@sr university
Professor & Head, Computer Science and Artificial Intelligence
SR University
Computer Engineering, Multidisciplinary
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Sohamkumar Chauhan, Damoder Reddy Edla, Vijayasree Boddu, M Jayanthi Rao, Ramalingaswamy Cheruku, Soumya Ranjan Nayak, Sheshikala Martha, Kamppa Lavanya, and Tsedenya Debebe Nigat
Springer Science and Business Media LLC
AbstractDeep learning is a highly significant technology in clinical treatment and diagnostics nowadays. Convolutional Neural Network (CNN) is a new idea in deep learning that is being used in the area of computer vision. The COVID-19 detection is the subject of our medical study. Researchers attempted to increase the detection accuracy but at the cost of high model complexity. In this paper, we desire to achieve better accuracy with little training space and time so that this model easily deployed in edge devices. In this paper, a new CNN design is proposed that has three stages: pre-processing, which removes the black padding on the side initially; convolution, which employs filter banks; and feature extraction, which makes use of deep convolutional layers with skip connections. In order to train the model, chest X-ray images are partitioned into three sets: learning(0.7), validation(0.1), and testing(0.2). The models are then evaluated using the test and training data. The LMNet, CoroNet, CVDNet, and Deep GRU-CNN models are the other four models used in the same experiment. The propose model achieved 99.47% & 98.91% accuracy on training and testing respectively. Additionally, it achieved 97.54%, 98.19%, 99.49%, and 97.86% scores for precision, recall, specificity, and f1-score respectively. The proposed model obtained nearly equivalent accuracy and other similar metrics when compared with other models but greatly reduced the model complexity. Moreover, it is found that proposed model is less prone to over fitting as compared to other models.
Shivani Gaba, Ishan Budhiraja, Vimal Kumar, Sheshikala Martha, Jebreel Khurmi, Akansha Singh, Krishna Kant Singh, S. S. Askar, and Mohamed Abouhawwash
Institute of Electrical and Electronics Engineers (IEEE)
In this current era, cyber-physical systems (CPSs) have gained concentrated consideration in various fields because of their emergent applications. Though the robust dependence on communication networks creates cyber-physical systems susceptible to deliberated cyber related attacks and detecting these cyber-attacks are the most challenging task. There is the interaction among the components of the cyber and physical worlds, so CPS security needs a distinct approach from past security concerns. Deep learning (DL) distributes better performance than machine learning (ML) due to its layered architecture and the efficient algorithm for extracting prominent information from training data. So, the deep learning models are taken into consideration quickly for detecting cyber-attacks in cyber physical systems. As numerous attack detection methods have been proposed by various authors for enforcing CPS security, this paper reviews and analyzes multiple ways of attack detection presented for CPS using deep learning. We will be putting the excellent potential for detecting cyber-attacks for CPS concerning deep learning modules. The admirable performance is attained partly as highly quality datasets are eagerly obtainable for the use of the public. Moreover, various challenges and research inclinations are also discussed in impending research.
Mothe Rajesh and Sheshikala Martha
Springer Science and Business Media LLC
Shivam Chaudhary, Rajat Chaudhary, Ishan Budhiraja, Aditya Bhardwaj, Anushka Nehra, and Sheshikala Martha
IEEE
Unmanned aerial vehicles (UAVs), which can help with high-speed communications and provide better coverage, are an important component of next-generation wireless networks. Because of its high mobility and aerial nature, it is suitable for a wide range of mobile wireless communications-based applications. However, low data rates with limited transmission power constitute a significant difficulty in wireless communication that lowers network performance. To overcome this issue, integrating a UAV with a relay device capable of delivering high data speeds while utilising minimum transmission power is a promising approach. In this research, we presented an edge-cutting framework called UAV-IRS, in which an Intelligent reflective surface (IRS) supports unmanned aerial vehicles (UAVs) that traverse areas with low signal strength. Furthermore, we discussed the applications, challenges and research directions of UAV-IRS in vehicle-to-everything (V2X) communication. We considered a case study of UAV-IRS in V2X communication. The performance evaluation demonstrates how the viable data rate and minimum transmission power decrease with distance as the number of IRS elements increases.
Balakrishna Bhukya, M. Sheshikala, and Balakrishna Bhukya
IEEE
Topic modeling is a popular unsupervised machine learning technique that can assist healthcare providers and policymakers in analyzing vast amounts of unstructured text data, such as patient reviews, to improve the quality of care and access to health care services. This research paper investigates the potential of topic modeling in health care by examining patient reviews collected from online healthcare platforms. The study employs state-of-the-art topic modeling algorithms, including Latent Dirichlet Allocation (LDA), to identify the most relevant topics related to quality of care and access to services. The results reveal that topic modeling can provide valuable insights into patient experiences and perceptions, and help pinpoint areas for improvement in health care delivery. By leveraging these insights, healthcare providers and policymakers can improve patient outcomes, optimize the patient experience, and tackle systemic issues in healthcare delivery. This research paper emphasizes the utility of topic modeling as a powerful tool for healthcare improvement, underscoring the importance of utilizing patient reviews as an essential source of valuable data in health care analytics.
Sheshikala Martha, Rithika Reddy Patti, Varsha Sri Paka, Saketh Pashikanti, and Saikeerthana Shanigarapu
IEEE
In the past few years, Human Action Detection has become more popular due to its applications in various fields. In this scenario, if a model can be created to detect human action then it will be very much useful to the application in Human Action Detection. Training a neural network from scratch may not be a feasible option. The main challenge to develop a neural networking model form scratch is the complexity of the model. The results of each model can be analyzed efficiently with Transfer Learning approach and also the results for the transfer learning models are much more promising than building the model from scratch when there is less amount of data. A comparative study is done to determine the best architecture for a transfer learning model.
Mothe Rajesh and M. Sheshikala
Springer Nature Singapore
M. Sheshikala, Sallauddin Mohmmad, D. Kothandaraman, Dadi Ramesh, and Ranganath Kanakam
Springer Nature Singapore
D. Kothandaraman, A. Balasundaram, E. Sudarshan, M. Sheshikala, and B. Vijaykumar
AIP Publishing
P. Kumaraswamy, M. Shashikala, Mamidala Sruthi, B. Vijay Kumar, and G. Nagaraju
AIP Publishing
Mothe Rajesh, Biswajit Senapati, Ranjita Das, and Sheshikala Martha
Elsevier BV
M. Sheshikala, Dadi Ramesh, Sallauddin Mohmmad, and Syed Nawaz Pasha
Springer Singapore
M. Sheshikala, P. Praveen, and B. Swathi
Springer Nature Singapore
Bonthala Prabhanjan Yadav, M. Sheshikala, N Swathi, Kanegonda Ravi Chythanya, and E Sudarshan
IOP Publishing
Mahesh Akarapu, Sheshikala Martha, Koteshwar Rao Donthamala, B Prashanth, G. Sunil, and K. Mahender
IOP Publishing
M Sridevi, N ManikyaArun, M Sheshikala, and E Sudarshan
IOP Publishing
Sheshikala Martha, SyedMusthak Ahmed, G. Jose Mary, Bonthala Prabhanjan Yadav, S. Gouthami, and M. Sridevi
IOP Publishing
Abstract Visual and Hearing impairments are the one of the major cause for accidents. In this paper we discuss the major impacts of Visual and Hearing Impact while driving and the remedies that can be used to provide the solution to the Problem. Impairment can be defined as the abnormality or mis-functionality of a physical body or a system. We discuss the types of disease that are affecting both vision and hearing impairment and the types of exercises that are required to overcome these diseases empowering Cognitive Abilities. Yoga plays a major in improving the strength of Visual and Hearing impairments, so we discuss the exercises that are required to improve the impairments.
Syed Musthak Ahmed, M. Sheshikala, Ankit Maurya, and Vinit Kumar Gunjan
Springer Singapore
Dr.M. Sheshikala, , Dr.D. Kothandaraman, Dr.R.Vijaya Prakash, G. Roopa, , , and
Blue Eyes Intelligence Engineering and Sciences Engineering and Sciences Publication - BEIESP
Detecting the author of the sentence in a collective document can be done by choosing a suitable set of features and implementing using Natural Language Processing in Machine Learning. Training our machine is the basic idea to identify the author name of a specific sentence. This can be done by using 8 different NLP steps like applying stemming algorithm, finding stop-list words, preprocessing the data, and then applying it to a machine learning classifier-Support vector machine (SVM) which classify the dataset into a number of classes specifying the author of the sentence and defines the name of author for each and every sentence with an accuracy of 82%.This paper helps the readers who are interested in knowing the names of the authors who have written some specific words.
Internet of Things (IoT) is a fast- growing technology in on-going research field that includes wireless sensor networks, cloud computing, big data analytics, ubiquitous computing, distributed decentralized systems, pervasive computing, embedded systems, mobile computing, machine learning etc. The above mentioned fields are mainly connected with IoT smart portable devices such as smartphones, home appliances, healthcare device, smart vehicle devices automation industry devices, etc. Though IoT enabled devices has been increased in many fields, the industries still faces many problem with connectivity issues because of several factors like mobility nature of devices; limited processing power and resource availability which includes energy, bandwidth constraints, routing cost and end to end delay; communication between node to node via intermediate mobile nodes towards destination may also fail links frequently, there by affecting the network performance. These limitations of existing topology based on reactive tree and mesh based routing protocols create challenging task while designing an optimized stable routing algorithm for IoT. In such a situation, resource optimization is an essential task to be performed by the IoT networks. In the proposed work resource optimization was done by Designed Optimized Multicast Routing Algorithm (DOMRA) for IoT. The DOMR algorithm implemented has route discovery process with nodes positions, directions of nodes, velocities of nodes, and then the path stability bases to overcome the connectivity issues. The proposed algorithm focusing to deploy various real time IoT enabled applications such as smart home automation, smart cites, smart agriculture, automation industry etc. To finalize the simulation results shows maximized system throughput, goodput, packet delivery ratio, network lifetime, network routing performance and reduced control overheads. The proposed algorithm hence produced better routing performance when compared with other existing algorithm in wireless networks.
R. Vijaya Prakash, S. S. V. N. Sarma, and M. Sheshikala
Institute of Advanced Engineering and Science
Association Rule mining plays an important role in the discovery of knowledge and information. Association Rule mining discovers huge number of rules for any dataset for different support and confidence values, among this many of them are redundant, especially in the case of multi-level datasets. Mining non-redundant Association Rules in multi-level dataset is a big concern in field of Data mining. In this paper, we present a definition for redundancy and a concise representation called Reliable Exact basis for representing non-redundant Association Rules from multi-level datasets. The given non-redundant Association Rules are loss less representation for any datasets.
M. Sheshikala, D. Rajeswara Rao, and R. Vijaya Prakash
IEEE
Given an application of a spatial data set, we discover a set of co-location patterns using a GUI (Graphical User Interface) model in a less amount of time, as this application is implemented using a parallel approach-A Map-Reduce framework. This framework uses a grid based approach to find the neighboring paths using a Euclidean distance. The framework also uses a dynamic algorithm in finding the spatial objects and discovers co-location rules from them. Once co-location rules are identified, we give the input as a threshold value which is used to form clusters of similar behavior. If the threshold value is too low more clusters are formed, if it is too high less clusters are formed. The comparison of the results shows that the proposed system is computationally good and gives the co-location patterns in a less amount of time.
M. Sheshikala, R. Vijaya Prakash, and D. Rajeswara Rao
IEEE
Privacy preserving of personal data is done by many techniques like k-anonymity, l-diversity, t-closeness etc., but the techniques proposed are implemented only when there is the availability of a laptop or a computer. Now-a-days people are interested to carry a mobile instead of a lab-top, because some of the works done by a lap-top can also be done by a mobile, like files sharing, images and videos sharing, and many more, but the sharing may lose the privacy of the data. With a specific end goal to give the protection to the information the strategy k-anonymity is utilized, which chooses the k-esteem where at any rate k-1 people whose data additionally shows up in the discharge. This technique is implemented using Android SDK. Whenever the user requests the information, instead of sending original information, the data is sent in an anonymized way. This paper presents the implementation of this technique and results are shown.