Volume 14, Issue 3, August 2023
Reinforcement Learning in Financial Markets: A Study on Dynamic Model Weight Assignment |
Pages: 1-8 (8) | [Full Text] PDF (394K) |
Akash Deep |
Department of Graduate Studies, Texas Tech Univer- sity Lubbock, TX, 79409, USA |
Abstract - This study innovates in financial market forecasting by employing reinforcement learning within a diverse ensemble of models, including Random Forest Regression, LSTM networks, linear regression, and sentiment analysis. Through dynamic, performance-based weight adjustments, our approach shows marked improvement over static strategies. Real-world data testing evidences enhanced prediction accuracy, hinting at potentially more profitable trading decisions. This work underscores the untapped potential of reinforcement learning for optimal ensemble model management in the ever-changing financial landscape. |
Index Terms - Ensemble Reinforcement Learning, Random Forest Regression, Long Short Term Memory (LSTM), Advan- tage ActorCritic (A2C) Algorithm and Financial TimeSeries Analysis |
Application of IoT-based Terrestrial Biodiversity Data for Sustainable Livelihoods among Rural Communities in Turkana County |
Pages: 9-18 (10) | [Full Text] PDF (654K) |
Jeremiah Osida Onunga, Anselemo Ikoha Peters, Peter Edome Akwee |
Department of Information Technology, Kibabii University Department of Biological and Physical Sciences, Turkana University College |
Abstract - This study researched on the different ways of application of IoT-based terrestrial biodiversity data for sustainable livelihoods in Turkana County. This study studied the application of IoT devices and Technologies, in order to enhance sustainable livelihoods in Turkana County. The purpose of the study was to establish the application of IoT-based Terrestrial Biodiversity Data for Sustainable Livelihoods among Rural Communities in Turkana County. The study used theoretical research findings to connect IoT theory and practice, while also connecting it to biodiversity data for sustainable livelihoods. The study was a case study. Case studies are real, rely on careful research, they foster the development of multiple perspectives by users. The study method was mixed method. This method was considered appropriate for this study because the quality of information that was yielded were valid. The study was carried out in Turkana County, which is a county in the former Rift Valley Province of Kenya. This study targeted 164,519 households who were the main respondents. The sample population for this research was 384 households. Data was collected using questionnaires, Focus Group Discussions and Key Informant Interviews. The data was analyzed using descriptive statistics and presented in tables. The study used inferential statistics to establish statistical significant relationship between the variables. The qualitative and quantitative findings supported the idea that pastoralists who had access to and were aware of data on terrestrial biodiversity and pastoral advice provided by IoTs used the information to influence their way of life. The findings of this research emphasized the need for initiatives that allow rural populations to employ IoT technologies to fully utilize terrestrial biodiversity data processing and sharing. The study's knowledge contribution assumed the shape of an enhanced SLF, where the varied responses and systematic analysis made the IoT application and use relevant for understanding the relationship between terrestrial biodiversity data sustainable livelihoods. The findings of this study can be utilized to provide policy recommendations and suggestions that can Kenya develop its future terrestrial biodiversity data plans, policies, and strategies. |
Index Terms - Application, IoT Technologies, Terrestrial, Biodiversity, Data, Rural, Sustainable and Livelihoods |
Internet of Things in Banking Industry: A Systematic Literature Review |
Pages: 19-26 (8) | [Full Text] PDF (395K) |
Muhammad Mazhar, Muhammad Abrar, Junaid Arshad |
Department of Computer Science, University of Engineering and Technology, Lahore, Punjab, Pakistan |
Abstract - One of the most sensitive and complicated sectors of the economy, the banking sector sees considerable changes every day. Most banks and financial institutions work hard to develop and diversify their payment procedures, making them harder and more secure in order to boost their digital skill. Sympathetic customer behavior has emerged as a crucial component of a successful business strategy, so why the internet of things (IOT) may offer the furthermost effective solution toward the issue of data collection and exchange among various items through the Internet. Depend on the ideas, the purpose of this review paper is development in e-banking sector in relation to IoT and big data, in addition as the effects of large data across different financial and banking sectors, particularly in relation to banking and financial markets, internet service provider, financial institutions threat scrutiny, and other areas. In this study the conclusion of the future research prospects will be discussed because the large data context in electronic banking and finance is a brand new idea. |
Index Terms - IoT in Finance, IoT Application, Data Management, Finance and Big Data |
Comparative Analysis of Different Machine Learning Algorithms to Classify Wanted and Unwanted IoT Traffic |
Pages: 27-33 (7) | [Full Text] PDF (434K) |
Wajid Hussain, Syed Muqadir Hussain Shah, Muhammad Kafayat |
Department of Computer Science, University of Engineering and Technology, Lahore, Pakistan |
Abstract - The Internet of Things (IoT) is becoming increasingly popular, and as more and more devices are connected to the Internet, it becomes more difficult to distinguish between wanted and unwanted traffic. Machine learning (ML) algorithms have been use to classify IOT traffic, but little research has been done on the comparative analysis of different ML algorithms to classify IoT traffic. In this research paper, we present a comparative analysis of different ML algorithms to classify IoT traffic into wanted and unwanted traffic. We evaluated different ML algorithms, including logistic regression, decision trees, random forests, support vector machines, neural networks etc. We used precision, recall, TP, FP and accuracy as performance metrics to compare the effectiveness of the algorithms. Our results show that the neural network algorithm outperformed the other algorithms in terms of all performance metrics. The decision tree and random forest algorithms also performed well, but not as well as the neural network algorithm. The Logistic Regression and Support Vector Machine Algorithm had the lowest performance. Overall, our research shows that the neural network algorithm is the most effective in classifying IOT traffic into wanted and unwanted traffic. These results can be useful for designing better intrusion detection systems for IoT networks and improving the security of IoT devices. |
Index Terms - Machine Learning, IoT, Comparative Analysis and Intrusion Detection |