Advances in Computing and Information

A Light Detection and Ranging (LIDAR) system is a very useful tool in the exploration of sparse environments, such as underground tunnels and during disasters. LIDAR can also be used to measure distances from the sensor to an object by illuminating that object with a laser light and then measuring time for the reflected light to return. The LIDAR is a device that emits and measures light in the form of a laser which reflects from the ground to the lidar. The aim of our research is to make use of LIDAR technology and explore unknown environments with autonomous navigation at ease. In our research, we used LIDAR and made a robot that can generate a 2D map of the surrounding environment and can help the operator to analyse the interior part of it. We have used Raspberry Pi to pre-process the data from the LIDAR and used Robot Operating System (ROS) to interpret the data on the graph to generate the map. This system is almost human independent and is risk free. Hence, generated 2D map is used for further process to inspect if the environment is safe for human to enter. In this research, we have also implemented the autonomous navigation part. Once the map is generated, the robot can move from one point to another point taking the shortest path. Suppose there is new obstacle in the path, the robot is capable of generating the next feasible alternative path to reach the destination Manoj Kumar and Nandakumar (Int J Grid Distrib Comput 13:1622–1627, 2020). SLAM technique is used to predict the position of the robot and simultaneously navigate. In this paper, we have explained in detail about how the data is processed and the functionality.

Knowledge dissemination had never before been hampered in the history of humanity until the World Wide Web's development and the rapid adoption of social media outlets. As a result of the growing usage of social media platforms, fake news is increasingly common in all kinds of circumstances. After the internet evolved, most of the people are utilizing Internet for their personal purpose only at the same time they are uncontrolled to read many of fake news, also. Automated classification of a text article as real or fake is a challenging task. In this situation, to detect such types of fake news and to provide well verified news to our society, the machine learning (ML) techniques such as support vector machine, linear regression, K-nearest neighbor, neuro-evolution of augmenting topologies (NEAT) and boosting NEAT are applied in this research. After preprocesses over the actual dataset methods effectively identify the fake news with collected dataset and evaluated by the metrics such as accuracy, precision, recall and F1-score.

Wordle is an online word guessing game that recently gained immense popularity. The goal of this game is to guess a five-letter word within six tries. A number of attempts have been made at finding the best Wordle opener or the best word to start the game with. This paper, however, focuses on building an algorithm that can win the game for a varied set of answer words. The purpose is not to obtain the best starting word but to develop a generic strategy to win the game. A cost-based approach has been utilized for the same. The algorithm has further been tested with a diverse set of five-letter words and gives an average accuracy of around 70% for different starting words.

The current starved voltage-controlled oscillator offers low power consumption, excellent integration capability and smaller area when compared to the other VCOs. A five-stage CSVCO for phase-locked loop (PLL) is designed using Gm/Id methodology. The tuning range is from 600 MHz to 1.3 GHz can be achieved by adjusting the control voltage of the VCO between 0.6 and 2 V. The supply voltage used is 1.8 V, and the circuit is designed using 180 nm CMOS technology. The circuit design and simulation of this work is carried out in Cadence Virtuoso EDA environment.

With the rapid evolution of deep convolutional neural networks (CNNs), major breakthroughs have been achieved in object detection in the field of computer vision. However, the majority of state-of-the-art detectors, in both one-stage and two-stage methods, have limits and are inadequate for usage in a real-world setting where each step must be thoroughly checked. This thesis investigates advanced object detection models and frameworks. It offers an in-depth analysis of the most recent object detection models, their frameworks, and the performance criteria used to evaluate such models. The object detection models selected are YOLOv5, Faster R-CNN using Detectron 2, and SSD using TensorFlow 2, and the dataset selected is the Vehicles-Open Images Dataset. The performance of the selected models in relation to many metrics is analyzed, and the findings are reported. In conclusion, the benefits and limits of the selected models, as well as their relative performance, are discussed.

Every magnetic induction wireless power transfer (WPT) or inductive power transfer (IPT) system operates on the two guiding principles of Faraday’s law of induction. A conductor can generate an electromotive force (EMF) that is inversely proportional to the strength and rate of change of the magnetic field when an alternating magnetic field is present, as stated by Faraday's law of induction. A conductor can generate an electromotive force (EMF) that is inversely proportional to the strength and rate of change of the magnetic field when an alternating magnetic field is present, as stated by Faraday's law of induction. Wireless power transmission technology provides a significant advantage over the conventional way of transferring power over a distance. Here, the electrical energy can be transferred wirelessly without using the cables. In this paper, based on the working principle of magnetic coupling resonant wireless power transmission (MCR-WPT), we have designed and analyzed the major circuit with the help of simulation. The simulation of the design is carried out by using Multisim software, and also, the hardware implementation is done.

In the recent years of development, deep learning (DL) is very useful in all fields with the growing availability of data. The main goal of DL technology is to make faster, reliable, and good decisions. Because of this ability, DL has found its use in healthcare, particularly with a focus on various types of medical images or images related to patients’ health. These fields have diagnostic processes that depend on gathering and processing large amounts of medical images. This work proposes a deep learning (DL) model based on “Convolutional Neural Network (CNN)” for identifying COVID-19 and Pneumonia using Chest X-Ray images. The result of this processing helps the radiologists to derive insights and make decisions to determine the correct diagnosis of the patient. This model helps in two ways. Firstly, to classify whether a chest X-ray shows any sort of variations with respect to COVID-19 and pneumonia or not. Secondly, to classify with the help of normal chest X-ray images. Experiments were carried out using InceptionV3 (IV3), and VGG16 models on the chest X-ray images. Results revealed that the VGG16 model outperformed IV3 model for COVID-19 and pneumonia identification on multiple performance metrics.

In today’s fast-growing Internet world, customer ratings and reviews play an essential role in online buying on e-commerce websites such as Amazon, Flipkart, and others. Sentiment analysis is crucial for increasing customer satisfaction on e-commerce sites since it contains a lot of consumer feedback. In this work, we have used Amazon Women's E-Commerce Clothing Reviews dataset. We have used CountVectorizer and TF-IDF and trained the data on five machine learning (ML) classifiers, namely logistic regression (LR), multinomial Naive Bayes (MNB), Bernoulli Naive Bayes (BNB), support vector machine (SVM), random forest (RF), and AdaBoosting (AB). When comparing the ML model’s accuracy scores concerning the CountVectorizer, it was discovered that MNB and LR models had the highest accuracy of 0.94, while RF had the lowest accuracy of 0.90. SVM achieved the maximum accuracy of 0.94 using the TF-IDF approach, and MNB achieved the lowest accuracy of 0.89. The accuracy, precision, recall, F1-score, and AUC-ROC curve help us to determine the performance of the ML algorithms. To examine the dataset’s attributes and comprehend the relationships between the variables, many statistical techniques were applied.

As new ways to generate renewable energy cost effectively are being developed, high load devices such as electric vehicles (EVs) are becoming common. EVs require high current and voltage for charging, due to the large size of the batteries and the requirement for fast charging. This has made intelligent smart grids necessary, as traditional grids are incapable of dealing with the disturbances in the grid produced by the load required by these devices. Our paper proposes an intelligent smart grid which is interconnected using LoRa network and uses a long short-term memory (LSTM)-based neural network for processing, forecasting and predicting the disturbances produced in the network. The LoRa modules will be used in a mesh configuration; this will remove the need for a repeater and also give us greater network coverage. A LoRa access point will be placed, which will collect data from the individual modules and send them to the power station. Here, the LSTM algorithm will be applied on the data collected thereby doing the necessary predictions. The use of LoRa has make the system independent of cellular networks and due to its low power consumption, and efficiency have been improved. LSTM algorithm is capable of dealing with not only individual data but also complete time series. It can also deal with missing data, which makes it much more beneficial. Overall, our proposed system can greatly benefit both consumers and electricity companies.

The development of an obstacle-avoiding robot is a crucial step toward the development of autonomous or unmanned vehicles. These vehicles are used in situations that do not allow for human interaction, such as transportation, surveillance, and rescue. This future strategy will advance humanity’s understanding of unmanned vehicles significantly. This article implements an obstacle-avoiding robot that can detect objects in front of it using an ultrasonic sensor and a microcontroller. The ultrasonic sensor is used in this prototype, and the microcontroller configures the Arduino Uno board to measure distances using the Arduino IDE software. It also encompasses a wireless camera for live video transmission, which can be accessed via an IP address by a variety of terminals including laptops, smart phones tablets, and personal computers. This is accomplished by programming the ESP 32 camera module, which also functions as a Wi-Fi Internet and Bluetooth module. The live streaming camera is an ESP 32 CAMERA module that was programmed with Arduino IDE software and a CP2102 USB-TTL converter.

Apart from the widely researched detection and segmentation domains, we reconnoitered the embodiment to audio-visual segmentation (AVS) which consists of visual object localization with connected audio patches for each frame. Recently, pixel level image categorization had achieved saturation levels with satisfactory results. But their corresponding audio segments were least explored dropping that valuable support information which makes the model less prone to multiple real-time scenarios. Hence, in this paper, a graphical neural network with attention-based audio-visual segmentation (GWA-AVS) model is proposed, which is a full attention-based graphical neural network (GNN) designed to segment object masks from the audio of the visual frames with single or multiple sound sources. Distinct approaches were investigated to fine-tune the proposed GWA-AVS frameworks to get superior results. Attention block is added to improve the learning of meaningful features of visual frames, thereby improving performance of the proposed framework. GWA-AVS is benchmarked on AVSBench dataset achieving better results than the existing SOTA models with less computational cost, i.e., by using less number of model parameters than existing AVS model. Qualitative results of GWA-AVS evidently show that the generated masks are clear with crisp shape of sounding objects depicting the robustness of our model in real-time scenarios.

Synthetic aperture radar, sometimes known simply as SAR, is a technique for imaging a target from space that uses microwave radiations to illuminate the object that is the focus of the picture. These brief bursts of microwave radiation are sent with the help of a piece of equipment known as RADAR. Denoising is an important pre-treatment step in image processing that has to be finished before an application-friendly picture can be produced. This step is necessary to create an image. They are versatile enough to be used for everything from digital photography to photographing satellites. Their range of applications is rather extensive. The imaging method known as synthetic aperture radar may be used in any weather, both during the day and at night, and it uses airborne radar to illuminate the earth’s surface. Satellites inspired the development of this particular technology. Processing must first be performed on the recorded backscattered signals before synthetic aperture radar images can be produced using them. Because microwaves can travel through clouds and dirt, they are often used in settings where traditional imaging would be difficult. Speckle noise is a natural occurrence that may be observed as granular patterns in synthetic aperture radar images. These images are produced by synthetic aperture radar (SAR) technology. It could be difficult to eliminate speckles, a random multiplicative noise. Speckle is a sort of noise. Speckle is a kind of noise that occurs in coherent systems and is formed when echoes interact with transmitted signals. This interaction may result in either constructive or destructive interference. The appearance of the speckle lowers the overall image quality, which means that the application cannot use it since it is improper. Denoising SAR pictures is a tough approach, but it is important since there is noise in the images and it has to be removed. Denoising is a strategy that should be used to eliminate the noise, but all of the essential visual information should be preserved. Denoising may be done in either the spatial domain or the transform domain, both of which are acceptable choices. Combining the dual tree curved wavelet transform with modified cellular neural networks is said to produce a one-of-a-kind filter, which has both been hypothesized and produced as a possibility (DTCWT-MCNN). The recommended filter is implemented using SAR images and put through its paces to see how well it works. The recommended approach is given a score based on its efficacy, which is determined by utilizing objective metrics, and its performance is analysed to see how effectively it functions. Performance evaluation metrics such as Noise Mean Value (NMV), Mean Square Difference (MSD), Equal Number of Looks (ENL), Noise Standard Deviation (NSD), and Speckle Suppression Index were utilized to carry out quantitative confirmation of the findings. These metrics were utilized to confirm the findings (SSI).

Corona virus is considered as a scourge all over world by which most of the people got infected. Since the virus was spreading very quickly governments had to bring in very tough rules such as lock down. Lock down disrupted almost all sector like education, business etc.… During lock down and after the lock down people have used social media as their platform to express the opining regarding the pandemic. Tweeter is one such social media where people had shared their thoughts on the pandemic. In this study tweets which are made by the public is considered to analyses the sentiments. And in this research machine learning models are used to classify the data into five different classes namely positive, negative, neutral, extremely positive, and extremely negative. Decision Tree, Random Forest, Support Vector Machine, Logistic Regression, and Naïve Bayes are the ML models used to classify the sentiment and are achieving on accuracy of 99%, 99%, 97%, 94%, and 73%, respectively. Further precision, recall, F1-score, and world clod are calculated for each and every class.

Sentiment Analysis is widely used in the process of mining the data, to predict emotion of a sentence through Natural Language Processing (NLP). The main aim is to find the accurate polarity of a sentence. Therefore, to find the polarity or sentiment of a user or customer for a product there is a need for automated data analysis techniques. In this paper, a detailed analysis of classification techniques is used in Sentimental Analysis of Amazon Product Reviews with recommendation for a best buy product on web. Multinomial Naïve Bayes, Random Forest, Logistic Regression, Decision Tree, and SVM classifiers are tested and compared. Random Forest gives the best accuracy of 94.94%. Web scraping extracts five Amazon products on Amazon.com and recommends the best buy product on the basis of polarity score of each product and here Samsung Galaxy M01 is recommended as the best buy product.

Access to the Internet is growing exponentially due to its ease of usability, flexibility, and lowering data plans. The diverse network service requirements encourage mobile operators to look for mechanisms that facilitate efficient use of network infrastructure, so that it can reduce the operational and expenditure costs. Use cases like the video streaming services requires high bandwidth, autonomous driving and remote medical surgery requires low latency, and various IoT applications work with low bandwidth to cater to the users needs. We simulate the RAN slicing using an emulator called eXP-RAN which effectively manages the allocation of different network resources to the created slices. The infrastructure, slicing, and service layers are the three distinct layers in the proposed system architecture. The isolation and abstraction of the network resources is also applied to the created slices by this emulator.

The term “Internet of Things (IOT)” has been contested because devices only need to be independently reachable and accessed through the Internet, not even the whole Internet. In the IOT the wireless sensor network (WSN) part of it. IOT WSN is used to connect the different sensors. Numerous uses for wireless sensor networks are currently being researched. The sensor network usually operates on the batteries. Reducing energy consumption during data transmission and node communication will increase network lifetime. One of the challenging criteria in the IOT is maintaining the energy level for a longer duration. The energy level of the sensor networks is consumed mainly during Base Station (BS) communication, inter-node communication, and data sensing. The proposed approach is a multimodal approach, creates a meta-heuristic optimization strategy to lower the communication's energy-level consumption of the sensor networks. The suggested model will segment the WSN into various clusters. The proposed approach has the combination of the modified LEACH and Modified Cuckoo algorithm to reduce the energy consumption of the WSN-IOT sensor nodes during communication. The modified LEACH algorithm is used to choose the cluster's head (CH). In order to choose the best route from CH to BS, the Modified CUCKOO algorithm is employed. With an ideal number of clusters, balanced energy dissipation, and low-energy consumption, the proposed model can improve network performance.

Artificial intelligence is gaining popularity in every aspect of business and development of organizations. In manufacturing sector, application of this technology is constantly evolving. However, application of artificial intelligence technology in the manufacturing sector is in the various sectors of manufacturing such as the supply chain management, production testing, quality assurance and engineering. The present study is undertaken to understand the areas of application of artificial intelligence in the manufacturing sector by considering the areas of quality assurance, product design and development, purchase, order-level management, maintenance, logistics and supply chain management. The results of the study show areas of application of artificial intelligence and future directions for enhancing this technology in the manufacturing sector.

In the twenty-first century, women participate in many activities on an equal footing with men. Women contribute greatly to the success of society and nation, and this is commendable. However, women’s freedom of movement is limited by sexual violence. Now, sexual harassment is one of the major barriers to women’s empowerment; hence, girls and women are placed in insecure environment. So, women should have some safety device that can be used during critical situations. The most challenging scenario includes situations in which a women may not have a mobile phone or unable to use mobile phone in dangerous situations. The proposed system overcomes these challenges by inserting a device into women's accessories to resemble a hidden device and can be activated in dangerous situations which sends the location of the user to her parents and guardians contacts thereby saving her from danger. The proposed system is designed in such a way that it works even without the user's mobile phone. Mobile phone is just used to update the current guardian’s information only. Using Android Studio, an app is developed, and using this app, parents and guardians contact information can be updated based on place such as guardians of city1, guardians of city2, etc. When the user is in city1, the app will be updated to city1 and also updates the cloud database. Hence, the cloud system will have a database of contacts of the user's parents and guardians in a particular city to save her from danger at the earliest. During danger condition user activates hidden hardware system, then distress signal is sent to the cloud which sends a signal to the parents and guardians mobile devices. In the parents and guardians mobile, the signal is in the form of an alert call with an email that contains the location of the user so that the user can be tracked easily and can be saved in time. The hardware includes ESP32 Wrover with an in-built SIM800 module which provides a data link to a remote network. Global Positioning System (GPS) is used to get the location of the user.

There are a variety of automated diagnostic techniques and models using numerous supervised learning models, but most of these models cater especially to the diseases that are seen in the Western countries, and they rarely see the outbreak of diseases such as malaria, dengue, etc. Early detection of these diseases can control the mortality rate and help save lives. Malaria while being a curable disease still has no vaccine available for it, so early detection of malaria can help determine the risk and can prove to be lifesaving. And with the time taken to collect, analyse and diagnose malaria in the blood is valuable time that can be the difference between a patient’s life or death. And less developed countries do not have the proper resources for fast response against the disease. In this project, we are hoping to develop an effective and efficient automated diagnostic model using machine learning models. For this, we have implemented a model based on CNN architecture to detect malarial parasites in blood cells and then use advance image processing techniques to contour and isolate the parasite to track the progression of the disease.

Internet of Things (IoT) has become an emerging technology in the twenty-first century. Using this technology everyday, devices can be connected to the internet using embedded devices. Using this technology one can improve security, reduce cost and create a sustainable environment in corporate offices, homes, etc. This will provide the best possible environment for employees to produce best environment to work in. In this project, three important problems are solved that are safety, security and seamless monitoring. GSM interface provides a vital role in connecting different components enabling communication between sensors and databases, which is based on the SS7 protocol stack. The database used here is Arduino Cloud. All IoT device communications to the cloud use the industry standard SSL protocol for encryption providing a secure GSM interface. In addition to security purposes for our Smart Cabin, we are using two technologies to ensure our security. Radio-frequency identification and keypad lock are used for opening/closing and ensures who enters and leaves the Smart Cabin. In this paper, an implementation of a smart office is done where the security module consists of a keypad lock and an RFID, environment monitoring system consisting of a temperature and humidity monitor and an air quality monitor and display. This implementation also consists of a smoke and fire alarm that sends SMS in presence of fire.

Computer Vision is one of the branches of computer science. It will detect and understand the images and scenes. This work suggests a real-time, immediate motion tracking system for devices that follows an object's attitude in space as represented by its 3D bounding box. Computer Vision includes different features such as image recognition, image production, object detection, high-resolution image processing etc.… Object detection is frequently utilized in self-driving cars, security systems, facial recognition, pedestrian counts, and online photos. The most accurate acquisition algorithms and techniques are used in this research. This covers the precision of each identifying technique. Images can contain objects that can be automatically located and recognized. One of the core issues with computer vision is object detection. This work will show that the most cutting-edge approach to object detection at the moment is R-convolutional neural networks. This is the major objective is to examine and evaluate convolutional object identification techniques. The main idea behind such system is to classify various images and to classify object’s position approximately in all the images in order to give a full information about the images and videos. The system will be able to detect, localize and classify several objects using given image or videos. It is very difficult to classify images into different classes.

This paper exploits some of the Industry standard applications for trading and gives new insights on few ML models, namely LSTM, ARIMA, SVR. Techniques like Bollinger Bands and Support and Resistance are implemented to help in understanding the stock data better. The usage of Optimal Portfolio has been demonstrated to aid in making buy/sell decisions. It also talks clearly about the 20 min window used by certain trading application when a new user tries to learn trading, and how the new users are being manipulated. It gives an alternate approach to paper trading with no real money involved and the most appropriate ML models that can be used for short term prediction through commodity hardware.

The main cause of death worldwide, which has a significant negative influence on the vast majority of people, is liver disease. Various factors that affect the liver are the cause of this disease. For instance, alcoholism, obesity, and untreated hepatitis. The cost and complexity of this disease's diagnosis are enormous. The proposed work aims to reduce the high cost of liver disease diagnosis through detection by comparing the efficiency of machine learning (ML) and deep learning (DL) algorithms. In the proposed study, variety of algorithms have been employed that include Convolution Neural Network (CNN), Artificial Neural Network (ANN), Gaussian Naive Bayes (GNB), Random Forest (RF), and Logistic Regression (LR). The effectiveness of the performance is assessed using a variety of metrics that include accuracy, precision, recall, F-1 score, train time, and test time. Proposed work primarily focuses on the use of medical data for the detection of disease related to liver. It has been noted that the proposed work performed better than state-of-the-art work.

Measurement of room temperature electrical conductivity of metals (both porous and composite metals) has been a subject of interest for researchers since it is useful in determining some important mechanical properties that are critical in industry especially aerospace structures. The most widely accepted method of D.C room temperature electrical conductivity is the Vander Pauw method that requires extremely sensitive ammeter capable of measuring currents less than a micro ampere. This paper reports the results of experiments with discs of varied materials of known conductivity using the rotating disc type energy meter and relate the conductivity with the rotational speed of the disc. Previous efforts have shown that replacement of discs inside the slot is difficult, and discs of different diameter are not possible. This is achieved by using a small Servo motor in place of energy meter. Discs of varied materials are fabricated, and the rotation of disc is accomplished by attaching them on to the shaft of the motor. Replacement of discs is made possible by providing a threading arrangement in the motor shaft. Experiments are conducted by measuring the change in rotating speed (retardation) of the disc. The work reported is a first step toward obtaining a calibration curve that can relate the electrical conductivity with a significant experimentally measurable parameter like slip. This novel technique is expected to lead toward the development of a low-cost device and technology to assess the electrical conductivity of metals and composite metals including materials with pores.

In this paper, an enhanced classification model using Naive Bayes with various similarity measures is proposed. In this work, the Bayes formula is used to vectorize a text document such that the vector gives the probability distribution of a document with respect to each predetermined probable category to which the document may belong to. The probability distribution may assign the document to any of the topics, for example, the ones that are found in ‘mini-20-newsgroup’. The proposed approach can be used on any document collections and will show an improved classification accuracy thereby competing with some of the well-known classifiers such as SVM.

There are several approaches to solving every coding algorithm in Computer Science. To achieve the same result, these methods may use various techniques and reasoning. The difficulty is that as the number of inputs increases, certain algorithms tend to perform poorly. Several metrics may be used to assess the quality of any code. The code runtime complexity is one of these measurements. To determine this runtime complexity, substantial study and a thorough understanding of algorithms are necessary, which is a challenging manual undertaking. In this study, the worst-case runtime complexity of codes in programming languages C, Java and Python are calculated as Big-O notations utilising code features like Abstract Syntax Trees, ML approaches and static code analysis. The novelty of the research is our labelled runtime complexity dataset which was constructed manually, implementing Deep Learning Algorithms like Bi-LSTM and calculating Code Runtime Complexity for the worst-case scenario as Big-O notations for codes in three languages, C, Java and Python. To predict the runtime complexity for a given code more accurately than the traditional legacy methods such as manually asserting code runtime complexity or running code for different amounts of input, we have presented a more effective manner for the same. The results portray that the XGBoost classifier outperforms the other models with an accuracy of 96%. The current study can also be extended to other high-level programming languages, including more training samples and making use of graph neural networks.

Plant ailment identification is the biggest job for farmers in recent times. Over the years, plant and leaf diseases have caused an overall yield loss of 14% globally. Misdiagnosis of plant leaf diseases can lead to misuse of chemical pesticides, leading to economic burden to farmers and also causing soil depletion. Currently, plant leaf disease detection requires experts to diagnose and recommend pesticides, which is a time-consuming and costly process. Hence, our work aims at suggesting a machine learning technique that is reasonably useful than the existing manual method. We propose a software prototype and natural insecticides to avoid banana plant infection in our proposed work. Our work makes use convolution neural network to detect diseases in sunflower and banana plants. The proposed work was able to identify sigatoka leaf spot, Panama wilt in banana leaves and leaf blight, and downy mildew in sunflower leaves. The work carried out gave an overall accuracy of 94% in identifying and classifying the disease. Organic and chemical pesticides were suggested to control the identified disease

Classification and analysis are the two major factors for the real-time automation systems. In the sector of farming, the cultivation of different paddy crops depends on the soil nature and the weather. We need to analyze the humidity level in the area to predict the type of paddy that can be cultivated. In this proposed work, a novel model of feature prediction and classification algorithm to estimate the humidity level of soil and its atmospheric temperature to analyze the type of crop that can be cultivate in the land. This is to classify the different types of rice paddy crop which is better to plant in the farming land in nature. Regression-based categorization algorithms are employed in this procedure to examine the temperature and moisture of the land. The dataset comprises collections of temperature and moisture readings from diverse data samples taken over a land moisture readings from diverse data samples taken over a land. The extracts of temperature and moisture from different day patterns are examined and framed as the pattern for the provided dataset using the feature analysis method. Regression-based categorization algorithms are employed in this procedure to examine the temperature and moisture content of the land. The dataset comprises collections of temperature and humidity readings from diverse data samples taken over a land. The extracts of temperature and humidity from different day patterns are examined and framed as the pattern for the provided dataset using the feature analysis method. The data pattern is then classified using a regression technique in order to forecast the class of paddy crop based on the dataset's attributes. The measurement of the data represents different varieties of paddy based on the atmosphere and other characteristics as a consequence of the categorization result. The kind of classification model aids in agricultural planting and guards against crop damage brought on by excessive heat or water. The contrast between the suggested work results and those of other cutting-edge data categorization techniques is shown in the result analysis.

Fake currency is a serious problem in India, and detecting it is crucial to maintain the integrity of the country's currency. There are various software programs available in India for detecting fake currency. These software programs use image processing techniques to analyze images of currency notes and identify any irregularities that indicate counterfeit notes. Existing methods for detecting counterfeit notes rely primarily on image processing techniques. A web-based detection system provides an easy-to-use interface for users to upload images of currency notes, send them to the CNN model (Venkata Raghu et al. in Int J Creat Res Thoughts (IJCRT) 10:2320–2882, 2004), and view the analysis results. Convolutional Neural Networks (CNNs) (Pallavi et al. in Int Res J Modernizat Eng Technol Sci 50:4076–4081, 2002) are used in the detection of counterfeit currency by analyzing the security features of currency notes and learning to differentiate between genuine and counterfeit notes based on those features. Using a real-time camera view (Selvi Rajendran and Anithaashri in IOP Conf Ser Mater Sci Eng 992:01201, 2020), this study will identify Indian banknotes by extracting features from notes, the model can identify counterfeit money. The model is trained with 80–20% training and test split, with each layer receiving the same learning rate of 0.001. After training the network for 200 cycles with 306 images, the training accuracy score 90.6%.

The market demand, consumer preferences, supply chain challenges, unstable geopolitical environment, and other business environmental factors have emerged to be highly dynamic. In this scenario, data-driven business decisions and support have become vital. The constant need for high volume, high variety, accuracy, and reliability of data has made big data imperative for businesses. The data being generated and used in abundance also brings a few challenges. This paper focuses on challenges faced by data scientists and big data analysts in using big data models in effective and efficient supply chain decisions. The data envelopment analysis (DEA) technique is used to analyze the variable and evaluate the performances of various entities engaged in numerous activities. The study’s findings indicate that variables like Sharing and Accessing Data, Privacy, Security, and Fault tolerance have proved efficient during the model deployment. Still, the factors like Analytical challenges, Quality of data, and Scalability of data have proved inefficient during the deployment of the defined big data model in the supply chain.

In the field of image processing content-based image retrieval is having research scope to retrieve the images which is stored in any location on the network. In the content-based image retrieval system to check images which are deposited on the network based on image attributes like image size, name of the image, shape, and color of the image. This work gives a novel method for system retrieve the image based on the text query given by the user. The scope of the paper also extends the scope to finding the similar combined images containing text and also graphical shapes. In the proposed system we can retrieve the images by using Optical Character Recognition system. The important scope of the proposed methodology is to produce accurate result even in the case of invariance to difficult graphics and background features, text design, and different font size. The approach has been tested for retrieving images of English text based on the provided text. We put forth a brand-new method for creating a content-based picture retrieval system based on locating images with the same textual information within an image.

For the purpose of identifying the felt emotions and intentions behind multimodal data, multimodal sentiment investigation seeks to semantic info acquired across different modalities. The primary focus of this field of study is the design of a novel fusion system that can efficiently and effectively aggregate data from several sources. However, the ability to use the independence and connection across modalities is lacking in prior work, preventing optimal performance. Consequently, the work suggests a unique approach to visual and textual sense modality and then fuses these features using different kernel learning techniques (MKL). After that, we feed the combined dataset into an Extreme Learning Machine (ELM), where we use the Vortex Search Algorithm to choose the most appropriate bias and weight (VSA). When everything is said and done, trials are run on two publicly accessible datasets, and the results are better than the current gold standard for multimodal sentiment analysis. For instance, whereas the current ML only manages 94% accuracy, the suggested ELM-VSA achieves 98.50%.

Task scheduling and resource allocation are the major issues in grid environment. Based on grid user’s requirements such as deadline, cost, and service type, tasks must be scheduled and appropriate resources are allocated for each user task. Previous works in this topic is failed to analyze all criteria for timely scheduling and resource allocation. Further, scalability and storage issues are other drawbacks in grid computing. Grid over Hadoop is a great solution for solving scalability and storage issues. When adding security to the system, we can address the traditional issues of grid computing high response and retrieval time, resource searching time, low scalability, and storage issues. In this paper, we proposed a new framework called as SecQSON which is a secure query scheduling and ontology-based searching in Map-Evaluate-Reduce model. There are three processes which are applied in this paper as authentication, scheduling, and data retrieval phases. In authentication phase, data owners (DO) and data users (DU) are authenticated by trusted authority (TA). For this purpose, Dual Bio-Key-based Random Authentication (DUBK-RA) algorithm is proposed. The security credentials are fingerprint, finger vein, ID, and password for authentication. For bio-key generation, BLAKE-3 hashing algorithm is generated in TA and this key is verified to validate whether the request is authorized or not. Then scheduling phase processes the authorized requests (DU’s) for query scheduling. In this task, Map-Evaluation-Reduce model is proposed that maps the users to optimum grid resources. The evaluation of the resources for user queries is evaluated using Spotted Hyena Optimizer algorithm. For evaluation purpose, various criteria are considered such as trust level, resource score (available bandwidth and queries), and time score (response time and execution time). Final phase is a data retrieval phase in which authorized DO’s records are stored in the grid-connected Hadoop server. In grid server, name node is processed and it constructs the index values for DO’s records by means of Dendrimer Order Statistic (Den-OS) index. Further, ontology is constructed for records stored in data nodes. Grid resources for user queries are dynamically searched, and the optimum search results are retrieved for grid users. Experiments are conducted and the performance is evaluated using several metrics such as response time, search accuracy, retrieval time, authentication time, latency, energy consumption, precision, recall, and f-measure.

Texture, a crucial aspect of an image, is something made up of components that are related to one another. Reliable feature extraction in image files requires the use of a texture-based categorization method, which is significant. This study proposes an effective method for classifying textures using machine learning (ML) approaches. Using these ML classifiers, which are in the form of artificial intelligence (AI), programmers can predict results exactly without providing instructed to do so explicitly. The proposed study focuses on the creation of own dataset in the form of CSV file, to do so Haralick features (contrast, dissimilarity homogeneity, energy, and correlation) extracted from the Brodatz texture dataset. Different ML algorithms are used like: K-Nearest Neighbor, Decision Tree Classifier, Random Forest Classifier, Gradient Boosting Classifier, and AdaBoost Classifier which are experimented on the created dataset to classify the texture of Brodatz dataset. Proposed approach exhibits better results with 100% accuracy with less computation time as compared to previous work in the literature.

In recent times, social media has played a major role in shaping the state of mind of young adults. This means that the content that is interacted with online may subconsciously have a significant effect on one’s mental health. This paper presents a novel approach to detect the mental health state of a user by analyzing their online activity over a period of time and generating a report indicating the same. The user is then allowed to ask any clarifying questions on the generated report, along with general queries on mental health via a chat-bot interface. In order to classify the mental state of the user based on the kind of content posted or interacted with by them, their tweets are scraped and feature vectors of the same are generated. Supervised machine learning algorithms like Support Vector Machines (SVM) and neural network-based models like Long Short-Term Memory (LSTM) are compared for their performance on prediction. A transfer learning approach is also attempted and gives promising results in predicting the classes of the tweets. Natural Language Processing techniques such as question similarity and extractive summarization are utilized in building the chat-bot framework.

The plant diseases have a direct impact on the quality and quantity of the crop, and by diagnosing them, the market value of agricultural products increases. This exemplifies the significance of healthy plants as well as the relevance of early identification of disease on leaves. The early detection has the difficulty of manpower’s’ inadequate knowledge on usage of sophisticated technology for plant disease detection. To automate this time-consuming process, this work proposes to build a device that takes pictures as input and detects damaged leaves while classifying the plant condition. Based on the disease detected, suitable pesticide is suggested from the database and the automatic spraying in the affected part is carried out. In this article, diseases in tomato and potato plants using image processing and machine learning techniques (convolution neural network) are used to automate disease identification and automated method for suitable pesticide spraying which is implemented. The identification of the disease has been implemented with 96% accuracy. The hardware implementation has the wheels controlled by 12 V DC motor R365 connecting spraying machine nozzle. The L293D motor driver is controlled by Arduino Uno.

Now a days, news on print and digital media has become very essential for everyone to stay aware with current happenings in the world. The increased amount of IT penetration and digitization has resulted in more number of people accessing various social media offerings like YouTube, Facebook, Twitter, Yahoo, etc. With the increased usage of social media, the amount of fake news also has increased disproportionately high. Fake news may cause economic and social problems. It is very difficult to find source of fake news and to make the concerned accountable for it. The accountability of authentic news is decreasing day by day. In this paper, we are proposing a Blockchain-based platform which acts as a media to share news. The source of news and each interaction will be recorded in the Blockchain and it is immutable. Users have read access to the news and they can also give rating to news based on its authenticity. Authors/journalists have write access, so that they can post their contents to media. Because all the interactions are available on Blockchain, it is easy to trace back the source of fake news if any.

Emotion recognition has been a problem in the field of brain–computer interface. Numerous ways are available for recognizing human emotions and one such technique is through Electroencephalogram (EEG) signals. EEG signals are recordings of the subject’s electrical activity in the brain. Feature extraction approaches such as Power Spectrum Density (PSD) and Discrete Wavelet Transform (DWT) are fed as features to various machine learning (ML) and deep learning (DL) models. This work aims to develop models that predict emotions from EEG data. In addition, the results of the above-mentioned feature extraction approaches are compared in this work. The proposed feature extraction methods and models are applied on the DEAP dataset.

Hand gesture control system is the most demanding needs in the today’s world. It is one of the important means of human and computer interaction as of today. As an important part of non-verbal communication, hand gestures play an important role in our daily life. Hand gesture recognition control systems offer an innovative, natural, and user-friendly way to interact with computers. Analyzing and studying various segmentation and tracking, feature extraction, and recognition techniques, in this research paper, we have introduced an approach to recognize the hand gestures and control of different system settings and applications with the help of gestures. This paper provides the overview of systematic development of the system control using hand gesture recognition system. Hand detection model has been developed and made with the help of the computer vision library, OpenCV and Media pipe; using this hand detection system as a module, we can add specific number of gestures as different modules that will be able to control different system settings. These modules can be linked with hand detection model for controlling the system settings. The main aim of our project is to make human–computer interactions as smooth as possible such that the requirement of physical devices like keyboard, mouse, etc., as input system might not even be necessary. Therefore, overview of hand detection module and description about how system control using gestures have been implemented is described.

Pomegranate is one of the fruits which is commercially sold and is mostly cultivated in semiarid and arid regions in India. Pomegranates are one of the highly consumed fruits as they are rich in carbohydrates, calcium, vitamin-c, iron, and citric acid. And, these fruits are also used as the medicinal agent for the diseases like diabetes, cancer, hypertension, heart, and kidney-related diseases. Though there are many advantages for pomegranate, but the major issue which is affecting the production of this fruit is the disease on fruit and leaves due to climate changes and pest. Identifying this disease on fruit and leaf in early stages will improve the production and also beneficial to farmers. The most common diseases for the pomegranate fruit will be Bacterial Blight, Anthracnose, Fruit Spot, Fruit Borer, and Fusarium Wilt, and coming to the leaves, the diseases are diverse bacterial blight symptoms observed on pomegranate leaves. In this proposed research, the concept of deep neural networks with image processing will be applied in order to identify the diseases at early stages. To imply this system, the data will be collected from the field images starting from leaf stage to the production stage of the fruit starting from 25 to 60 days. There is no limitation on the number of images which are going to be captured, as many images as possible will be captured as it will be given as an input to the machine learning model. As machine learning mechanism says that the size of the input is big, the accuracy levels will be always better. Once it is loaded, same will be applied with various ML algorithms in order to find out the diseases which are affecting fruits and leaves. The proposed research is majorly to identify the diseases of pomegranate fruit and leaves at early stages.

The adoption of cloud computing has skyrocketed in a lot of companies. One of the biggest selling points of cloud computing is how cheap it is and how easily you can access your data. Since users routinely save sensitive data with cloud storage providers despite the fact that they may not be reliable, cloud computing security is a vital concern in the cloud computing ecosystem. Due to concerns around current service failure and damaging entrants in the single-cloud, consumers are predicted to shift away from engaging with “single cloud” suppliers. Multiple clouds, or “inter-clouds” as they are sometimes called, are becoming increasingly popular. In this study, we investigate the problem of securing multiple cloud environments and discuss some potential approaches to the problem. Researchers have paid less effort to studying how to use several cloud providers while keeping data secure than they have to studying individual clouds. Multi-clouds can reduce vulnerability to some types of attacks, hence this initiative aims to increase their adoption. However, there are still major worries about the security and trustworthiness of cloud-based data storage. We present DEPSKY, a system that encrypts, encodes, and replicates data over numerous clouds to increase availability, integrity, and confidentially stored in the cloud. Our infrastructure, together with Planet Lab, has been deployed to four commercial clouds, where users from all around the world can use the service. Our protocols improved perceived availability and, for most cases, access latency when compared to those of individual cloud providers.

In the year 1950, the concept of patch antenna was initiated. Improvement in the field of patch antenna was observed and it is employed in Printed Circuit Board. Extensive applications originated in antenna domain due to its compact size, cost-effective, low profile etc., The major drawback of patch antenna is that it has narrow bandwidth. However, the problem can be reduced by using slots, different feed techniques and substrates. Slots can be in different shapes and feed technique must be integrated depending on the design and applications. This study intend at the sights of slot in Rectangular Microstrip Patch Antenna (RMPA) to improve the parameters of simulated and fabricated design. In this paper, rectangular shaped patch antenna with U-slot, U-I slot structure, co-axial feed, Rogers RO4350 substrate, and ground plane are presented. The U-slot and U-I slot Rectangular Microstrip Patch Antenna are operated at resonant frequency of 2.45 GHz and it has desired resonant input impedance. The main advantage of co-axial feed technique is that feed can be placed in any desired location. Rogers RO4350 substrate has better return loss and the main advantage is low manufacturing cost. The projected U-slot antenna is operated at the frequencies between 2.35 GHz and2.55 GHz. The gain and VSWR values are 4.41 dB and 1.03 which have been achieved in the simulation using HFSS 2017. The projected antenna offers a maximum Bandwidth (BW) of 200 MHz and Return Loss of − 34.41 dB. The projected U-I slot antenna is operated at the frequencies between 2.35 GHz to 2.56 GHz. The gain and VSWR values are 4.55 dB and 1.03 which have been achieved in the simulation using HFSS 2017. It offers a maximum Bandwidth (BW) of 210 MHz and Return Loss of − 36.22 dB. The proposed work provides a noble agreement connecting the simulated and fabricated results. From the results of U-I slot patch antenna, it can be observed that 20% wider Bandwidth; 3% increment in Gain, improved Return Loss of 4%, and 2% VSWR are obtained compared to U-slot patch antenna.

The Breast cancer is the a type of a carcinogenic cells that is developed in the breast and is more commonly found in women. This type of cancer has been found more fatal amongst women after the lung cancer. The paper proposes a study where it helps in identifying the breast cancer at faster rate using Convolution Neural Network (CNN). The paper uses a technique of identifies and differentiates the mammography pictures into three types, first one to be benign, second one is malignant, and the third one to be normal. This technique helps the medical professional in identifying the type of cancer. The CNN model used uses VGG-16 architecture for cancer detection and classification. The proposed system uses huge dataset of 275,000 RGB image patches. The proposed prediction is a deep learning technique where it’s architecture has two stages, first is image processing and the second is image classification. The result of using this technique reaches the accuracy level of 95.72% in classifying the different types of cancer.

Search engine algorithms work by taking the important elements of a web page which includes the title of the page, the content of the page, and keyword density. By using these key elements, search engines use their proprietary algorithms to come up with a ranking for each web page such that users can get most relevant results. Each search engine’s algorithm is unique, so a top ranking on Yahoo! does not guarantee a prominent ranking on Google, and vice versa. The aim of this study is to develop a search engine mainly for handling programming related queries based on Evolutionary Learning. This paper also focuses on feature snippet as an add on for displaying source code for relevant code search. The proposed system works on a plethora of domains including machine learning and big data analytics for improving the search results and refining them.

For a very long time, people have fantasized about making devices that can peer inside another person's mind and communicate with technology just by thought. These ideas have captivated people's imaginations in both present and ancient mythology. However, the progress of cognitive neuroscience and brain imaging technologies has only recently made it possible for people to connect with the human brain. Proponents have used these approaches to create brain-machine interfaces (BMI), which are methods of interaction independent of the greater utilization of the brain's deductive networks of peripheral muscles and nerves. The expanding societal understanding of the needs of those with physical impairments was the main driving force behind this research. Users in these systems just track their cognitive function rather than producing signals with their muscles which might be utilized to drive machines or communications systems. The importance of this research is immense, specifically for those who have endured catastrophic neuromuscular injury issues and neurological conditions like amyotrophic lateral sclerosis, which gradually deprive them of their ability to navigate independently while maintaining cognitive performance.