Shekhar Rambhau Suralkar

Scopus Publications

Performance Evaluation of Light Weight Cryptographic CLEFIA Algorithm
Atul H. Karode, Shekhar R. Suralkar, and Vaishali B. Patil
Seventh Sense Research Group Journals

Prediction of Epilepsy Seizures by Machine Learning Methods
Mayuri Tushar Deshmukh and Shekhar R. Suralkar
Auricle Technologies, Pvt., Ltd.
According to the Globe Health Organization (WHO), more than 50 million people throughout the world are living with a diagnosis of epilepsy, making it perhaps of the most widely recognized neurological issue. Epileptic seizures are a leading cause of hospitalization and mortality across the globe. Accurate and prompt diagnosis is more crucial than ever given the increase in epileptic seizures all through the globe and their effect on individuals' lives. Epilepsy, cancer, diabetes, heart disease, thyroid disease, and many more are only some of the diseases for which machine learning approaches are being applied in prediction and diagnosis. Epilepsy is one ailment that may be treated early on to save a person's life. The main objective of this research is to use feature label extraction to the dataset in order to obtain the best ML models for epileptic seizures. In order to predict epilepsy, we used the techniques of logistic regression, SVM, linear SVM, KNN, and RNN in this study. The models employed in this research are accurate to varying degrees and have attributes including precision, recall, f1-score, and support. This study demonstrates that the model is able to accurately predict the occurrence of epilepsy. Our discoveries demonstrate that involving Examination highlight extraction in the dataset, the Regional Neural Network (RNN) model with 99.9998 % Training data accuracy and 97.78% Test data accuracy and 100% prediction probability of epilepsy seizure produces the best results and also the feature characteristics of RNN is better as compared to other models used in current research work.

Machine Learning Classifiers to Detecting Epileptic Seizures
Mayuri Tushar Deshmukh and Shekhar R. Suralkar
IEEE
Examining the neuronal signals that brain neurons produce may help detect the serious chronic neurological disease known as epilepsy. Neurons are tightly connected to one another in order to transmit signals and communicate with internal organs. To identify these brain impulses, electroencephalogram (EEG) in addition electrocorticography (ECoG) media are usually employed. The signals are intricate, noisy, not stationary, and not linear, also they a plenty of data to create. As a consequence, finding knowledge about the brain and identifying seizures are tough tasks. Classifiers for machine learning can categorize EEG data, identify seizures, also showcase significant patterns without losing performance. As a consequence, several academics have developed a range of seizure detection techniques that make use of statistical traits then machine learning classifiers. The selection of appropriate classifiers and attributes presents the most challenges. This paper's goal is to provide a summary for the many different variations of these approaches over the past several years depending on the classification of statistical characteristics also “black-box” also “non-black-box” classifiers for machine learning. Cutting-edge techniques and concepts discussed will provide a meticulous clutch of seizure detection, enumeration, and future research prospects.

Deep Learning for Crowd Image Classification for Images Captured Under Varying Climatic and Lighting Condition
Hemant T. Ingale, Shekhar S. Suralkar, and Anil J. Patil
IEEE
Most of recent events have attracted a lot of attention towards importance of automatic crowd classification and management. COVID-19 is the most setback for the entire world. During these events proper breakout and public crowd management leads to the requirement of managing, counting, securing as well as tracking the crowd. But automatic analysis of the crowd is very challenging task because of varying climatic and lighting conditions, varying postures etc. During this paper we have developed PYTHON based system for automatic crowd images classification using Deep learning. This paper is the first attempt for automatic classification of crowd images. We have prepared the dataset of crowd classification consisting of three categories. The proposed methodology of crowd classification starts with preprocessing during which we have used median filtering for noise removal. Deep learning models are developed using 70% training images. The performance of the system is evaluated for various deep learning algorithms including one block VGG, two block VGG and three block VGG. We have also evaluated the performance of three block VGG using dropout. VGG16 transfer learning based crowd classification is developed using PYTHON. Using VGG16 transfer learning we achieved the accuracy of 69.44.% which is highest among all deep learning classification models during this study

Performance Monitoring of High Voltage Power Supply using Flyback Converter with Extra High Tension

Evaluating the Performance of POX and RYU SDN Controllers Using Mininet
Nafees M. Kazi, Shekhar R. Suralkar, and Umesh S. Bhadade
Springer International Publishing

Image encryption based on matrix factorization
Vivek Khalane, Shekhar Suralkar, and Umesh Bhadade
International Information and Engineering Technology Association
In this paper, we present a matrix decomposition-based approach for image cryptography. The proposed method consists of decomposing the image into different component and scrambling the components to form the image encryption technique. We use two different type of matrix decomposition techniques to check the efficiency of proposed encryption method. The decomposition techniques used are Independent component analysis (ICA) and Non-Negative Matrix factorization (NMF). The proposed technique has unique user defined parameters (key) such as decomposition method, number of decomposition components and order in which the components are arranged. The unique encryption technique is designed on the basis of these key parameters. The original image can be reconstructed at the decryption end only if the selected parameters are known to the user. The design examples for both decomposition approaches are presented for illustration purpose. We analyze the complexity and encryption time of cryptography system. Results prove that the proposed scheme is more secure as it has less correlation between the input image and the encrypted version of the same as compared to state-of-art methods. The computation time of the proposed approach is found to be comparable.

Implementation of real time digital watermarking system for video authentication using FPGA
Ashish S. Bhaisare, A. H. Karode, and S. R. Suralkar
IEEE
This paper presents a hardware implementation of a digital watermarking system that can insert invisible, semi fragile watermark information into compressed video streams in real time. The watermark embedding is processed in the discrete cosine transform domain. Hardware implementation using field programmable gate array has been done, and project was carried out using a custom versatile breadboard for overall performance evaluation which become a viable target for the implementation of real time algorithms suited to video image processing applications. This System is developed on Xilinx Spartan3 Field Programmable Gate Array (FPGA) device using embedded development kit (EDK) tools from Xilinx. The results showed that the proposed algorithm has a very good hidden invisibility, good security and robustness for a lot of hidden attacks.

A compact hexagonal shaped patch antenna for UWB applications using CPW feed
Y.S. Santawani and S.R. Suralkar
IEEE
In this paper, the hexagonal shaped patch antenna feed by Co-Planar Waveguide (CPW) has been designed and analyzed using method of moments based electromagnetic simulator IE3D.The antenna is fabricated using a substrate having a dielectric constant of 4.4 (relative permittivity=4.4) and thickness of 1.6mm. The simulated results such as return loss, efficiency, Voltage Standing Wave Ratio (VSWR)and Gain are investigated. The antenna having a return loss less than -10dB from 2.9GHz to 12GHz which is applicable to Ultra Wide Band (UWB) Application. The point of attraction of this antenna is the use of single patch which makes it easy to fabricate.

Fingerprint verification based on fixed length square finger code
P.M. Patil, S.R. Suralkar, and H.K. Abhyankar
IEEE
In this paper a novel method is presented for generation of a fixed length square finger code of size 16times16times4. It uses a set of Gabor filters for extracting fingerprint features from gray scale image cropped in the size of 128times128 pixels using its core point as the center. Experimental results show that the recognition rate based on the Euclidean distance between the two corresponding Gabor filter finger codes have a verification accuracy of 93%. Since, the fingerprint matching is based on the Euclidean distance between two corresponding finger codes, it is extremely fast. This reveals that by setting the parameters to appropriate values, the finger code generated is more efficient and suitable than conventional methods for a small-scale fingerprint recognition system

Rotation invariant thinning algorithm to detect ridge bifurcations for fingerprint identification
P.M. Patil, S.R. Suralkar, and F.B. Sheikh
IEEE
In this paper we have modified the thinning algorithm proposed by Ahmed and Ward (2002). The unique feature that distinguishes the algorithm is its ability to thin any symbol or fingerprint to its central line taking care that the shape of the symbol is preserved while being rotation invariant. Our modified algorithm also incorporates a process to thin zigzag diagonal lines having a width of two pixels which was not considered in "a rotation invariant rule-based thinning algorithm for character recognition" (Ahmed and Ward, 2002). The algorithm is iterative and makes use of parallel processing to speed up execution. The system has 21 rules in its inference engine which are applied simultaneously to every pixel in each iteration. The algorithm is implemented for thinning fingerprints, fonts and symbols to a single pixel width. We also introduce a 24 rule based mask for detection of ridge bifurcations, which can be helpful for recognition/authentication of fingerprints as a biometric

Shekhar Rambhau Suralkar

RESEARCH, TEACHING, or OTHER INTERESTS

Scopus Publications