The rapid rise in popularity of social media sites and blogs, has prompted reviews and comments on daily activities. Sentiment analysis is the process of gathering and analysing opinions, thoughts, and impressions about subjects, goods, and services. SA produces accurate predictions in business and politics. Millions of bytes of data are uploaded and exchanged by people on product review sites, blogs, social networking sites and other online forums within the Internet realm that penetrates nearly every human being on our globe. Despite the numerous approaches for SA which is proposed, there is still a scope for effective methods which can reliably extract and produce correct sentiment analysis. Although there has been significant advancements made with ML techniques like SVM and Naïve Bayes, Maximum Entropy is the most notable method. This paper aims to contribute a comprehensive assessment report that can serve as a foundation for future research by surveying SA approaches and methodology.
The rapid growth of the Internet of Things has ledto an unprecedented increase in the volume of data generatedand shared across various devices and platforms. Consequently, the surge in data generation and sharing has heightened theneed for robust security measures, as these interconnected IoT devices are often susceptible to a multitude of cyber threats, ranging from unauthorized access and data breaches to malicioustampering and disruption of critical services. This paper presentsIoTChain, an advanced decentralized framework for secure storage and data sharing in Internet of Things (IoT) ecosys-tems. The proposed model ensures end-to-end encryption and precise, fine-grained access control through a role-based access control (R-BAC) system. IoTChain incorporates the Ethereum blockchain as a trusted, decentralized access management layer, leveraging smart contracts to guarantee secure and transparent access rights. By integrating Ethereum with the InterPlanetary File System (IPFS), the model provides decentralized, scalable, and fault-tolerant storage. Security is reinforced using Advanced Encryption Standard (AES) for data encryption, while secure keyexchange is achieved through the elliptic curve Diffie-Hellman (ECDH) protocol. To optimize efficiency, the traditional proof- of-work (PoW) consensus mechanism is replaced with proof- of-authority (PoA), significantly reducing transaction costs and boosting throughput.
Amidst the dynamic landscape of agriculture, the integration of robotic technologies emerges as a transformative force. This paper unveils a robotic plant health monitoring and intervention system, integrating machine learning and networking, tailored to revolutionize automation of agricultural practices. The centralized network server, in the system, utilizes Python scripts for socket communication with robots, as well as for implementing algorithms for optimized robot selection and path planning. Surveillance robots, equipped with ESP32-S with camera and Wroom modules navigate around the field and capture real-time images. The server on receiving these images does crop disease detection using Convolutional Neural Networks (CNNs), achieving high accuracy. Spray bots then target affected crops and spray pesticides and fertilizers as necessary using a 5V pump with precision. Unlike the surveillance robots, the spray bots are equipped with only ESP32 Wroom which controls both the movement and spray action. The system also features a software implementation that illustrates tasks like robot selection, networked communication, surveillance, and pesticide application autonomously executed in an agricultural field, highlighting the potential for scalable field deployment. The integrated solution ensures timely interventions, enhances yield quality, reduces environmental impact, and demonstrates potential of the concept - Internet of Robots in driving sustainable agricultural practices. Future iterations aim for full autonomy to further optimize system operation.
Improving road safety and streamlining traffic represent critical goals achievable through effective lane change prediction in intricate traffic situations. To achieve this, the paper delves into forecasting lane shift decisions across diverse and challenging traffic scenarios, comparing and contrasting various machine learning methods. By harnessing visual data capturing crucial environmental parameters such as weather conditions, traffic signals, and impediments like cars and buses, the research endeavours to develop prediction algorithms capable of identifying optimal lane change maneuvers. The accuracy and dependability of five well-known machine learning algorithms, including Random Forest, Logistic Regression, Adaboost, Gradient Boosting, and Bagging Classifier, are meticulously evaluated in predicting lane change occurrences. These models undergo rigorous training and assessment using real-world traffic statistics, providing a comprehensive understanding of their behavior in various contexts. Through this comparative analysis, insights into the strengths and weaknesses of each paradigm are gleaned, offering valuable contributions to the advancement of intelligent transportation systems.
Hardware Trojan (HT) detection using test pattern generation is very difficult because HTs are intended to be activated by infrequent state of logic at internal circuit nodes. The functionality of these types of circuits is stealthily and can cause severe disruptions even if it is set off. Applying appropriate test vectors is necessary to detect their presence. However, the test generation task requires a vast search space to be explored. Considering all these, Nature Inspired Algorithms (NIAs) are a promising option, as they can effectively explore this search space with test vectors. This paper aims to introduce a test pattern generation methodology using Binary Learning Differential Evolution (BLDE) and the Firefly Algorithm (FA), which are very effective in achieving success. Attacking any arbitrary node (payload) of the circuit cannot be done without a specific internal trigger condition. Logic malfunctions, detected by test vectors at feasible payload nodes is selected later based on their compactness. The validity of the proposed approach can be verified by performing several ISCAS '85 benchmark circuits. Our proposed method gives higher Trojan coverage than that of the previously reported logic testing based techniques.
