Proceedings of the 7th International Conference on Communications and Cyber Physical Engineering

Table of Contents

1. Image Classification Based on Convolutional Neural Networks

   Samala Nandini, Potlakayala Deepthi, Manchala Bhavani, Kasapaka RubenRaju, BommaReddy Sindhuja, Aluka Madhavi

   Abstract

   Images serve a vital role and transmit a wealth of information. Finding important image information in a fair period of time is critical for many photos. A variety of factors influence image categorization results, including the algorithm's remarkable performance. To categorise an image, first input it into a particular classification technique. Preprocessing photos, identifying features, and building classifiers comprise the majority of image classification. The deep learning model's convolutional neural network can automatically extract local features and share weights, which is an advance above traditional machine learning's manual feature extraction. Picture classification produces better outcomes than more traditional machine learning algorithms. This study focuses on image categorization algorithms based on convolutional neural networks. Deep belief network approaches are also contrasted and evaluated, and the study finishes with an overview of the algorithms' application aspects.

2. Machine Learning and Artificial Intelligence in IoT: Integration Techniques and Applications

   Vemula Shiva Kumar, Pechetti Sujani, Sargari Swapna, Appari Lakshmi Prasanna, Maggidi Mounika, Sadula Sai Prasanna

   Abstract

   The convergence of Machine Learning (ML) and Artificial Intelligence (AI) with the Internet of Things (IoT) has unleashed a paradigm shift in the capabilities of interconnected devices. This paper explores the integration techniques and diverse applications of ML and AI in the IoT ecosystem. The synergy between these technologies empowers IoT systems to process vast amounts of data, make intelligent decisions, and enhance overall efficiency. The integration techniques encompassed in this study include data preprocessing, feature engineering, and model optimization tailored for IoT environments. Furthermore, the paper delves into the challenges and solutions associated with deploying ML and AI algorithms on resource-constrained IoT devices. In terms of applications, the research covers a broad spectrum, including predictive maintenance, anomaly detection, smart healthcare, industrial automation, and energy management. Case studies and real-world examples illustrate how ML and AI bring about transformative changes in these domains, paving the way for more sustainable and intelligent IoT solutions.

3. Machine Learning Applications in VLSI Design and Testing

   Pechetti Sujani, Sargari Swapna, Vemula Shiva Kumar, Sadula Sai Prasanna, Appari Lakshmi Prasanna, Maggidi Mounika

   Abstract

   Machine Learning (ML) has emerged as a transformative force in the realm of Very Large Scale Integration (VLSI) design and testing. This paper explores the integration of ML techniques to address the intricate challenges inherent in the creation and validation of integrated circuits. This paper provide a comprehensive overview, covering key aspects such as automated layout generation, circuit design optimization, fault detection, testing strategies, and manufacturing yield enhancement. The synergy between traditional VLSI methodologies and ML applications is examined, showcasing how ML algorithms can enhance efficiency, reduce time-to-market, and improve overall reliability. Real-world case studies and practical applications highlight successful implementations of ML in diverse VLSI scenarios. The ethical considerations of deploying ML in semiconductor technology are also explored. As the semiconductor industry advances, hardware acceleration for ML in VLSI becomes a focal point, discussing the integration of custom hardware architectures and accelerators in VLSI chips. The paper concludes by delving into future directions, including quantum computing, neuromorphic computing, and emerging technologies that promise to redefine the landscape of VLSI design and testing. This comprehensive exploration serves as a valuable resource for researchers, engineers, and students seeking a deeper understanding of the evolving intersection between ML and VLSI.

4. Neural Networks for Natural Language Processing: Techniques and Applications

   Sadula Sai Prasanna, Maggidi Mounika, Appari Lakshmi Prasanna, Vemula Shiva Kumar, Sargari Swapna, Pechetti Sujani

   Abstract

   Neural Networks (NNs) have become indispensable in the realm of Natural Language Processing (NLP), revolutionizing the way machines comprehend and generate human language. This paper explores the diverse applications and integration strategies of NNs within NLP, shedding light on the transformative impact of these technologies. The fundamental components, such as recurrent and convolutional architectures, and more recent advancements like Transformer models, are discussed in the context of sequential data processing. Transfer learning emerges as a crucial strategy, allowing models pre-trained on vast datasets to be fine-tuned for specific NLP tasks, overcoming data scarcity challenges. The applications span a broad spectrum, including text classification, named entity recognition, machine translation, text summarization, question-answering systems, chatbots, and conversational agents. Additionally, NNs play a pivotal role in speech recognition, information retrieval, and document classification. As the field evolves, ethical considerations, interpretability challenges, and the pursuit of explainable AI are scrutinized. This research contributes to the comprehensive understanding of the integration of NNs in NLP and provides insights into the future trajectories of this dynamic field.

5. Unlocking Cardiac Health: Prognosis of Cardiovascular Disease with Machine Learning and Deep Learning Techniques

   Divya Lalita Sri Jalligampala, Gangadhar Rao Kancherla

   Abstract

   Among the main causes of mortality for the general public is cardiovascular illness. The likelihood of survival for patients with cardiac problems is significantly impacted by late detection. Life-threatening heart disorders are recognized to be impacted by many elements, like age, gender, blood sugar, cholesterol, and heart rate. However, because there are so many variables, it can be challenging for an expert to assess each patient while considering this information. The opportunity to apply machine learning (ML) and deep learning (DL) algorithms to predict cardiac diseases is investigated in this review paper. The benefits and drawbacks of traditional machine learning methods, such as Decision Tree classifier, Support Vector Machine algorithm, Random Forest classifier, ANN and few more in the context of the forecast of heart disease, are examined and evaluated. Additionally, the research explores one of the rapidly developing subjects of deep gaining knowledge (DL), looking at developments in convolutional and recurrent neural networks (RNNs) that are especially suited for analyzing medical information, including Electronic Health Records (EHRs) and electrocardiograms (ECGs). The effectiveness of ML and DL techniques, emphasizing the precision, comprehensibility, and applicability of each are being evaluated in this paper. The state of heart disease prediction using ML and DL is critically examined in this review, with an emphasis on both the field's achievements and shortcomings. Conclusions are drawn by highlighting the significance of these AI-powered strategies will transform the prevention of heart disease and enhance patient outcomes, as well as ethical considerations and future goals for this quickly developing subject.

6. Plant Leaf Disease Detection Using Machine Learning

   B. Tirapathi Reddy, V. N. Janardhan K, N. Likhitha, P. Keerthana, P. Dhaarini, Siva Sankar Namani

   Abstract

   To enhance crop yield, it is important to identify and prevent crop diseases. This paper utilizes deep convolutional neural networks (CNNs) to detect and diagnose plant diseases from their leaves. Conventional CNN models typically need a substantial number of parameters and computationally expensive operations. Therefore, we employed transfer learning to replace the standard CNN with models such as VGG-16 and RESNET-34 using the PYTORCH framework. The trained model was tested using a dataset comprising 14 different plant species and 38 categorical classes, achieving an accuracy of 90.42%.

7. Intelligent Speed Assistance with Embedded System Technology

   M. Pavani, D. Jayasri, M. Shreyan, P. Kowstubha

   Abstract

   India is considered one of the most congested countries globally in terms of road traffic. The nation comprises approximately one percent of the total number of vehicles worldwide, yet it is responsible for six percent of all road traffic accidents globally. These numerous counts of road accidents are majorly due to over-speeding, and loss of control of the vehicle. That is the reason Speed is one of the most studied topics related to road accidents. The main objective of this paper is to control the Speed of Electric Vehicles with pivotal technology - Intelligent Speed Assistance (ISA). Speed limits can often result in slower traffic and frustrating delays but ISA Electric Vehicle is a game changer, offering drivers a way to enjoy the speed and efficiency as they need without compromising safety. ISA systems in EVs utilize a combination of advanced sensors to monitor the vehicle’s speed and maintain compliance with speed limits. ISA adapts to varying speed ranges to optimize safety and efficiency, especially in three modes: Low speed, Moderate Speed, and high speed using road-specific data, driver feedback, and a few other sensors. The findings of this paper could help enhance the awareness of drivers regarding the cognitive factors that impact road safety.

8. Machine Learning Models for Salary Prediction: A Comprehensive Literature Survey

   Sunitha Tappari, Arshiya, Ambati Nirmala, Silveru Sowmya, Karedi Suchithra

   Abstract

   In today’s business scene, predictive engines are commended for their accuracy and cost-effectiveness in tackling a wide range of difficulties. This study emphasizes the importance of optimal data utilization. Machine Learning (ML) emerges as an effective approach for improving prediction accuracy in software applications. Fair and competitive salary structures are critical for retention in compensation planning. The study investigates machine learning models for wage prediction, to optimize compensation planning in modern enterprises.

9. Ethical Implications of Machine Learning in Data Mining

   Manchala Bhavani, Kasapaka RubenRaju, BommaReddy Sindhuja, Aluka Madhavi, Samala Nandini, Potlakayala Deepthi

   Abstract

   As machine learning techniques become increasingly integral to data mining processes, ethical considerations become paramount in ensuring responsible and fair use of technology. This paper explores the ethical implications associated with the intersection of machine learning and data mining, shedding light on the challenges and concerns that arise in this evolving landscape. We examine issues such as privacy infringement, algorithmic bias, and the potential societal impact of automated decision-making systems. The paper also discusses the importance of transparency, accountability, and inclusivity in mitigating ethical risks. By analyzing real-world examples and existing frameworks for ethical guidelines, we aim to provide insights into fostering a responsible approach to machine learning in data mining.

10. Neural Networks in Healthcare Applications, Challenges, and Future Trends

    Appari Lakshmi Prasanna, Sadula Sai Prasanna, Maggidi Mounika, Pechetti Sujani, Vemula Shiva Kumar, Sargari Swapna

    Abstract

    Neural Networks (NNs) have emerged as powerful tools in revolutionizing healthcare applications, demonstrating the potential to enhance diagnostics, disease detection, and treatment planning. This paper provides a comprehensive overview of the current state, challenges, and future trends of NNs in healthcare. We delve into the significant strides made in medical imaging, where NNs have proven adept at analyzing complex datasets and aiding clinicians in accurate diagnoses. Furthermore, NNs play a crucial role in predicting disease risks, formulating personalized treatment plans, and expediting drug discovery processes. Despite these advancements, challenges such as interpretability, data quality, and ethical considerations persist. The interpretability of NN models remains a critical concern, requiring further exploration to ensure transparency and user trust. The availability of high-quality, diverse datasets is essential for the robustness of NNs, necessitating continued efforts in data collection and curation. Ethical considerations, including biases in algorithms, fairness, and accountability, demand ongoing research to mitigate potential pitfalls. In terms of future work, the paper proposes research directions to address these challenges and enhance the integration of NNs into healthcare. This includes the development of interpretable NN models, optimizing for real-time applications, and fostering collaboration between machine learning experts and healthcare professionals. Continuous learning mechanisms, rigorous validation through clinical trials, and the exploration of human-machine collaboration further underscore the roadmap for future research.

11. Robotic Patroller for Women Safety

    T. Aparna, Datta Manvita, Kunuru Nikitha, Renikuntla Nayani

    Abstract

    The safety of women is a significant concern worldwide, especially in remote areas. To address this issue, we propose the implementation of a security patrolling robot. This innovative system integrates advanced cameras and sound sensors onto a wheeled robot, enhancing security measures within designated premises. Operating along a predefined path, the robot continuously monitors its surroundings for any signs of activity. Equipped with cameras, it captures real-time images, while sound sensors enable the detection of unusual sounds. Upon detecting such sounds, the robot autonomously redirects its path towards the source, utilizing its camera system to capture images of any individuals present. These images are then transmitted to a server for review by nearby police center which can take appropriate actions as necessary. This system effectively safeguards large premises and provides a proactive approach to ensuring women’s safety.

12. Cutting Edge Mechanism for Child Retrieval from Risky Pits

    B. Venkateshulu, K. Varsha, Ch. Nainitha, M. Nikitha, G. Bhanusri

    Abstract

    The purpose of this study is to successfully rescue the kids from dangerous places like exposed borewells. The kid retrieval mechanism for borewell dimensions that we employed in this work has a gripper-based gripping system and a camera positioned beneath the bottom plate that is powered by a winch that is connected to the top motor shaft. A processing unit with an 89c52 controller chip H Bridge IC is used to govern the mechanism, which is powered by two DC motors and controlled by an RF remote unit for effective kid rescue and surveillance. The wired camera enables live video transmission. For further safety, the youngster is also given oxygen, and a motion sensor sounds an alert for the presence of a borehole. An open bore well will be simulated using a 2-foot plastic pipe (8 inches in diameter) for practical validation.

13. Reinforcement Learning in Data Mining - Applications and Emerging Trends

    Kasapaka RubenRaju, BommaReddy Sindhuja, Aluka Madhavi, Samala Nandini, Potlakayala Deepthi, Manchala Bhavani

    Abstract

    Reinforcement Learning (RL) has gained significant traction in the field of data mining, offering a paradigm shift in how algorithms learn and adapt from interactions with an environment. This paper explores the application of RL in data mining and highlights emerging trends. RL excels in scenarios where systems make sequential decisions, making it suitable for dynamic and evolving data environments. Its ability to optimize decisions over time has found applications in diverse domains such as finance, healthcare, and autonomous systems. In data mining, RL contributes to improved pattern recognition, anomaly detection, and decision-making processes. The paper delves into key applications, including fraud detection, recommendation systems, and resource allocation, showcasing how RL enhances the efficiency and adaptability of data-driven models.

14. Precision Weed Control for Sustainable Agriculture

    Ashok Kumar Malhotra, Syed Ismail

    Abstract

    The global agricultural landscape is facing unprecedented challenges, with farmers grappling to maintain optimal crop yields, indiscriminate usage of herbicides with prolonged repeat cycles creates health risks for humans, use of chemical herbicides to control weeds and contaminate soil, water sources, and even the harvested crops, leading to a range of health and ecological issues. Interfering in the ecosystem. This project aims to use artificial intelligence for image processing to categorize weeds and limit the use of chemicals for weed control. A custom data set annotating various weed category images was created to train the neural network. Users must feed the field area image to the trained model as input. This image passes through a neural network to detect weeds or crops in the specified image. Users get a weed category or healthy crop image as an output. Based on the detected weed timely intervention by the user to remove the weed can be initiated. This image can be a satellite image at a fixed duration from the field to send an alert message to the landowner for weed detection in the crop life cycle. Timely intervention of pest infestation etc. can also be detected. Removing weeds or weed plants will reduce the use of blanket chemical herbicides. This will not only reduce the use of chemicals but also improve the crop yield significantly.

15. AI-Powered Data-Driven Resource Allocation and Scenario Planning for Healthcare Systems

    Karpagam Anandan, Syed Ismail

    Abstract

    The research harnesses the capabilities of Generative AI (Gen AI) to dynamically predict resource needs and proactively allocate critical healthcare resources, such as beds and ventilators, in real-time during unpredictable pandemics. Through its ability to uncover hidden patterns in data, Gen AI assists in generating diverse outbreak scenarios, enabling us to anticipate resource demands and optimize resource allocation strategies effectively. Moreover, Explainable AI (XAI) is seamlessly integrated into our framework to provide transparent model insights, ensuring trust and collaboration between AI technologies and healthcare professionals. The novel fusion of Gen AI and XAI not only enhances real-time resource allocation but also fosters a culture of trust and transparency in predictive analytics, thereby promoting more effective healthcare resource management. By anticipating resource bottlenecks and implementing preventative measures based on transparent model insights, our approach proactively mitigates future challenges, creating resilient and adaptable healthcare systems equipped to address crises with efficiency and foresight.

16. Medicare Chatbot

    Ambidi Naveena, Afifa Zohreen, T. Vyshnavi, S. Likitha, Sai Sharanya Ponnala

    Abstract

    The Medicare Chatbot improves healthcare efficiency and accessibility for users by providing quick, personalized Medicare information. This paper describes a Chatbot aimed to aid in disease diagnosis by analysing user-input symptoms and making predictions about likely diseases. The Chatbot also provides advice on healthy eating and makes it easier to schedule doctor appointments. To accomplish this, The Convolutional Neural Network (CNN) algorithm has been trained using a vast dataset that maps disease names (in the ‘Source’ column) to associated symptoms (in the ‘Target’ column). By assessing these symptoms, the Chatbot can reliably forecast diseases and provide personalized food suggestions and home cures. According to performance measures, the Chatbot using the CNN algorithm outperforms the Chatbot using the SVM algorithm. The integration of the CNN algorithm enhances the Chatbot’s diagnostic accuracy by efficiently recognizing complex patterns in symptom data. This unique technique provides a potential step toward accessibility.

17. Detecting Fake Identities with Machine Learning: A Comprehensive Literature Survey

    Vijaya Lakshmi, Aishwarya Rao, Kalyani Boddulah, Ojaswi Cheekati, Varsha Vadla

    Abstract

    The exponential growth of fake profiles on social media platforms has spurred a concerning surge in cyber threats, encompassing misinformation dissemination, phishing attacks, and identity theft. The study investigates machine learning techniques to address this urgent problem by differentiating between genuine and fake profiles by carefully examining characteristics like follower numbers and posting frequency. This helps to reinforce efforts to establish a more secure online environment and promote authenticity, strengthening digital trust.

18. Mute Mate – The Comfort of Automation

    A. Sharada, M. Sindhuja, S. Tejaswini, S. Sruthi, Adhya Indraganti

    Abstract

    Television advertisements are integral to sustaining broadcasting networks and presenting products or services to a wide audience. Nonetheless, the impact of commercial interruptions can be far from ideal. The loud, abrupt transitions from program to advertisement can startle and disturb viewers. For individuals with sensory sensitivities, such disruptions can be especially overwhelming. In response to these concerns, a novel approach seeks to revolutionize the television experience by automating the muting process during ad breaks, creating a seamless and inclusive viewing environment. The proposed model introduces an ingenious solution that enhances the television viewing experience by automatically muting the audio during ad breaks. Upon turning on the television, the device seamlessly activates alongside it, equipped with a sophisticated speech recognition algorithm and keyword spotting technology. As the program progresses, the model continuously listens for specific phrases, such as “We’ll be back after a short break,” commonly associated with ad interruptions. When these phrases are detected, the model promptly sends an infrared signal to mute the television’s audio, ensuring a smooth transition and allowing viewers to enjoy their content uninterrupted. With its fully automated and seamless functionality, the proposed method offers a seamless experience like never before.