Browse Items (2150 total)
Sort by:
-
Automatic Generation Control of Multi-area Multi-source Deregulated Power System Using Moth Flame Optimization Algorithm
In this paper, a novel nature motivated optimization technique known as moth flame optimization (MFO) technique is proposed for a multi-area interrelated power system with a deregulated state with multi-sources of generation. A three-area interrelated system with multi-sources in which the first area consists of the thermal and solar thermal unit; the second area consists of hydro and thermal units. The third area consists of gas and thermal units with AC/DC link. System performances with various power system transactions under deregulation are studied. The dynamic system executions are compared with diverse techniques like particle swarm optimization (PSO) and differential evolution (DE) technique under poolco transaction with/without AC/DC link. It is found that the MFO tuned proportional-integral-derivative (PID) controller superior to other methods considered. Further, the system is also studied with the addition of physical constraints. The present analysis reveals that the proposed technique appears to be a potential optimization algorithm for AGC study under a deregulation environment. 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
Food Recommendation System using Custom NER and Sentimental Analysis
In today's fast-paced lifestyle, the need for efficient and personalized solutions is paramount, especially in the category of dining experiences. This research responds to this demand by proposing a better food recommendation system for Zomato reviews. It targets the audience who are not aware of the best cuisines and search for user reviews online. Utilizing custom Named Entity Recognition (NER) and sentiment analysis, the system seeks to understand and cater to individual food preferences extracted from user Reviews. Specifically, improving the analysis by extracting reviews for ten restaurants in the city of Kolkata. By providing a specific solution to address the current research gap in the area of restaurants recommendation systems, the system recommends top choices for neighboring restaurants and best food based on the sentimental analysis of the chosen menu items. 2024 IEEE. -
Reinforcement Learning for Language Grounding: Mapping Words to Actions in Human-Robot Interaction
Within the domain of human-robot communication, effective communication is paramount for seamless and smooth collaboration between humans and robots. A promising method for improving language grounding is reinforcement learning (RL), which enables robots to translate spoken commands into suitable behaviors. This paper presents a comprehensive review of recent advancements in RL techniques applied to the task of language grounding in human-robot interaction, focusing specifically on instruction following. Key challenges in this domain include the ambiguity of natural language, the complexity of action spaces, and the need for robust and interpretable models. Various RL algorithms and architectures tailored for language grounding tasks are discussed, highlighting their strengths and limitations. Furthermore, real-world applications and experimental results are examined, showcasing the effectiveness of RL-based approaches in enabling robots to understand and execute instructions from human users. Finally, promising directions for future research are identified, emphasizing the importance of addressing scalability, generalization, and adaptability in RL-based language grounding systems for human-robot interaction. 2024 IEEE. -
Prevention of Data Breach by Machine Learning Techniques
In today's data communication environment, network and system security is vital. Hackers and intruders can gain unauthorized access to networks and online services, resulting in some successful attempts to knock down networks and web services. With the progress of security systems, new threats and countermeasures to these assaults emerge. Intrusion Detection Systems are one of these choices (IDS). An Intrusion Detection System's primary goal is to protect resources from attacks. It analyses and anticipates user behavior before determining if it is an assault or a common occurrence. We use Rough Set Theory (RST) and Gradient Boosting to identify network breaches (using the boost library). When packets are intercepted from the network, RST is used to pre-process the data and reduce the dimensions. A gradient boosting model will be used to learn and evaluate the features chosen by RST. RST-Gradient boost model provides the greatest results and accuracy when compared to other scale-down strategies like regular scaler. 2022 IEEE. -
Advancing Credit Card Fraud Detection Through Explainable Machine Learning Methods
The world of finance has experienced a significant shift in the way money flows, due to the advancements in technologies such as online banking, card payments, and QR-based payment systems. These innovative banking payment facilities are offered by ensuring the safety of the transaction and ensuring that only the authorized customer can access and utilize these banking services. Credit card fraud is innovative way to cheat the user of the card. Government all over the word encouraging to the people for the uses of digital money. This research work focuses on analyzing the machine learning database by using a labelled dataset to classify legitimate and fraudulent business transactions with explainable AI. This study is based on decision tree, logistic regression, support vector machine and random forest machine learning techniques. 2024 IEEE. -
Customer Lifetime Value Prediction: An In-Depth Exploration with Regression, Regularization and Hyperparameter Tuning
In today's dynamic business environment, companies have been strategically shifting towards a customer-centric approach from their traditional product-centric focus. The main goal of this paper is to estimate customer lifetime value of 5,000 customers in the retail industry. This research follows a step-by-step approach to construct a multiple regression machine learning model. The model used in the study is based on the nine features to predict the customer life time value. First basic train-test split model is developed, which predicted 74% of variation in the customer lifetime value. This necessitates to improve the model performance, hence to address the multicollinearity problem lasso regularization is used. After lasso regularization , final model is trained with hyperparameter turning for further model performance improvement. The results show significant improvements in predicting customer lifetime value with the final model. This study suggests that the machine learning regression models can help to businesses to better understand how much value they can generate from individual customer.This deep understanding about customers helps retail businesses to align their customer engagement strategies to create a positive impact on the profitability and maximizing overall value offered to the customers. 2024 IEEE. -
Decoding Customer Lifetime Value to Unlock Business Success with Predictive Machine Learning Approach
This study highlights how crucial customers are for a company's success who directly impacts revenue and overall business value. This study focuses on analysis of customer lifetime value, the research uses data from 5000 customers with 8 important features with the main goal of predicting customer lifetime value. Business leaders often face choices about where to invest in marketing, like loyalty programs, incentives and ads or nothing. The study suggests that customer lifetime value is a key metric for making smart decisions, which measures how much a customer spends over their time with a company. To predict this value, the research explored different machine learning models - linear regression, decision tree regressor, random forest, and AutoML regressor. Each model is checked for how well it predicts customer spending habits. The results show that AutoML regression stands out for its accuracy without overcomplicating things. This study offers insights for businesses looking to improve their customer-focused strategies and long-term success. 2024 IEEE. -
The Development of Structured Tele Based Medicine Concept Using Programmable System
In the medical field, clinics and hospitals frequently use dispersed applications like telediagnosis. These apps must nevertheless provide information security in order to properly transit security measures like firewalls and proxies. The User Datagram Protocol (UDP) is often recommended for videoconferencing applications because of its low latency; nevertheless, security problems occur when UDP tries to pass through firewalls and proxies without a specified set of fixed ports. In order to overcome these obstacles, this study presents a revolutionary platform that uses Transmission Control Protocol (TCP) rather of UDP: VAGABOND, which stands for 'Video Adaptation framework, across security gateways, based on transcription,' Adaptation Proxies (APs) that are designed to accommodate user preferences, device variations, and dynamic changes in network capacity comprise VAGABOND. This platform's versatility at the user and network levels guarantees seamless operation in a range of scenarios. VAGABOND uses a binomial probability distribution to start making adaptation decisions. This distribution is formed from the retention of video packets inside a certain time period. VAGABOND gets beyond firewall and proxy constraints by using ordinary TCP ports (like 80 or 443) to provide videoconferencing data via TCP. But even though TCP is a dependable transport protocol, it can occasionally have latency and socket timeout problems. VAGABOND has clever adaptation techniques to deal with these problems and ensure smooth data transfer. 2024 IEEE. -
Front-End Security Analysis forCloud-Based Data Backup Application Using Cybersecurity Tools
In this challenging, demanding, daunting, and competitive business world, the rise, and growth of cybercrimes are very high. With the proliferation of Cloud Computing techniques, usually in industrial arenas, business information and important clients data are stored and managed using cloud platforms. Application programs are developed to handle such valuable information assets of the organizations. Cloud backups are provided for these client data where security is the most concerning aspect. There are many vulnerabilities in the current scenario where intruders can cause havoc. Destruction of the product can happen by exploiting vulnerabilities that can put the company and the product in jeopardy. It may create a bad impression about the organization among the customers, competitors, and the public world. This paper shows the work done by a cyber security team whose main objective is to run vulnerability analysis and mitigate threats on an application that backs up the clients data to the cloud. Cyber Security is an important aspect in all types of businesses because it protects all categories of data such as fragile data, private information, intellectual property data, and other data including governmental and industrial information systems from theft and damage which concludes in huge financial loss and loss of client data. 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
A Framework for Dress Code Monitoring System using Transfer Learning from Pre-Trained YOLOv4 Model
Maintaining a proper dress code in organizations or any environment is very important. It not only imbibes a sense of discipline but also reflects the personality and qualities of people as individuals. To follow this practice, some organizations like educational institutions and a few corporations have made it mandatory for the personnel to maintain proper attire as per their regulations. Manual checks are performed to adhere to the organizations' regulations which becomes tedious and erroneous most of the times. Having an automated system not only saves time but also there is very little scope of mistakes and errors. Taking this into context, the main aim and idea behind the project is to propose a model for detecting the dress code in such workplaces and educational institutions where the attire needs to be regularly monitored. The model detects Business Formals (Blazer, Shirt & Pants) worn by the personnel, for which CNN has been considered, along with YOLOv4, for performing the detection, due to its nature of giving the highest accuracy in comparison to the other object-detection models. Providing the Mean Average Precision of around 81%, it becomes evident that the model performs quite well in performing the detections. 2023 IEEE. -
The effect of airline service quality on customer satisfaction and loyalty in India
Indian Aviation Industry has been one of the world's fastest-growing aviation industries with private airlines representing more than 75 percent of the domestic aviation industry. With an 18 percent compound annual growth rate (CAGR) and 454 airports and airstrips in place in the country, 16 of which are designated as international airports, it has been stated that by 2011 the aviation sector will be witnessing a revival. In 2009, with traffic movement rising and revenues rising by nearly US$ 21.4 million, India's Airports Authority appears expected to earn better margins in 2009-10, as indicated by the Civil Aviation Ministry's latest estimates. The most crucial step in identifying and providing high-quality service is to understand exactly what customers expect. Quality of service is one of the best models for measuring customer expectations and perceptions. A company's performance results in customer satisfaction with a product or service. Passenger satisfaction is important to customer sovereignty. Customers can be loyal without being highly satisfied and being highly satisfied and yet not being loyal. Companies are required to gain a better understanding of the online environment relationship between satisfaction and behavioural intention, and to assign online marketing strategies between satisfaction initiatives and behavioural intention programme. In addition, the findings of this research will assist airline managers to better serve their customers, track and improve quality of service and achieve the highest level of satisfaction for their passengers. 2020 Elsevier Ltd. All rights reserved. -
Performance Analysis of YOLOv7 and YOLOv8 Models for Drone Detection
Drone detection techniques are used to detect unmanned aerial systems (UAS) also commonly known as drones. A rapid increase in these drones has limited the airspace safety and so the research for drone detection has emerged. This study compares between the two widely used deep-learning models, previously used YOLOv7 and the latest YOLOv8. The overall finding of this study suggests that the YOLOv8 deep-learning model appears to be more promising and may make valuable contributions on their own. We got the result that for 10 epochs YOLOv8 gave 50.16% accuracy while YOLOv7 gave 48.16% accuracy making YOLOv8 more promising for the task. As a practical application for future work, we intend to deploy YOLOv8 on edge devices to achieve real-time drone detection in critical security applications. 2023 IEEE. -
Performance Analysis of Various Machine Learning Classification Models Using Twitter Data: National Education Policy
With the exponential growth of social networking sites, people are using these platforms to express their sentiments on everyday issues. Collection and analysis of people's reactions to purchases of products, public services, etc. are important from a marketing and innovation perspective. Sentiment analysis also called opinion mining or emotion extraction is the classification of emotions in text. This technique has been widely used over the years to determine sentiment within given text data. Twitter is a social media platform primarily used by people to express their feelings about specific events. In this paper, collected tweets about National Education Policy which has been a hot topic for a while; and analyzed them using various machine learning algorithms such as Random Forest classifier, Logistic Regression, SVM, Decision Tree, XGBoost, Naive Bayes. This study shows that the Decision tree algorithm is performing best, compare to all the other algorithms. 2023 IEEE. -
Genetic Algorithm-Based Optimization ofUNet forBreast Cancer Classification: A Lightweight andEfficient Approach forIoT Devices
IoT devices are widely used in medical domain for detection of high blood sugar and life threatening disease such as cancer. Breast cancer is one of the most challenging type of cancer which not only affects women but in some cases men also. Deep learning is one of the widely used technology which provides efficient classification of cancerous lumps but it is not useful for IoT devices as the devices lack resources such as storage and computation. For the suitability in IoT devices, in this work, we are compressing UNet, the popular semantic segmentation technique, for the pixel-wise classification of breast cancer. For compressing the deep learning model, we use genetic algorithm which removes the unwanted layers and hidden units in the existing UNet model. We have evaluated the proposed model and compared with the existing model(s) and found that the proposed compression technique suppresses the storage requirement to 77.1%. Additionally, it also improves the inference time by 3.82without compromising the accuracy. We conclude that the primary reason of inference time improvement is the requirement of less number of weight and bias by the proposed model. The Author(s), under exclusive license to Springer Nature Switzerland AG 2024. -
An Efficient andOptimized Convolution Neural Network forBrain Tumour Detection
Brain tumour is a life threatening disease and can affect children and adults. This study focuses on classifying MRI scan images of brain into one of 4 classes namely: glioma tumour, meningioma tumour, pituitary tumour and normal brain. Person affected with brain tumours will need treatments such as surgery, radiation therapy or chemotherapy. Pretrained Convolution Neural Networks such as VGG19, MobileNet, and AlexNet which have been widely used for image classification using transfer learning. However due to huge storage space requirements these are not effectively deployed on edge devices for creation of robotic devices. Hence a compressed version of these models have been created using Genetic Algorithm algorithm which occupies nearly 3040% of space and also a reduced inference time which is less by around 50% of original model. The accuracy provided by VGG19, AlexNet, MobileNet and Proposed CNN before compression was 92.18%, 89.45%, 93.75% and 96.85% respectively. Similarly the accuracy after compression for VGG19, AlexNet, MobileNet and Proposed CNN was 91.34%, 88.92%, 94.40% and 95.29%. 2023, Springer Nature Switzerland AG. -
Whale Optimization Based Approach toCompress andFasten CNN forCrop Disease andSpecies Identification
In recent years deep learning and machine learning have been widely researched for image based recognition. This research proposes a simplified CNN with 3 layers for classification from 39 classes of crops and their diseases. It also evaluates the performance of pre-trained models such as VGG16 and ResNet50 using transfer learning. Similarly traditional Machine Learning algorithms have been trained and tested on the same dataset. The best accuracy using proposed CNN was 87.67% whereas VGG16 gave best accuracy of 91.51% among Convolution Neural Network models. Similarly Random Forest machine learning method gave best accuracy of 93.02% among Machine Learning models. Since the pre-trained models are having huge size hence in order to deploy these solutions on tiny edge devices compression is done using Whale Optimization. The maximum compesssion was obtained with VGG16 of 88.19% without loss in any performance. It also helped betterment of inference time of 44.13% for proposed CNN, 56.76% for VGG16 and 63.23% for ResNet50. 2023, Springer Nature Switzerland AG. -
Stock Price Prediction Using RNNs: A Comparative Analysis of RNN, LSTM, GRU, and BiRNN
Stock price prediction is a crucial area of financial market research, having significant implications for investors, traders, and analysts. However, given the dynamic and intricate nature of financial markets - which are impacted by a wide range of variables such as economic statistics, geopolitical developments, and market sentiment - accurately projecting stock prices is intrinsically difficult. Conventional techniques frequently fail to fully capture these dynamics, producing predictions that are not ideal. Recurrent Neural Networks (RNNs), one of the most recent developments in machine learning, provide potential methods to overcome these obstacles. Despite their potential, the effectiveness of different RNN architectures in stock price prediction remains an area of active research. This study compares four Recurrent Neural Network (RNN) architectures - Simple RNN, Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional Recurrent Neural Network (BiRNN) - for forecasting the Nifty 50 index values on the Indian National Stock Exchange (NSE) from the year 2000 to 2021. Using a comprehensive dataset spanning two decades, we assess each model's performance using the metrics Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Mean Squared Error (MSE). The data shows that the BiRNN model regularly outperforms the other models in all criteria i.e., MAE, MAPE, and MSE, indicating higher predictive accuracy. This study adds to the existing research by offering useful insights into the usefulness of RNN models, especially that of the BiRNN model for predicting stock prices, specifically in the setting of the Nifty 50 index. Our findings emphasize BiRNN's potential as a stock price prediction model and open new options for future research in this area. 2024 IEEE. -
A Deep Convolutional Kernel Neural Network based Approach for Stock Market Prediction using Social Media Data
Several economists and social scientists have held a longstanding fascination with the practice of stock market prediction. As the stock market is essentially uncontrollable chaos, many experts believe that trying to predict it is futile. Due to the complexity of the numerous factors, accurate stock price predictions are notoriously difficult to achieve. While the market behaves more like a scale than a voting machine over the long run, its behavior may be predicted with some certainty. Information from Twitter is used into the algorithm. In this proposed method, a convolutional extreme learning machine model with kernel support was introduced (CKELM). To improve feature extraction and data classification, the CKELM model builds on the KELM's hidden layer by adding convolutional and subsampling layers. The convolutional layer and the subsampling layer do not employ the gradient technique to fine-tune their parameters because some designs worked well with random weights. When compared to popular models like CNN and KELM, The proposed model fares quite well, with an accuracy of around 98.3 percent. 2023 IEEE. -
A Comprehensive Review of Linear Regression, Random Forest, XGBoost, and SVR: Integrating Machine Learning and Actuarial Science for Health Insurance Pricing
Actuarial science and data science are being studied as a fusion using Industry 4.0 technologies such as the Internet of Things, artificial intelligence, big data, and machine learning (ML) algorithms. When analyzing earlier components of actuarial science, it could have been more accurate and quick, but when later stages of AI and ML were integrated, the algorithms weren't up to the standard, and actuaries experienced some accuracy concerns. The company requires actuaries to be precise with analysis to acquire reliable results. As a result of the large amount of data these companies collect, a choice made manually may turn out to be incorrect. We will, therefore, examine alternative models in this article as part of the decision-making process. Once we have chosen the best path of action, we will use our actuarial expertise to evaluate the risk associated with specific charges features. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024. -
Enhanced Multi-Model Approach for Motion and Violence Detection using Deep Learning Methods Using Open World Video Game Dataset
For today's environment, it is extremely important to understand hostility and motion in a variety of contexts, particularly where accidents are concerned. There's also a high safety risk in public places if there is no proper identification of suspicious activities that occur fast and cannot be accurately observed through traditional surveillance systems that rely on constant human monitoring. Although deep learning algorithms have proven useful for detecting anomalies such as fraud recently, there has been little research on real-time crime detection because of issues related privacy when using live data sets. To tackle the key problem of motion and violence detection with current deep learning methods, this work exploits the Open World Game Dataset which provides realistic activities. The reliance on only one technique undermined the previous models' accuracy while this study comes up with various models to raise the detection precision and real-time processing capability. This work applies MobileNet SSD, YOLOv8 (You Only Look Once), and SSD (Single Shot MultiBox Detector) techniques to create a more accurate movement detection system. To identify violent or illegal behavior from videos, 3D convolutional neural networks (3DCNN) will be used alongside attention approaches. A diverse inexpensive training environment that enables simulating. 2024 IEEE.