proteins. A balanced dataset with 6000 mesophilic and 6000 psychrophilic sequences for training, and with 8432 psychrophilic and 3169 mesophilic sequences for testing was created and used for experiments. Discrimination using only the statistically significant amino acids taken from previous literature was also experimented. For the first time 70.3% testing accuracy is being reported with 71.3 % correctly predicted psychrophilic and 67.7 % correctly predicted mesophilic proteins. © 2012 IEEE.
Enhanced classification of hyperspectral images using improvised oversampling and undersampling techniques
(Springer Science and Business Media B.V., 2022) Pangambam Sendash Singh; Vijendra Pratap Singh; Manish Kumar Pandey; Subbiah Karthikeyan
In the era of climate change, monitoring and effective retrieval of soil, water bodies, vegetation parameters etc. are of utmost importance which is successfully being executed using remote sensing from last few decades. The advancement of technologies has enabled us to reach effective decision making through these sensors. The advantage of acquiring multitemporal spatially continuous data sometimes turns into a disadvantage due to class imbalance where minority class instances are often misclassified by most of the classifiers. The current work explored the solution to handle this problem by resampling the datasets before the application of classification algorithms by proposing a new computationally efficient class wise resampling technique which is based on SMOTE and centroid-based clustering. The experiment was conducted on two benchmarked publicly available hyperspectral datasets. The output of the current work shows the superiority of the current work over past studies based on the performance evaluation metrics, accuracy, precision, recall and kappa values. © 2021, Bharati Vidyapeeth's Institute of Computer Applications and Management.
Enhanced classification of remotely sensed hyperspectral images through efficient band selection using autoencoders and genetic algorithm
(Springer Science and Business Media Deutschland GmbH, 2022) Pangambam Sendash Singh; Subbiah Karthikeyan
Hyperspectral images (HSIs) contain significant number of contiguous dense spectral bands which often have large redundancy and high correlation that subsequently results into “curse of dimensionality” in HSI analysis. Therefore, efficient band selection techniques are crucial for dimensionality reduction of HSIs without any significant loss of spectral information contained in it. In this paper, deep learning autoencoders and genetic algorithm (GA) are used for efficient selection of the most revealing bands from a remotely sensed HSI. The proposed method formulates the HSI band selection process as a GA-based evolutionary optimization that minimizes the reconstruction error of an autoencoder which uses a few informative bands for HSI reconstruction. The proposed approach starts with spectral segmentation of the bands in an HSI into a number of spectral regions, and then, different autoencoders are trained on each segment with the original input band vectors contained in the segmented region. Finally, GA-based search heuristics is applied on each region in order to find out sparse sub-combination of spectral bands in such a way that the trained autoencoders would reconstruct the original segmented spectral vectors from the resulting band sub-combinations with least reconstruction errors. The final band selection is carried out by aggregating all the band sub-combinations returned from the segmented regions. Finally, the effectiveness of the proposed method is verified through selected bands validation by a support vector machine classifier. Experimental results on three publicly available HSI datasets depict the consistently superior effectiveness of the proposed band selection method over other state-of-the-art methods in land cover classification of remotely sensed HSIs. © 2021, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.
Hybrid MLP-GRU Model With PCA-Optimized Feature Selection for DDoS Attack Detection in IoMT Networks
(Institute of Electrical and Electronics Engineers Inc., 2025) Uthayanathan Priyatharsan; Subbiah Karthikeyan
The Internet of Medical Things (IoMT) has revolutionized healthcare, but faces significant cybersecurity challenges, particularly distributed denial of service (DDoS) attacks. This research introduces a novel and efficient hybrid deep learning model for DDoS attack detection by combining a Multi-Layer Perceptron (MLP) and a gated recurring unit (GRU). The proposed framework incorporates data standardization and Principal Component Analysis (PCA) for dimensionality reduction and feature selection. The framework was evaluated using the CICIoMT2024 dataset, which specifically contains the traffic from IoMT security attacks. The experimental results demonstrate that proposed MLP-GRU model outperforms several existing classifiers and relevant models, achieving 99.99% accuracy, 100% recall, 99.99% precision and 99.99% F1 score for 6 PCA components. This hybrid model offers improved computational efficiency and robustness for IoMT systems. This research contributes to improving the security and operational integrity of IoMT systems against the growing threat of DDoS attacks. © 2025 IEEE.
Local Binary Ensemble based Self-training for Semi-supervised Classification of Hyperspectral Remote Sensing Images
(Instituto Politecnico Nacional, 2020) Pangambam Sendash Singh; Vijendra Pratap Singh; Manish Kumar Pandey; Subbiah Karthikeyan
Supervised classification of hyperspectral remote sensing images is still challenging due to the scarcity of enough labelled samples. Semi-supervised methods have been adopted to handle this issue. Self-training is a popular semi-supervised technique which is widely used for training a classifier with limited labelled data and a large quantity of unlabeled data. However, traditional self-training approaches often give poor classification results in high dimensional data. In the current work, a novel efficient self-training approach for handling the deficiency of labelled samples for semi-supervised classification of hyperspectral remote sensing images is proposed. The proposed method first trains an ensemble of locally specialized supervised binary classifiers independently by using the dimensionally reduced spectral feature vectors of few available labelled samples. The trained local binary classifiers are then used to extend the labelled set by iterative addition of highly informative unlabeled samples to it by exploiting both the spectral and spatial information of the hyperspectral image. The classifiers are then retrained with the extended dataset in a batchwise manner and the procedure is repeated until adequate quantity of labelled samples are generated. Finally, a supervised multiclass classifier is trained on the extended dataset to produce the final classification map. Experimental results on two benchmark hyperspectral image datasets prove the effectiveness of the proposed method over supervised and traditional self-training based semi-supervised pixelwise classification approach in terms of different classification measures. © 2020 Instituto Politecnico Nacional. All rights reserved.
Neural Net Time Series Forecasting Framework for Time-Aware Web Services Recommendation
(Elsevier B.V., 2020) Vijendra Pratap Singh; Manish Kumar Pandey; Pangambam Sendash Singh; Subbiah Karthikeyan
The convergence of Social Mobility Analytics and Cloud (SMAC) technologies resulted in an unexpected upsurge of web services on the internet. The flexibility and rental approach of the cloud makes it an attractive option for the deployment of web services-based applications. Once a number of web services are available to gratify the similar functionalities, then the choice of the web service based on personalized quality of service (QoS) parameters plays an important role in deciding the selection of the web service. The role of time is rarely being discussed in deciding the QoS of web services. The delivery of QoS is not made as declared due to the correlated behavior of the non-functional performance of web services with the invocation time. This happens because service status usually changes over time. These limitations have affected the performance of neighborhood-based collaborative filtering. Hence, the design of the time aware web service recommendation system based on the personalized QoS parameters is very crucial and turns out to be a challenging research issue. In the current work, various neural network models like Levenberg Marquardt (LM), Bayesian-Regularization (BR) and Scaled-Conjugate-Gradient (SCG) are used for experimentation with the input time series to find the best fit model for the prediction of personalized QoS based web services recommendation. The Pearson's Correlation Coefficient is used as an evaluation metric and their value for the prediction of Response time is found to be 0.84985 and for throughput (TP) is found to be 0.99082 with the Levenberg Marquardt algorithm. Thus, the experimental results show that the with the Levenberg Marquardt model of Time Series Forecasting for Web Services Recommendation Framework is performing better in case of Response time as well as throughput. © 2020 The Authors. Published by Elsevier B.V.
One-class Classifier Ensemble based Enhanced Semisupervised Classification of Hyperspectral Remote Sensing Images
(Institute of Electrical and Electronics Engineers Inc., 2020) Pangambam Sendash Singh; Vijendra Pratap Singh; Manish Kumar Pandey; Subbiah Karthikeyan
The scarcity of labelled training data as well as uneven class distribution among the limitedly available labelled data have posed a critical issue in supervised hyperspectral remote sensing image classification. Semisupervised methods can be an easy solution to this critical problem. However, traditional self-training based semi-supervised approaches often give poor classification results in high dimensional multiclass classification problems. This paper proposes a novel efficient one-class classifier ensemble based self-training approach for semisupervised classification of hyperspectral remote sensing images with limited labelled data. The proposed method initially trains an ensemble of locally specialized one-class classifiers independently by using the dimensionally reduced spectral feature vectors of the available labelled samples. The trained one-class classifiers are then used to extend the labelled set by iterative addition of high quality unlabelled samples to it through the exploitation of both spectral and spatial information. The classifiers are then retrained with the extended dataset in a batchwise fashion. The procedure is repeated until an adequate quantity of labelled samples are generated. Finally, a supervised multiclass classifier is trained on the extended dataset for the final image classification purpose. Experimental results on two benchmark hyperspectral images verify the effectiveness of the proposed method over supervised and traditional self-training based semisupervised pixelwise classification in terms of different classification measures. © 2020 IEEE.
Salient object detection in hyperspectral images using deep background reconstruction based anomaly detection
(Taylor and Francis Ltd., 2022) Pangambam Sendash Singh; Subbiah Karthikeyan
Salient object detection is a significant task that forms the basis for many image processing and computer vision applications. In recent years, the research in this area is being extended beyond RGB images to applications involving multispectral and hyperspectral images. But most of the existing algorithms for salient object detection often yield an incomplete representation of the object and often produce saliency maps with blurred edges. In this paper, we propose an efficient hyperspectral image salient object detection method through anomaly detection by combining deep learning autoencoders with one-class support vector machines. Here, the saliency detection problem in hyperspectral images is formulated as an unsupervised deep background spectral reconstruction-based anomaly detection. Our proposed method first employs deep autoencoders to model the background of an input hyperspectral image in terms of spectral reconstruction residuals of the autoencoders and then detect the salient objects from the image through a one-class support vector machine-based anomaly detection. The proposed method was evaluated on a publicly available hyperspectral image dataset for salient object detection. The experimental results show that our proposed method is found to be more efficient and superior over other previous methods in terms of various performance measures. © 2021 Informa UK Limited, trading as Taylor & Francis Group.
Unsupervised deep autoencoder-based reconstruction for ink mismatch detection in hyperspectral document images
(Springer Science and Business Media B.V., 2025) Pangambam Sendash Singh; Subbiah Karthikeyan; Govind Murari Upadhayay; Sounak Sadhukhan; Pramod Kumar Soni; Timothy Malche
Ink analysis is crucial for finding ink mismatches in handwritten documents to determine their authenticity and identify potential forgery. Traditional chemical-based methods like thin layer chromatography are effective for this purpose, but they are destructive, irreversible, time-consuming, and sensitive to environmental factors. Hyperspectral document images (HSDIs) which capture the spectral information in multiple bands can reveal the material composition, thereby allowing the identification of different inks by their distinct spectral characteristics, even when the inks visually appear to be the same colour. Hyperspectral imaging thus shows great promise for forensic document analysis and authentication. Despite its potential, research in this area is still emerging. Existing HSDI-based methods show promise in detecting ink mismatches but these methods often require prior spectral information and ground truth data for the ink pixels, limiting their practical applicability. Unsupervised methods present a solution by removing the need for prior information. This work proposes a novel unsupervised ink mismatch detection method in HSDIs using deep learning-based hyperspectral analysis. The proposed method formulates ink mismatch detection as a decision problem that determines whether a deep autoencoder trained on the spectral features of specific ink pixels would be able to reconstruct the spectral features of unseen complementary ink pixels. The proposed framework works in a fully unsupervised manner, learning the spectral representations directly from data without any labeled samples or prior spectral information of the inks present. Unlike most existing methods which are supervised, the proposed unsupervised method is more practically applicable for real world forensic document analysis. Experimental results on a publicly available HSDI dataset demonstrate the superiority of the proposed method over existing methods in identifying ink mismatches in potentially fraudulent document images. © The Author(s) 2025.