Dr. Bhukya Madhu

@trr.org.in

Associate Professor-HOD-IT
TRR College of Technology



                    

https://researchid.co/madhu0525

RESEARCH, TEACHING, or OTHER INTERESTS

Computer Engineering, Artificial Intelligence, Agricultural and Biological Sciences, Biomedical Engineering

11

Scopus Publications

Scopus Publications


  • Moving object detection using modified GMM based background subtraction
    S. Rakesh, Nagaratna P. Hegde, M. Venu Gopalachari, D. Jayaram, Bhukya Madhu, Mohd Abdul Hameed, Ramdas Vankdothu, and L.K. Suresh Kumar
    Elsevier BV

  • A model for multi-attack classification to improve intrusion detection performance using deep learning approaches
    Arun Kumar Silivery, Ram Mohan Rao Kovvur, Ramana Solleti, LK Suresh Kumar, and Bhukya Madhu
    Elsevier BV

  • Network Intrusion Detection using ML Techniques for Sustainable Information System
    K. Chandra Mouli, B. Indupriya, D. Ushasree, Ch.V. Raghavendran, Babita Rawat, and Bhukya Madhu
    EDP Sciences
    Network intrusion detection is a vital element of cybersecurity, focusing on identification of malicious activities within computer networks. With the increasing complexity of cyber-attacks and the vast volume of network data being spawned, traditional intrusion detection methods are becoming less effective. In response, machine learning has emerged as a promising solution to enhance the accuracy and efficiency of intrusion detection. This abstract provides an overview of proper utilization of machine learning techniques in intrusion detection and its associated benefits. The base paper explores various machine learning algorithms employed for intrusion detection and evaluates their performance. Findings indicate that machine learning algorithms exhibit a significant improvement in intrusion detection accuracy compared to traditional methods, achieving an accuracy rate of approximately 90 percent. It is worth noting that the previous work experienced computational challenges due to the time-consuming nature of the utilized algorithm when processing datasets. In this paper, we propose the exertion of more efficient algorithms to compute datasets, resulting in reduced processing time and increased precision compared to other algorithms to provide sustainability. This approach proves particularly when computational resources are limited or when the relationship between features and target variables is relatively straightforward.

  • IOT NETWORK ATTACK SEVERITY CLASSIFICATION
    Bhukya Madhu, Sanjib Kumar Nayak, Veerender Aerranagula, E. Srinath, Mamidi Kiran Kumar, and Jitendra Kumar Gupta
    EDP Sciences
    Lack of network security is a major roadblock for Internet of Things (IoT) implementations. New attacks have emerged in recent times, taking advantage of vulnerabilities in IoT gadgets. The sheer scale of the IoT will also make standard network attacks more potent. Machine learning has found a lot of use in traffic classification and intrusion detection. We present a methodology in this piece that can be used to spot fraudulent communications and determine the identity of IoT devices. To determine the origin of the generated traffic, the nature of the traffic, and the presence of network hazards, this framework collects features per network flow. To achieve this, it relocates the network’s brains to its periphery. In order to discover which of several Machine Learning algorithms is superior to random forest, a number of them are pitted against one another. Using these Machine Learning methods, attacks can be ranked in terms of their potential damage. After running the tests, it was determined that TABNET has the highest accuracy (94.62%) for categorizing the network severity (93.51%) and that CNN has the lowest accuracy (93.51%) of the two.

  • Incorporating Sustainability: A Comprehensive Review of Factors Influencing Consumer Acceptance of Mobile Wallets
    Himanshi Bhardwaj, Pooja Kapoor, Avnish Kumar, N.V. Ganapathi, and Bhukya Madhu
    EDP Sciences
    Mobile wallets have gained widespread popularity as a convenient, secure, and user-friendly payment method embraced by consumers. However, the pace of mobile wallet adoption has exhibited variations across different markets, primarily due to a range of factors. In this study, we present an all-encompassing examination of existing literature, aiming to pinpoint the fundamental elements influencing the sustainable acceptance of mobile wallets by consumers. Through an exhaustive analysis of 80 research papers published between 2010 and 2022, we discern the prevailing factors that hold sway over the adoption of mobile wallets. Our scrutinization highlights that factor such as perceived sustainability, usefulness, ease of use, security, social influence, trustworthiness, and compatibility stand out as the most formidable propellants of mobile wallet adoption. Furthermore, our investigation uncovers hurdles that hinder the wider acceptance of mobile wallets, encompassing insufficient awareness, perceived intricacies, and lingering uncertainties regarding the technology’s sustainability. Our in-depth evaluation underscores the necessity of comprehending consumer perspectives and dispositions towards mobile wallets to galvanize their adoption sustainably. The culmination of our inquiry involves a discourse on the implications drawn from our discoveries, catering to researchers and practitioners vested in fostering the sustainable adoption of mobile wallets.

  • Techniques of Machine Learning for the Purpose of Predicting Diabetes Risk in PIMA Indians
    Bhukya Madhu, Veerender Aerranagula, Riyaz Mahomad, V. Ravindernaik, K. Madhavi, and Gopal Krishna
    EDP Sciences
    Chronic Metabolic Syndrome Diabetes is often called a “silent killer” due to how little symptoms appear early on. High blood sugar occurs in people with diabetes because their bodies have a hard time maintaining normal glucose levels. Care for a recurrent sickness would be permanent. The two most common forms of diabetes are type 1 and type 2. A better prognosis can help reduce the high risk of developing diabetes. In order to better predict the likelihood that a PIMA Indian may develop diabetes, this study will use a machine learning-based algorithm. The demographic and health records of 768 PIMA Indians were used in the analysis. Standardisation, feature selection, missing value filling, and outlier rejection were all parts of the data preparation process. Machine learning techniques such as logistic regression, decision trees, random forests, the KNN model, the AdaBoost classifier, the Naive Bayes model, and the XGBoost model were used in the study. Accuracy, precision, recall, and F1 score were the only metrics utilised to assess the models' efficacy. The results demonstrate that. The results of this study reveal that diabetes risk may be reliably predicted using machine learning-based models, which has important implications for the early detection and prevention of this illness among PIMA Indians.

  • Modelling the Impact of Road Dust on Air Pollution: A Sustainable System Dynamics Approach
    Sarah Khan, Quamrul Hassan, Kaushal Kumar, Saurav Dixit, Kshama Sharma, Vivek Kumar C., Navdeep Dhaliwal, and Bhukya Madhu
    EDP Sciences
    Road dust contributes significantly to air pollution by releasing fine particulate matter (PM) and other pollutants into the air, which can cause respiratory and cardiovascular problems and premature death. This dust is generated through the wear and tear of vehicle tires and road surfaces, as well as the accumulation of dirt and debris on the road, primarily from construction activities and cargo trucks carrying building materials. Wind, weather conditions, and vehicle movement play crucial roles in the distribution and concentration of these particles in the air. To address this issue, this paper focuses on identifying various variables that are connected to road dust operations and their interrelationships with air pollution variables, representing the dynamic pattern of the entire system. The paper proposes the establishment of a sustainable causal-loop model using system dynamics (SD) modeling in Vensim, connecting feedback mechanisms to effectively control the road dust concentration. Additionally, the paper suggests different policy interventions applied to the whole system to achieve optimized results. In the future, this research aims to convert and simulate the causal-loop model to a stock-flow model and compare the effectiveness of different policy interventions to further reduce road dust contributing to air pollution.

  • Automated Brain Tumour Classification using Deep Learning Technique
    M. Kiran Kumar, D. Sree Naga Sreeja, Samiya Sadiq, D. Manisha, Abhishek Jain, and Bhukya Madhu
    EDP Sciences
    Brain Tumour is a severe condition caused due to abnormal growth of cells in the brain. Brain Tumour is broadly classified into two categories namely Malignant (Cancerous) and Benign (Non-Cancerous). As tumour grows, the pressure within the skull can increase which can damage the brain and be life-threatening. Early detection and classification of the brain tumours is important as it helps to select the most appropriate treatment for saving the patient’s life. Usually, Brain Tumour Detection can be done manually by the doctors or use machine learning models in case of MRI images of the brain. In literature, it is identified that deep learning techniques such as CNN, DCNN and RNN show good results in image processing applications. This paper aims to detect and classify the Brain Tumours effectively using CNN deep learning techniques. The dataset is collected from Kaggle. The proposed method achieved an accuracy of 93.5% and 98.4% with CNN and Resnet50 respectively.

  • Intrusion detection models for IOT networks via deep learning approaches
    Bhukya Madhu, M. Venu Gopala Chari, Ramdas Vankdothu, Arun Kumar Silivery, and Veerender Aerranagula
    Elsevier BV

  • IoT based Intelligent Attendance Monitoring with Face Recognition Scheme
    S A Sivakumar, Tegil J John, G Thamarai Selvi, Bhukya Madhu, C Udhaya Shankar, and K P Arjun
    IEEE
    This article includes understudy participation and workforce participation. The understudy participation is set apart by face acknowledgment. Face identification and face acknowledgment are performed by the raspberry-pi module. The pin camera is associated with the raspberry-pi serial USB port catch of the researchers who are accessible inside the class for face location. The selected images perceive with stored images and will perceive the essences of each understudy, and enrollment will be given to that subject class based on that perception. This interaction is done in each class and understudies are given participation appropriately. The attendance will be set apart with date and time. The participants can check whether there is human intervention or not.

RECENT SCHOLAR PUBLICATIONS