Proceedings of 6th International Conference on Recent Trends in Machine Learning, IoT, Smart Cities and Applications

Inhaltsverzeichnis

1. Frontmatter

2. Assessment of Improvising Breast Cancer Detection Through Machine Learning and Cross-Validation Methods

   G. Srikanth, R. Saikrishna, M. Ravindran, Y. Aruna Suhasini Devi, S. Krishnaveni, B. Sivaiah

   Abstract

   Major health problem affecting millions of women worldwide is breast cancer. Enhancing patient outcomes and lowering death rates are directly related to early identification and precise prognosis of breast cancer. So as to improve breast cancer prediction, this research compares machine learning algorithms as well as cross-validation methods. The study makes use of an extensive dataset that includes the clinical characteristics of patients with breast cancer. To create prediction models, a assortment of machine learning methods are used, like decision trees, KNN, random forests, logistic regression, and support vector machines. In addition, an assortment of cross-validation procedures are worn to evaluate model presentation, including leave-p-out cross-validating, stratified k-fold cross-validation, and k-fold cross-validation. The goal of the comparison analysis is to calculate about a mixture of cross-validation methods and machine learning algorithms predict breast cancer. Performance indicators are used to consider every model's predictive power, including recall, accuracy, precision, and f1 score.

3. A Specific Processing of Device to Device Communication Through Multihop for Equalizing and Minimizing Sar

   Vivekanand Aelgani, T. Bhaskar, S. Kirubakaran, Voruganti Naresh Kumar, Yedlla Satyam, Bejjanki Pooja

   Abstract

   The fifth generation wireless network, termed as 5G, is the most highly developed wireless telecommunication standard that has replaced all previous ones—1G, 2G, 3G, and 4G. The innovation brings a revolutionary network architecture to enable the connectivity of a multitude of entities across electronic devices, consumer goods, and other digital systems. Despite what it can offer, 5G comes with a high Specific Absorption Rate (SAR), which sparks concerns over electromagnetic field (EMF) exposure. To answer this challenge, we introduce a new and innovative solution. The current project outlines an advanced method of band identification that uses best data rates in multi-hop packet routing, aimed at minimizing and dispersing the transmission power. This method lessens the SAR effects due to EMF exposure. Theoretical analysis sets an upper limit on the number of hops allowed in a particular spectral band with linear constraints. Simulation results on randomly created network topologies show that the method proposed here provides better SAR performance compared to standard direct communication with base stations.

4. Design of Modified Booth Multiplier Using Reversible Logic Gates

   A. Pradeep Kumar, B. Vamsi Krishna, R. Bhargav Ram, G. Divya, Suraya Mubeen, Attuluri Uday Kiran

   Abstract

   The Reversible Multiplier is a computational apparatus employed for the purpose of multiplying two binary values through the utilization of reversible addressing. The significance of the development of a reversible multiplier arises from its extensive applicability in the realization of computer networks utilizing emerging technology. In the majority of multipliers that have been shown thus far, there is a unique separation between the partial product generators and the adders. This results in an augmentation of the quantity of blocks comprising the ultimate circuit. In the realm of reversible computing, it is imperative for circuits to incorporate ancilla inputs and garbage outputs. As the amount of blocks in a circuit increases, there is a corresponding increase in the quantity of inputs as well as outputs. The present work involves the creation of a column-wise structure for the multiplier, with the aim of decreasing the quantity of individual blocks comprising it. The expense associated with enlarging reversible circuits in modern devices is substantial. The quantity of blocks utilized in the framework proposed in this research for a reversible multiplier exhibits a notable reduction in comparison to preexisting designs. Additionally, the blocks in question are designed to limit the quantity of ancilla inputs and trash outputs. Hence, the suggested multiple criteria exhibit significantly lower values compared to those of the prior studies. In the suggested methodology, the multiplier is not built in two distinct stages; rather, every multiplier column is created as a cohesive block. Hence, the quantity of blocks comprising the circuit is equivalent to the quantity of columns in the multiplication process.

5. Impact of Soil Type on Sunflower Growth in Punjab: SPCC and Google Colab Analysis

   Harpreet Kaur Channi, Ramandeep Sandhu, Chander Prabha, Durgesh Nandan

   Abstract

   In India, and especially in Punjab, soil-based study examined the impact of loamy, clayey and sandy soil textures on the growth of sunflower. For performing the statistics, SPSS is used and google colab is used for managing and visualizing the data. Also, the sample collection of soil is performed below the ground and upto 15cmbased on the parameters such as pH, organic matter, electrical conductivity as well as analysis of other necessary nutrients such as phosphorous, potassium and nitrogen. In this work, Randomized Complete Block Design (RCBD) is applied for assessing the impact of soil type on sunflower development parameters such a height, stem girth and flower diameter. In this study, based on a time period of 12 weeks, sunflowers were cultivated and growth record was captured. Here, Google colab and seaborn and matplotlib are used for tracking the growth measurements. Various statistical tests have been performed using ANOVA and post-hoc in SPSS. It is well examined that loamy and clayey both outperformed than sandy soil for sunflower growth.

6. Multi-scale Decomposition for Multi-exposure Image Fusion

   B. Premalatha, M. Nagaraju Naik, P. Ravindra Babu, Sireesha Pendem, M. Ravindran, J. Prasannababu

   Abstract

   Significant advancements have been achieved in modern computational photography techniques to address the constraints associated with ordinary digital cameras while capturing scenarios where you can move around a lot. One such technique is High Dynamic Range (HDR) photography. Capturing all the intricate features of real sceneries in a single photograph poses a significant challenge due to the limited range of motion of typically employed picture recorders. In order to mitigate this issue, it is possible to employ a compilation of photographs taken under varying illumination circumstances to record the event. Subsequently, using the process of photo fusion, these photographs can be merged into a cohesive and informative image. MEF is a technique employed in the field of fusion of images, wherein multiple photographs of a given situation, each taken with different illumination options, are combined. The recent attention of MEF approaches can be attributed to the significance of producing high-dynamic variety images. In accordance with prior research, it has been shown that the multi scale structural patch decomposition-based MEF (MSPDMEF) technique offers superior fusion quality and faster processing time. However, a limitation of this approach is the potential loss of fine details of order to tackle this concern, we initially integrate edge preserving elements into our methodology to maintain the intricate characteristics of the amalgamated images within a singular framework.

7. Diabetes Prediction in Healthcare with Ensemble Learning

   Salliah Shafi, Gufran Ahmad Ansari, Mohd Dilshad Ansari, Vinit Kumar Gunjan

   Abstract

   Diabetes is chronic disorder it affects millions of people throughout world. It can be cured diagnosis, with proper treatment. With aid of various techniques developed and created using machine learning algorithms can identify, predict type of diabetes. Machine learning is becoming more and more popular today. Therefore, technique has been used in various medical scenarios. This work, we used Machine Learning Approach’s (MLA) and different measurement metrics such as precision, recall, Fi Score. This work, numerous supervised classifiers using machine learning is compared based on efficiency of multiple factors for early diagnosis of diabetes. Six MLA have successfully used in experiment research. With an accuracy rating of 98% Random Forest Classification (RFC) outperforms other classifiers for predicting early diabetes mellitus. In this study a framework for early diabetes prediction is developed. In addition, the dataset classification skew has been eliminated using the ten-fold cross-validation procedure.

8. Multiscale Modeling of Piezoelectric Polycrystals: Impact of Microstructure and Domain Arrangement on Macroscopic Properties

   Pankaj Gupta, Bhagat Singh, Yogesh Shrivastava, Durgesh Nandan

   Abstract

   As technology continues to shrink in size, the grain sizes and other microstructural scales of the materials used to make electronic components are starting to approach the size of the components themselves. When investigating the mechanical behaviour in this context, it is important to consider the impact of microstructure. The paper introduces a three-scale model for piezoelectric materials that incorporates multigrain and multidomain structures, based on homogenization theory. Intergranular domains were represented as a sub-microstructure at a very small scale. A collection of grains with random orientations was considered a microstructure at the mesoscale. The previous comprehensive model also featured a macrostructure exposed to external forces. A dual-domain structure was created for the case study analysis. This structure comprises positive and negative directional domains separated by a 180-degree orientation gap. The study examined how domain arrangement affects the macroscopic material properties of a polycrystal using a two-step homogenization process.

9. Application of Classification and Prevention Methods for Image Analysis with the CNN Model and Its Versions

   Avala Raji Reddy, Pavan Kumar Panakanti, P. Sravanthi Reddy, Rajesh Tiwari

   Abstract

   The procedure of labeling a picture is image classification, and deep learning is well suited to this area. This is because images are metaphysical and can use the parallel structure to learn a variety of characteristics. In this research, we define three basic arrangements for the Convolution Neural Network model. One setup is straightforward, and the other two are advanced variations on the fundamental level, which uses the overtaking avoidance method. CIFAR-10, a dataset of 60000 image sets of 10 different kinds of products, has been trained and tested. Based on the various performance matrices, comparisons of model variants indicate that the accuracy of the model may be significantly enhanced by employing the drop-out regularization and also, fewer batches could also get better results than bigger ones.

10. The Reduction of Hadoop Map While Maintaining the Privacy and Balancing Dynamic Demand Across Data Notes

    Nuthanakanti Bhaskar, M. Srirama Lakshmi Reddy, B. K. Chinna Maddileti, Rajesh Tiwari

    Abstract

    Generally, we work in Hadoop with two focal components: Map Reduce and HDFS. In opposition to the flooding capacity of data nodes, the client information is saved by Hadoop based on the group’s room usage of data nodes. Up until now, Hadoop has continued that way since not every datanode is running similar systems. Therefore, if jobs continue to operate on an identical Hadoop group, erratic workload hallmarks will generate poor functionality. Given the Hadoop log documents, I recommend accordingly a dynamic computation to modify the loading on a diverse rack that is being spread through another rack in an identical Hadoop group. Although, given any Hadoop group, transmit any knowledge to the unsecured group with information including assignments that work with delicate or straightforward data, and your privacy will be compromised because of your protected rack. We propose a way to distribute information sharing across different racks while maintaining privacy. Like this, reassigning assignments from the most intensely powered rack to a separate rack enhances the execution of Map Reduce jobs. Our reproductions suggest that the suggestion reduces the operational time of a job on the most heavily powered rack by over 50%.

11. Early Identification of Diabetic Retinopathy from Fundus Images Using a Multi-scale Feature Fusion Network Built on MobileNetV3

    S. Percy Deborah, K. Martin Sagayam, Kevin Philip Abraham, Shajin Prince

    Abstract

    Diabetic retinopathy (DR) is a major cause of vision loss in diabetic individuals, and early identification is essential to prevent progression to blindness. This work presents a machine learning approach using convolutional neural networks (CNN) to detect DR severity automatically. The model is designed to extract detailed features from retinal images by using a multi-scale feature fusion network built on MobileNetV3 using machine learning. Experimental results indicate that this approach surpasses existing methods, including various machine learning parameters, achieving high accuracy, sensitivity, and specificity. The proposed method starts by taking a retinal image as input. It then extracts three key features that help identify diabetic retinopathy. After that, the model calculates each feature’s importance (weight) and merges them using a weighted fusion process to improve accuracy. Finally, the classifier analyzes the refined data and determines the severity of DR. The experimental results highlight its potential to revolutionize DR screening by offering faster, more precise, and widely accessible diagnosis, particularly for remote and underserved regions. The approach facilitates prompt and accurate identification of diabetic retinopathy, supporting early intervention and enhanced patient outcomes.

12. Advanced Deep Learning Techniques for Parkinson’s Disease Diagnosis Using Voice Biomarkers

    Shaily Jain, Karan Bajaj, Chander Prabha, Durgesh Nandan

    Abstract

    The disease called Parkinson’s when progresses, it causes nerve cells to die. The symptoms can be motor and non-motor types and its early detection is an important concern. The study suggests a new way for deep learning to work using a Long Short-Term Memory (LSTM) along with a 3D-CNN. The training process for these (LSTM) layers involves a Genetic Algorithm (GA) when selecting features, to use the given voice signal data to diagnose PD dataset. The database consists of 195 voice recordings that each have 23 features. Measured values such as fundamental frequency, jitter, shimmer and nonlinear parameters. Spectrograms of voice signals are used as input in our hybrid model. Identifies how events are distributed over space and time and perform well Instead of using traditional machine learning or single deep learning models. The results of experiments show an accuracy percentage of 94.8% and an accuracy of 95.2% and a specificity of 93.5% are achieved which is higher than baseline results provided by Support the methods used are SVMs and two-dimensional Convolutional Neural Networks (2D-CNNs). It draws attention to the ways technology could be used by using advanced deep learning, it is setting the stage for machines that are small, easy to use and suitable for many people.

13. Backmatter