Kanegonda Ravi Chythanya
@sru.edu.in
Assistant Professor, School of Computer Science and Artificial Intelligence
SR University
RESEARCH INTERESTS
Wireless Sensor Networks
Bigdata Analytics
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Goli Sunil, Srinivas Aluvala, Chinthala Sujitha, Akarapu Mahesh, Areefa, Kanegonda Ravi Chythanya, and Gadde Aruna
AIP Publishing
Jhalak Agrawal, Kanwarpartap Singh Gill, Rahul Chauhan, Hemant Singh Pokhariya, and Kanegonda Ravi Chythanya
IEEE
By using a Deep Learning-Based Support Vector Machine (SVM) model, this work presents a novel method for the sorting and visualisation of raisins. A vital component of the agricultural sector, the processing of raisins, has historically relied on labour-intensive, human error-prone manual sorting techniques. The suggested approach makes use of deep learning to improve this process' accuracy and efficiency. The Raisins dataset in the form of two classes namely Besni and Kecimen which is extracted from the Kaggle open source environment or platform, is pre-processed using the SVM model and the model is further designed accordingly so that it may predict and accuracy of 92 %. The cultural history of raisins is extensive and rich, going all the way back to antiquity. They are utilized in many different culinary applications, such as savory meals and classic desserts. Vitamins, minerals, and dietary fibre are just a few of the essential components that are concentrated in raisins. They can be used in many dishes as a healthier alternative to processed sugars because of their natural sweetness. In the context of food processing, this work investigates the use of a deep learning-based Support Vector Machine (SVM) model for sorting and displaying raisins. The manufacturing of raisins entails complex quality control procedures that have historically relied on labour-intensive techniques. This study automates the categorization of raisins according to quality factors in an effort to improve sorting accuracy and efficiency by utilizing SVMs. Known for its resilience in classification tasks, the SVM model is used to identify intricate patterns in raisin samples, which enhances sorting accuracy. Furthermore, the research delves into the visual element, illuminating the structural features of raisins and offering valuable perspectives on quality criteria.
Taifa Ayoub Mir, Sheifali Gupta, Sonal Malhotra, Swati Devliyal, Deepak Banerjee, and Kanegonda Ravi Chythanya
IEEE
This model has been categorized as an electronic model that is an excellent tool for discriminating between various sorts of disorders as part of our study that focuses on diagnosing illnesses in apple leaves. Healthy leaves, Apple Scab, Powdery Mildew, Cedar Apple Rust, Apple Mosaic Virus, Apple Leaf Spot, Fire Blight, and frog-eye leaf Spot are some of the eight distinct disease categories that may be seen in Apple leaves. After getting special training, this model was able to spot some illnesses. These illnesses were put into the right groups by the computer, as shown by the findings. As an example, the model was able to tell the difference between healthy leaves 93.75% of the time. It got 92.02% of the real healthy leaves right, with an F1-Score of 92.88 percentage points. On the other hand, Apple Scab got an F1-Score of 93.54 percent, with a remembering rate of 94.41% and an accuracy rate of 92.68%. The model was always very good at all kinds of sickness, with accuracy and memory rates above 92%. The mean scores for our model always stuck around 93.41%, no matter if the scores were global, weighted, or micro. This showed that the model was stable and could find diseases well. Also, the model was very accurate overall (95%), which was a great finding. The fact that the model can tell the difference among several diseases that affect apple leaves shows that our method works. By making a tool that can correctly diagnose diseases in farming settings, the results of this study will help to improve how apple crops are cared for in terms of their health.
Deepak Banerjee, Neha Sharma, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
Growing sunflowers (Helianthus annuus) is very important in farming around the world. But it faces a big problem from downy mildew - mostly caused by Plasmopara halstedii. This pathogen brings damage to these beautiful flowers' growth and development process where millions all over earth suffer similar conditions just like them as well- known scientists. We need to stop this farming problem. It needs a smart way, so we can quickly find and get rid of sunflower leaf infections from something called downy mildew. This will help us manage the sickness better before it spreads more widely than necessary due to early detection efforts on farm lands known today as crop fields growing certain kinds such cotton etc., where usually grow potatoes only. To meet this need, our research begins a new adventure. It creates a mix of deep learning models that effortlessly uses both Convolutional Neural Networks (CNN) and AdaBoost to sort different types of downy mildew scales in complex ways. Our main plan is to pick out carefully 3300 big pictures that cover many different areas. We will use these images for study. These photos show the different kinds of downy mildew. They include many types and how it is in various places, which helps us understand more about a disease called Downy mildew. This big data set is not just for fun, but it gets painted on by our mix model. It sees small changes and hard things in photos of tiny bugs on plants that cause disease called downy mildew scales. It is been so accurate it could show correct results 98.16%. This numbers win is not only a big number standard; it shows how well the model really can understand and group soft white mildew spots into about four different levels. These are all bad situations. This careful view of badness levels is very important for farming workers. It lets them make their fixes according to how serious and fast-growing the bug problem might be. Our mix method always beats single CNN and AdaBoost models when comparing different models. Accuracy, remembering, and F1 score measures all show that joining deep learning's power to find features with AdaBoost's help for changing collections works together. This comparison shows our mix model does better. It also makes it clear we need combined ways of dealing with the complications of crops and their diseases problems.
Prerna Negi, V.K. Shrivastava, Shweta Pandey, Kanegonda Ravi Chythanya, Prafful Negi, and Manish Gupta
IEEE
Artificial Intelligence (AI) can speed up the progress, provide novel teaching and learning techniques, and address some of the largest issues in education. It combines cognition, machine learning, interaction of human and computer, storage of data, recognition of emotion, and decision making to make machines think and act like humans. Digitalization is a process that involves orientation, digital preparation, interactive presence, teamwork, follow-up and performance, reflection and motivation, and examination evaluation. Technology Enhanced Learning (TEL) is being developed to help students in their academic pursuits with AI-supported time and task planning (AI4TTP). AI can be used to influence people in a variety of ways, such as through chatbots, e-assessment, and anti-cheating systems. Learning analytics (LA) is a new area that aims to improve education and instruction by assessing raw information analytically and identifying trends to recognize behaviour of students, anticipate learner actions, and present quick feedback. Individualized instruction is tailored to each learner's individualized learning inclinations. There has been discussion of suggestions based on the analysis.
Varun Kumar, Deepak Banerjee, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
This study paper digs into the classification of marigold leaf disease, using cutting-edge machine-learning approaches to ensure accurate and prompt detection. Convolutional neural network (CNN) along with Random Forest methods were employed in our study to determine the precision, recall, and F1-Score of diseases that included Theodore Powdery Mildew, Downy's Mould, Botrytis Blight, and the Fungal Leaf Spot, Aster Oranges or Yellows Greens Rust, Bacterial Plant Spot, or Leaf Curl Virus. The precision values show the percentage of favorable predictions for each disease class, ranging from 95.02% to 95.68%. Recall data show that the model regularly captures true positives, with scores of more than 95.19% in all classes. The F1-Score, a proportional combination of recall and accuracy, demonstrates a balanced performing range of 95.28 to 95.70 percent. The support values demonstrate the dataset's layout for each disease class, as well as the average number of incidents. The research reported a total modeling accuracy of 98%, demonstrating the classification method's durability and reliability. The macro stages, weighed averages, or micro averages all contribute to the model's consistent and equitable performance across courses. These numerical results offer a complete understanding of the model's ability to precisely recognize and categorize marigold leaf diseases. The study offers crucial insights into agriculture pathology or precision agriculture, setting the framework for informed decisions about disease prevention and farming strategies. The findings provide novel techniques for maintaining the health and brightness of marigold petals within agricultural contexts.
Varun Kumar, Deepak Banerjee, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
As developments in image recognition and machine learning continue to alter the natural environment of plant pathology, this study investigates the classification of rosemary leaves according to various illnesses using deep neural networks (CNN) convolutions as well as Random Forest techniques. Dusty Mildew, Botrytis Blight, Root Decay, Leaf Spot Rosemary Beetle Aphid Infection, Rust, or Cercospora Leaf Spot is one of eight forms of rosemary disease of the leaves evaluated in this study. The classification model's efficacy has been thoroughly evaluated for each sickness class, including accuracy, recall, or F1-Score assessments. Precision ratings ranging from 93.88% to 94.62% demonstrate the system's ability to accurately recognize incidents within each sickness category. The accuracy values range from 93.92% to 94.59%, showing that the model is capable of recognizing genuine positive instances within each class. In addition, F1-Score results with a steady average of 94.26% support the model's harmonic balance of recall with accuracy. The support column shows the total amount of cases of every circumstance in the educational setting, which gives insight into the dataset's structure. This data is critical for estimating the prevalence of various ailments among the rosemary leaves. The model's resilience is supported by an overall precision of 98%, proving its ability to forecast accurately across all classes. The study employs macro, weighted, or micro average indicators, along with class-specific information, to conduct an in-depth assessment of the model's efficacy across all categories. The large-scale, weighted, and micro mean accuracy, memory, or total F1-Score values have stayed consistent at 94.26%, indicating that the model applies to others. This paper contributes to the field of agricultural pathology and also to the larger context of computerized crop illness recognition.
Varun Kumar, Deepak Banerjee, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
The present research extensively analyses the use of methods based on machine learning to classify symptoms of fenugreek leaves. this investigation focuses on the complicated topic of fenugreek sickness, using artificial neural networks like CNN's network and random forests to improve disease identification both in precision and efficacy. The Powdery mildew and downy mildew, anthracnose (which affected plants where it occurred), rust, bacterial plant spots (Fusarium Wilt), and botrytis blight were among the most frequent fenugreek leaf diseases studied. In addition, support metrics provide a breakdown of occurrence for each disease class, which is useful for measuring the prevalence or impact of various illnesses in fenugreek farming. The macro, balancing, and micro-average data give a summary of the algorithm's effectiveness. The integrated approach to classifying illnesses demonstrates security, with average outcomes closely matching specific class requirements. The simulation offers a remarkable mean precision of 93.84%, suggesting its ability to provide accurate estimates across the entire fenugreek disease of leaves range. Finally, this study improves the automated disease classification of fenugreek leaves, demonstrating the potential of AI approaches for precision agriculture. The findings given herein not only provide critical information for fenugreek cultivation but also lay the groundwork for future research into the interface of technology and sustainability in agriculture.
Nishant Pritam, Kanwarpartap Singh Gill, Mukesh Kumar, Ruchira Rawat, and Kanegonda Ravi Chythanya
IEEE
This research investigates the use of a K-Nearest Neighbours (KNN) model and further Machine Learning methodologies for doing Exploratory Data Analysis (EDA) on the quality of red wine. The dataset consists of a wide range of chemical and sensory characteristics that are linked to red wines. The study utilises rigorous data preparation techniques, such as addressing missing values and outliers, to prepare the dataset for analysis. The K-nearest neighbours (KNN) model, known for its simplicity and ability to capture local patterns, is used to reveal complex linkages within the dataset. This research assesses the efficacy of the KNN model and many other Machine Learning approaches by using appropriate measures, therefore elucidating their predictive capacities and ability to generalise to unfamiliar data. The purpose of exploratory investigations is to investigate and access the potential benefits of combining feature engineering and dimensionality reduction techniques in order to improve model interpretability and overall performance. The findings of this study provide significant contributions to our understanding of the intricate connections between chemical and sensory characteristics in red wines. These insights have practical consequences for professionals in viticulture, winemaking, and research within this domain. Additionally, this work highlights and employs the adaptability of Machine Learning in deciphering patterns within complex datasets, hence offering prospects for further investigation in the field of wine quality analysis.
Varun Kumar, Deepak Banerjee, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
Papaya production faces significant challenges as a result of numerous leaf diseases that impair yield and quality. This study focuses on the automated identification of prevalent papaya leaf diseases using a model for classification constructed using neural networks using convolution, often known as CNNs and Random Forest algorithms. The research includes an in- depth assessment of illness classification efficacy, with a focus on recall, precision, F1-score, support, and metrics for accuracy across a variety of groups, including PRSV, Dusty Mildew, bacterial Leaves Spot, which is Anthracnose and Black Spot, Cercospora The leaves Spot, as well as Angular Leaf Spot. The results show that the algorithm performs well, with precision values consistently exceeding 93% for each disease class. The high recall values of the model, which reach 92%, reflect its ability to detect true positive occurrences. The F1-score, which assesses the balance of recall and accuracy, frequently exceeds 94%, proving the persistence of the categorization model. The backing numbers offer data on the occurrence distribution of every illness in the dataset. The recommended model's effectiveness in differentiating between various papaya leaf ailments is evidenced by an average weighted precision of 94.49%. The micro, as well as macro averages, demonstrate the model's stability in performance across various courses, resulting in the micro average yielding a value of 94.49%. These findings help to develop automated assessment and treatment in papaya growing, providing farmers with a useful tool for making early and precise choices regarding management. The proposed methodology generates positive results, paving the path for better disease management and papaya yields.
Muskan Agarwal, Kanwarpartap Singh Gill, Rahul Chauhan, Hemant Singh Pokhariya, and Kanegonda Ravi Chythanya
IEEE
Plant diseases present a substantial threat to global food security, and the timely detection of such diseases remains a challenging and time-consuming task. The accurate determination of a plant's health status and the identification of specific infections typically require the expertise of professionals. The advent of Deep Learning has significantly transformed the field of computer vision, offering highly efficient techniques for image analysis and categorization. This study specifically focuses on utilizing the MobileNet50 Convolutional Neural Network (CNN) model to visually represent and categorize images of maize. The identification of maize, a widely cultivated crop, is intricate due to its diverse array of varieties and growth stages. This research aims to leverage the capabilities of the advanced MobileNet50 CNN architecture to enhance the precision and effectiveness of maize classification. The model achieved an impressive accuracy of 97%, demonstrating its robust performance in distinguishing various types and health states of maize plants. By employing MobileNet50, this study contributes to the advancement of computer vision applications in agriculture, facilitating the prompt and accurate identification of maize diseases. The utilization of a deep learning approach reduces the dependency on human expertise, making it more accessible and efficient for large-scale agricultural monitoring. Ultimately, the integration of MobileNet50 in maize classification holds promise for revolutionizing plant disease detection and contributing to global efforts in securing food resources.
Varun Kumar, Deepak Banerjee, Deepak Upadhyay, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
The abstract summarizes the key findings and performance indicators of the customized model that is intended for the categorization of illnesses affecting Gingerleaves. Utilizing criteria such as precision, recall, and F1-score, the model exhibited remarkable efficacy in discerning a range of ailments, such as Downy and Powdery mildew, Anthracnose, Angular Leaf Spot, GingerMosaic Virus, Bacterial Wilt, and Target spot. The precision numbers, which range between 95.85% to 98.04%, demonstrate the model's ability to precisely identify cases of each condition and demonstrate its accuracy in reducing false positives. At once, recall values from 96.01% to 97.74% underscore how good it is at spotting diseases and avoiding false negatives. Overall, the model performs a fantastic 96.79% in the classification of Gingerleaf diseases in a dataset that had 7583 images, while the combined F1-scores of precision and recall are balanced as they go from 96.14% to 97.42%→ Harmonized F1-scores, reflecting just the great the overall performance of the model. With a weighted average and macro-average of 96.80%, its stellar 99% Accuracy does nothing but help with the confidence that it will be just as consistent across many illness classes. The impressive recent outcomes illustrate how well the model worked at distinguishing between unhealthy and healthy Gingerleaves, which is crucial to making sure it's a valuable tool for precision disease management in agriculture. What's more, the abstract indicates just how useful and adaptable the model is as “the weighted and macro average for all disease scenarios” continues to perform well. Overall, however, it's clear that by giving the class of diseases that affect Gingerleaves a precise and reliable model for their categorization, this work will make a significant contribution to sustainable farming development as it says. This is another way of saying that this model has broken new ground to provide a state-of-the-art - and highly accurate and consistent-way to control diseases very quickly in agricultural settings.
Muskan Singla, Kanwarpartap Singh Gill, Deepak Upadhyay, Swati Devliyal, and Kanegonda Ravi Chythanya
IEEE
In the realm of evaluating financial risks, accurately predicting bankruptcy holds paramount importance for maintaining the stability of economic systems. A longstanding challenge in this field revolves around imbalanced datasets, where instances of bankruptcy are notably fewer than those of non-bankrupt cases. Mitigating this issue involves implementing a robust strategy that combines the Synthetic Minority Over-sampling Technique (SMOTE) with DenseNet121 CNN classification, resulting in heightened model accuracy. Financial data often exhibits an imbalance in class distribution, with occurrences of bankruptcy being significantly outnumbered by instances of solvency. This imbalance poses a substantial hurdle for machine learning algorithms, as they tend to display a bias toward the dominant class, leading to suboptimal accuracy in forecasting the minority class (bankruptcy). The primary objective of this study is to leverage DenseNet121 CNN for bankruptcy detection classification. This goal will be accomplished through the application of SMOTE Analysis, specifically designed to address the challenges posed by unbalanced data. The study aims to achieve an anticipated accuracy of 84 percent, employing the classification report and the confusion matrix as visualization tools in the proposed research. These metrics will contribute to a comprehensive understanding of the results and further enhance the reliability of the bankruptcy prediction model.
Khushi Mittal, Kanwarpartap Singh Gill, Mukesh Kumar, Ruchira Rawat, and Kanegonda Ravi Chythanya
IEEE
This research investigates the application of the ResNet50 Convolutional Neural Network (CNN) within a ResNet50 model framework for the purpose of classifying musical genres. The objective is to enhance the accuracy and efficiency of automated music genre categorization systems through the utilization of deep learning techniques. The proposed model employs a methodology that processes raw audio data, involving the extraction of relevant innovative features through convolutional layers. These layers are designed to capture hierarchical patterns inherent to specific genres. The incorporation of the ResNet50 architecture in machine learning facilitates the capture of temporal relationships, allowing the model to recognize subtle nuances and variations in musical compositions. The study utilizes a diverse dataset encompassing multiple genres to enhance the robustness and adaptability of the model. The primary goal is to validate the effectiveness of the CNN ResNet50 Model in accurately classifying musical genres. Through rigorous experimentation and assessment, this research aims to contribute significantly to the advancement of automated music analysis and classification systems. The findings of this study have noteworthy implications for various applications, including music recommendation systems, content tagging, and music streaming services.
Khushi Mittal, Kanwarpartap Singh Gill, Deepak Upadhyay, Vijay Singh, and Kanegonda Ravi Chythanya
IEEE
Using a Sequential Convolutional Neural Network (CNN) model, this research investigates the use of neural networks for the categorization of oil spills in satellite data. The increasing occurrence of oil spills presents a substantial danger to marine ecosystems, hence requiring expeditious and precise methodologies for detection. The suggested technique utilises the hierarchical feature learning capabilities of Convolutional Neural Networks (CNNs) to autonomously identify pertinent patterns from satellite photos. This enables the differentiation between oil spills and natural water characteristics. The use of a sequential design significantly improves the network's capacity to grasp spatial relationships and underlying patterns present in the picture. The article presents an evaluation of the efficacy of the constructed Convolutional Neural Network (CNN) model by conducting thorough testing on various datasets. The results highlight the model's strong performance in accurately categorising oil spills, achieving a high accuracy rate of 96 percent. The use of neural networks in this particular context presents a potentially effective strategy for the prompt identification and monitoring of environmental risks, hence enabling timely intervention and mitigation measures.
Muskan Singla, Kanwarpartap Singh Gill, Hemant Singh Pokhariya, Ramesh Singh Rawat, and Kanegonda Ravi Chythanya
IEEE
Within the domain of financial risk assessment, the accurate anticipation of bankruptcy has significant importance in safeguarding the stability of economic systems. One of the longstanding issues in this particular field pertains to unbalanced datasets, whereby the number of instances representing bankruptcy is much lower in comparison to non-bankrupt cases. The combination of Synthetic Minority Over-sampling Technique (SMOTE) with logistic regression classification presents a robust approach to tackle the issue of class imbalance, resulting in improved predicted accuracy of models. Financial statistics often demonstrate a disparity in class distribution, whereby occurrences of bankruptcy are comparatively few in comparison to instances of solvency. The existing disparity and social policy are a significant obstacle for machine learning algorithms, since these models tend to exhibit a bias towards the dominant class, leading to less-than-ideal accuracy when forecasting the minority class (bankruptcy). The main aim of this study is to access the Logistic regression classifier in order to classify Bankruptcy detection. The target will be achieved by the use of SMOTE Analysis, a technique designed to address the issue of unbalanced data. The anticipated accuracy of 84 percent will be employed using the classification report and the confusion matrix, both of which will be utilised in the proposed research study to visualise the results.
Muskan Agarwal, Kanwarpartap Singh Gill, Deepak Upadhyay, Sarishma Dangi, and Kanegonda Ravi Chythanya
IEEE
Cryptocurrency is a novel form of digital or virtual currency that employs cryptographic techniques to guarantee secure financial transactions, control the creation of new units, and verify the transfer of assets. Cryptocurrency signifies a fundamental change in the understanding of currency and financial transactions. The fundamental tenets of decentralisation, cryptographic security, and limited supply seek to revolutionise the conventional financial environment, providing fresh opportunities for financial inclusion, transparency, and innovation. Nevertheless, the path of cryptocurrencies is being influenced by ongoing challenges, such as regulatory uncertainties and market volatility, as they progressively establish themselves as a vital component of the global economy. It can be deduced that Dogecoin lacks the ability to supplant Bitcoin. Ethereum and Bitcoin exhibit a notably higher level of security compared to Dogecoin and Bit connect. That is the rationale behind their ability to withstand the decline in 2018 and also endure the current decrease in price. The depreciation of dogecoin is inevitable. Dogecoin is a valid and authentic form of digital currency. Nonetheless, the cultural structure of dogecoin ultimately undermines its own triumph.
Ricky Rajora, Himakshi Gupta, Mukesh Kumar, Ruchira Rawat, Kanegonda Ravi Chythanya, and Saransh Bansal
IEEE
Childhood leukemia, a formidable health challenge, demands innovative strategies for timely detection and intervention. This study leverages the formidable capabilities of cutting-edge deep learning models, VGG16 and EfficientNetB3, to intricately classify a comprehensive dataset comprising 15,135 cell images from 118 patients. Resultantly, VGG16 achieves a commendable classification accuracy of 77%, while the EfficientNetB3 model excels with an exceptional 91% accuracy. Beyond classification proficiency, this research underscores the urgency of early detection in childhood leukemia, shedding light on the transformative potential of deep learning models in enhancing diagnostic capabilities. The findings not only pave the way for refined classification methodologies but also illuminate promising avenues for timely, personalized, and targeted therapeutic interventions. This holistic approach holds promise for significantly improving outcomes and quality of life for young leukemia patients, emphasizing the indispensable role of technology in propelling advancements in pediatric oncology.
Rahul Chauhan, Tushar Sharma, Ruchira Rawat, Rudresh Pillai, and Kanegonda Ravi Chythanya
IEEE
This paper aims to present the implementation of two machine learning algorithms, Linear Regression and Lasso Regression for the task of predicting the price of the house located in the city of Bengaluru, India. The task of predicting house price accurately is quite difficult as it depends on a lot of factors, and needs of the buyer. For predicting these prices, we can try to create datasets which contains details of the factors that people look for while buying Houses such as size, location, bathroom etc. To analyses the data, we can usecertain algorithms such as linear regression, lasso regression etc. By using algorithms like this, we can reduce the margin of error of our model and try to make it more accurate and usable. Such models can be used by real estate agents,sellers as well as the people who want to buy house to get the best deal for them. In future,researcher can also integrate models like this with the real estate's websites like MagicBricks,99Acres to give better and more accurate recommendations to people.
Mayank Sharma, Rahul Chauhan, Swati Devliyal, and Kanegonda Ravi Chythanya
IEEE
This research paper aims to predict house prices with more efficiency and accuracy. The research paper involves three machine learning algorithms, "linear regression, lasso regression, and ridge regression," that are applied to the data set of Bangalore and have various information about different places in Bangalore, like BHK, number of bathrooms, area, etc. The house price prediction is quite a difficult task, as predicting the price of the house involves many factors like area, number of bathrooms, climate condition, age of the property, and many other factors. To make predictions according to all the factors, the author has applied three algorithms, one by one. From these machine learning algorithms, we have developed a model that helps predict the prices of houses with more accuracy and a lower error percentage. This model can be very useful for both buyers and sellers, as it can justify the price of the house for both of them.
Hassan M. Al-Jawahry, rd Kanegonda, Ravi Chythanya, th K.Maharajan and Akuthota Swathi
IEEE
The extensive application of Cyber-Physical Systems (CPS) requires efficient optimization of computational units and physical plants. Task scheduling (TS) is critical to improving resource utilization and system efficacy within CPS. Standard task schedulers in embedded real-time techniques fall short of achieving CPS performance criteria due to task diversity and system heterogeneity. This research proposes a novel task scheduling technique called Whale Optimization Algorithm based Energy-Efficient TS (WOA-EETS) that is specifically designed for the CPS context. WOA-EETSS involves assigning number of autonomous tasks to various resources during the TS process. The objective function focuses on reducing task completion time and optimizing resource use. A series of simulations were run to demonstrate WOA-EETSS' improved performance. The comparison analysis demonstrated that the WOA-EETSS system performed better across a variety of parameters. The proposed approach attains the better performance in Task Execution (ms) of 347ms, Convergence Rate of 511ms and Average Load Balancing Ratio (%) of 69.91% from 100 iterations.
Retinderdeep Singh, Neha Sharma, Priyanshi Aggarwal, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
Acute lymphoblastic leukaemia (ALL) is a group of blood malignancies that need very sophisticated detection methods and constitute a major danger to public health across the world. With the inceptionV3 model developed and implemented, this research study significantly advances blood cancer diagnostics. This model is built on top of a massive Blood Cells Cancer (ALL) dataset that includes a spectrum of cellular morphologies linked to ALL. This dataset and deep learning approaches were used to train the model being considered. A key component in enabling prompt and successful medical intervention, the major objective of this research activity is to enhance the precision and dependability of blood cancer diagnosis. Images of blood cells may reveal subtle aberrations and patterns, which may be recognised using a fine-tuned inceptionV3 model. The inceptionV3 model can use this to better differentiate between healthy and cancerous cells. The training procedure involves refining the model’s parameters to reach better degrees of sensitivity and specificity. After the training phase is over, the suggested inception model achieves an astonishing accuracy rate of 98.46%, demonstrating an excellent level of performance. The model demonstrates remarkable accuracy in diagnosing blood cancer with a high degree of precision, highlighting its potential as a crucial diagnostic tool in clinical settings.
Retinderdeep Singh, Neha Sharma, Priyanshi Aggarwal, Mukesh Singh, and Kanegonda Ravi Chythanya
IEEE
Blood cancers categorised as acute lymphoblastic leukaemia (ALL) pose a substantial threat to global public health and require extremely nuanced diagnostic techniques. This research study provides a substantial advancement in the field of blood cancer diagnosis through the development and implementation of an innovative Convolutional Neural Network model. On the basis of a large dataset of Blood Cells Cancer (ALL), which encompasses a variety of cellular morphologies associated with ALL, this model is constructed. The model under consideration is trained using deep learning methodologies and this dataset. The primary aim of this research endeavour is to improve the accuracy and reliability of blood cancer detection, a critical factor in facilitating timely and effective medical intervention. A fine-tuned CNN model is capable of identifying minute irregularities and patterns that may be detected in images of blood cells. This can assist the CNN model in distinguishing between benign and malignant cells accurately. To achieve higher degrees of specificity and sensitivity, the model’s parameters are refined during the training process. The proposed CNN model exhibits an exceptional degree of performance, achieving an astounding accuracy rate of 96.92% after the training phase concludes. Blood cancer is accurately diagnosed with a substantial degree of precision by the model, which is substantial evidence of its extraordinary accuracy and underscores its potential as a critical diagnostic tool in clinical settings.
Muskan Agarwal, Kanwarpartap Singh Gill, Mukesh Kumar, Ruchira Rawat, and Kanegonda Ravi Chythanya
IEEE
Accurately identifying product variations is critical for industrial operations due to the rising demand for automated quality control and effective inventory management. Convolutional Neural Networks (CNNs) offer the best option; they are famous for their exceptional performance on image recognition and classification jobs. This research explores the use of cutting-edge deep learning methods to the problem of Nespresso capsule image categorization using CNNs. In order to carry out the study, a dataset is assembled and produced using several photographs of Nespresso capsules. Care is taken to ensure that the images encompass a range of lighting conditions, perspectives, and backdrops. The EfficientNet design is notable for its exceptional balancing act between processing economy and accuracy, boasting a depiction accuracy of 94%. To improve generalizability and accelerate convergence, one may use transfer learning to start the model with pre-trained weights on a large dataset. Several criteria are employed to evaluate the model’s efficacy during training, such as loss, training and validation accuracy, and overall performance. Careful consideration is given to the computation of epochs in order to avoid overfitting and yet acquire complicated patterns from the training data. We also examine the model’s confusion matrix and classification report in detail after running it on a test dataset to determine its strengths and shortcomings. The goal of this project is to create a system that can correctly categorise Nespresso capsules in order to enhance inventory management and manufacturing quality control. This study’s results have wider implications for convolutional neural networks (CNNs) in picture classification tasks across several industries, demonstrating the effectiveness of deep learning methodologies in real-world contexts.
Andrzej Stateczny, Hirald Dwaraka Praveena, Ravikiran Hassan Krishnappa, Kanegonda Ravi Chythanya, and Beenarani Balakrishnan Babysarojam
MDPI AG
The increasing amount of rain produces a number of issues in Kerala, particularly in urban regions where the drainage system is frequently unable to handle a significant amount of water in such a short duration. Meanwhile, standard flood detection results are inaccurate for complex phenomena and cannot handle enormous quantities of data. In order to overcome those drawbacks and enhance the outcomes of conventional flood detection models, deep learning techniques are extensively used in flood control. Therefore, a novel deep hybrid model for flood prediction (DHMFP) with a combined Harris hawks shuffled shepherd optimization (CHHSSO)-based training algorithm is introduced for flood prediction. Initially, the input satellite image is preprocessed by the median filtering method. Then the preprocessed image is segmented using the cubic chaotic map weighted based k-means clustering algorithm. After that, based on the segmented image, features like difference vegetation index (DVI), normalized difference vegetation index (NDVI), modified transformed vegetation index (MTVI), green vegetation index (GVI), and soil adjusted vegetation index (SAVI) are extracted. The features are subjected to a hybrid model for predicting floods based on the extracted feature set. The hybrid model includes models like CNN (convolutional neural network) and deep ResNet classifiers. Also, to enhance the prediction performance, the CNN and deep ResNet models are fine-tuned by selecting the optimal weights by the combined Harris hawks shuffled shepherd optimization (CHHSSO) algorithm during the training process. This hybrid approach decreases the number of errors while improving the efficacy of deep neural networks with additional neural layers. From the result study, it clearly shows that the proposed work has obtained sensitivity (93.48%), specificity (98.29%), accuracy (94.98%), false negative rate (0.02%), and false positive rate (0.02%) on analysis. Furthermore, the proposed DHMFP–CHHSSO displays better performances in terms of sensitivity (0.932), specificity (0.977), accuracy (0.952), false negative rate (0.0858), and false positive rate (0.036), respectively.