Recent Advances in Computer Science and Communications - Volume 17, Issue 2, 2024

Plant diseases are one of the major contributors to economic loss in the agriculture industry worldwide. Detection of disease at early stages can help in the reduction of this loss. In recent times, a lot of emphasis has been done on disease detection due to the overall increase in production as well as the loss of grape number. With deep learning, having a promising future and having the advantages of automatic learning and feature extraction, the use of these techniques has now been widely spread. This paper reviewed the existing deep-learning techniques available for grape disease detection. Firstly, covering the various steps in a grape disease detection model ranging from the various sources of image acquisition, the different image augmentation techniques and the various models used, and the parameters required to evaluate. Secondly, the study summarizes the important findings of all literature available on the theme. The paper also tries to highlight the various challenges faced by the researchers and the common trend among them, so that future research on the topic can achieve higher performance.

Background: The concept of pill reminders has been discussed and developed throughout the decade. It varies from cascaded plastic pill boxes to complicated robust dispensers. This proposed smart pill reminding system based on IoT is being designed by considering ease-to-use and cost-effectiveness. Method: A smart pill reminding system is a system that will alert the patient to take their respective pill at the desired time. It will also track the motion of the patient’s hand while taking the pill and will also display the pill count on an LCD Screen. In case a patient forgets/ignores the reminder provided by the system, the system will automatically display the status on the application that will be installed in the relative/caretaker’s phone and through an email on the patient's relative/ caretaker’s email address to take subsequent action. The system will monitor the real-time using an RTC module, and as and when the current time matches the medicines time, it will activate its mechanism, and the patient will have a buffer time to take their medicine. In case a patient does take the medicine in the buffer time provided by the system, then one mechanism of the system will be activated. In another case, if a patient does not take the medicine in the stipulated time, further actions will be initiated by the system to benefit the patient. Results: It was tested and found that out of ten times, the system worked accurately nine times, with calculated accuracy as high as 90%. Initially, the Blynk application will display “Welcome Patient” and “You will be updated”. Once the RTC matches the scheduled time to take medicine, the buzzer starts buzzing. If the IR sensor detects the movement of the user’s hand, the LCD will update the pill count, and the pill count is reduced by one. The LCD will also display the message “Medicine Taken”. If the IR sensor does not detect the movement of the user’s hand, the LCD will display the same pill count. The LCD will also display the message “Med1 not Taken”. The Blynk will also be updated and will display the same messages. The green LED shows the status of the consumption of the pill. The email will sent to the caretaker in this case. Conclusion: In this work, all the problems related to the system are overcome in a systematic manner with the help of IoT and basic electronic applications. Thus, the system will help the user take all the prescribed medicines on time and will also update the user’s caretaker via application and email services. This will largely help the working class as well as the senior citizens. Thus, this proposed system can be commercialised as a handy and cost-effective device.

Background: Ontology matching provides a solution to the semantic heterogeneity problem by finding semantic relationships between entities of ontologies. Over the last two decades, there has been considerable development and improvement in the ontology matching paradigm. More than 50 ontology matching systems have been developed, and some of them are performing really well. However, the initial rate of improvement was measurably high, which now is slowing down. However, there still is room for improvement, which we as a community can work towards to achieve. Method: In this light, we have developed a Large Scale Ontology Matching System (LSMatch), which uses different matchers to find similarities between concepts of two ontologies. LSMatch mainly uses two modules for matching. These modules perform string similarity and synonyms matching on the concepts of the ontologies. Results: For the evaluation of LSMatch, we have tested it in Ontology Alignment Evaluation Initiative (OAEI) 2021. The performance results show that LSMatch can perform matching operations on large ontologies. LSMatch was evaluated on anatomy, disease and phenotype, conference, Knowledge graph, and Common Knowledge Graphs (KG) track. In all of these tracks, LSMatch’s performance was at par with other systems. Conclusion: Being LSMatch’s first participation, the system showed potential and has room for improvement.

This journal paper examines the transformative role of Artificial Intelligence (AI) and Machine Learning (ML) in shaping human psychology. It investigates how cognitive processes, emotional states, and social interactions are impacted by AI and ML technology. The use of AI and ML in psychology is covered in this study, covering social behaviour analysis, emotion identification, mental health assessment, and personalised therapies. It also explores the moral issues and prospective effects of AI and ML in comprehending and influencing human psychology. This paper emphasises the enormous influence of AI and ML on the comprehension and research of human psychology through a thorough analysis of pertinent literature and empirical evidence. This paper seeks to offer a thorough explanation of the profound effects that AI and ML have had on psychology. We will offer insight into the possible advantages, difficulties, and ethical issues that occur when integrating AI and ML into the study of human psychology by looking at recent developments and implementations of these technologies in psychological research. We will also look at how other areas of psychology, such as cognitive psychology, clinical psychology, social psychology, and neurology, have been impacted by AI and ML.

The advent of powerful tools like ChatGPT has opened up exciting new possibilities for augmenting law enforcement capabilities, elevating its efficiency and expanding its capacity. The AI-driven capabilities of ChatGPT can be harnessed to usher in novel applications in diverse areas such as language translation, customer service, content generation, and even law enforcement. In the realm of law enforcement, ChatGPT's potential is boundless, ranging from the generation of detailed reports and the transcription and translation of vital documents to the profiling of criminals and the automation of legal documents. However, the deployment of such technology in law enforcement raises critical issues of privacy, safeguarding fundamental rights, and the perpetuation of societal biases. Furthermore, the deployment of emerging technologies like ChatGPT could also open up new avenues for committing cybercrimes. Therefore, it is crucial to address these issues and establish regulatory frameworks that govern the use of such advanced technologies. In this paper, we provide legal and technical recommendations to regulate the use of ChatGPT and mitigate the potential risks associated with its usage, including the perpetration of various cybercrimes.

In the field of meteorology, temperature forecasting is a significant task as it has been a key factor in industrial, agricultural, renewable energy, and other sectors. High accuracy in temperature forecasting is needed for decision-making in advance. Since temperature varies over time and has been studied to have non-trivial long-range correlation, non-linear behavior, and seasonal variability, it is important to implement an appropriate methodology to forecast accurately. In this paper, we have reviewed the performance of statistical approaches such as AR and ARIMA with RNN, LSTM, GRU, and LSTM-RNN Deep Learning models. The models were tested for short-term temperature forecasting for a period of 48 hours. Among the statistical models, the AR model showed notable performance with a r² score of 0.955 for triennial 1 and for the same, the Deep Learning models also performed nearly equal to that of the statistical models and thus hybrid LSTM-RNN model was tested. The hybrid model obtained the highest r² score of 0.960. The difference in RMSE, MAE and r² scores are not significantly different for both Statistical and Vanilla Deep Learning approaches. However, the hybrid model provided a better r² score, and LIME explanations have been generated for the same in order to understand the dependencies over a point forecast. Based on the reviewed results, it can be concluded that for short-term forecasting, both Statistical and Deep Learning models perform nearly equally.

This study conducts a comprehensive review of Deep Learning-based approaches for accurate object segmentation and detection in high-resolution imagery captured by Unmanned Aerial Vehicles (UAVs). The methodology employs three different existing algorithms tailored to detect roads, buildings, trees, and water bodies. These algorithms include Res-UNet for roads and buildings, DeepForest for trees, and WaterDetect for water bodies. To evaluate the effectiveness of this approach, the performance of each algorithm is compared with state-of-the-art (SOTA) models for each class. The results of the study demonstrate that the methodology outperforms SOTA models in all three classes, achieving an accuracy of 93% for roads and buildings using Res-U-Net, 95% for trees using DeepForest, and an impressive 98% for water bodies using Water Detect. The paper utilizes a Deep Learning-based approach for accurate object segmentation and detection in high-resolution UAV imagery, achieving superior performance to SOTA models, with reduced overfitting and faster training by employing three smaller models for each task.

Background: In the last decades, the development of Internet activities has been significantly accelerated, particularly with the emergence of the Internet of Things (IoT). Heterogeneous devices in the IoT can seamlessly and feasibly inter-connect with each other without human interaction. Due to this revolution, many applications have been adopted in the arena of smart healthcare, e-commerce, environmental and habitat monitoring, etc. In order to promote and facilitate people's standards of living around the world. However, these unbounded applications bring more challenges to the storage capabilities of devices, and their security and privacy preservation. Moreover, security issues suffer from weak authentication protocols. Methods: To address these issues, suitable and secure lightweight mutual authentication schemes based on Elliptic Curve Cryptography (ECC) are required for the approval of Identity Management (IDM) of devices in the IoT. In this paper, we will propose an improved mutual authentication scheme based on ECC, coupled with a relevant seminal work considered as a reference in the field. This scheme is combined with U-quark, a lightweight hash function, to guarantee the security needed in the IoT environment. Results: We will compare our amended protocol with a seminal scheme as an established reference in terms of computation cost, storage cost, and executing CPU time to demonstrate that our version can ensure the most favorable performance during the authentication process. Conclusion: Finally, our proposed mutual authentication scheme has demonstrated its effectiveness in enhancing the security of IoT devices when compared to the seminal work in the same computational environment.