International Journal of Advances in Applied Computational Intelligence
Journal DOI
2833-5600ISSN (Online)
Stock is a financial product considered by flexible trading, high risk, and high return that can preferred by several investors. Investors may get an abundance of returns through the accurate prediction of stock price trends. Nevertheless, the stock price can be influenced by certain factors including market conditions, companies’ managerial decisions, macroeconomic situation, and investors’ preferences for major economic and social events. Econometric and Statistical models are widely utilized in classical stock price prediction; however, these techniques could not handle the complex and dynamic environments of the stock market. Researchers have begun using deep learning (DL) and machine learning (ML) to estimate stock fluctuations and prices with the rapid evolution of artificial intelligence (AI), serving investors to define investment strategies to increase returns and decrease risk. Therefore, this manuscript presents a new dung beetle optimization with deep learning based stock price prediction (DBODL-SPP) methodology. The purpose of the DBODL-SPP algorithm is to predict the rise or fall of stock prices using the optimal DL model. In the DBODL-SPP technique, the min-max scalar can be deployed for pre-processing the input data. Besides, the DBODL-SPP approach applies the DBO algorithm for electing an optimal subset of features. The DBODL-SPP technique makes use of a multi-head attention long short-term memory (MHA-LSTM) model for the stock price prediction. Finally, by using the equilibrium optimizer (EO) algorithm, the parameter tuning of the MHA-LSTM algorithm can be carried out. A detailed set of experimentations has been applied to evaluate the enriched performance of the DBODL-SPP technique. The simulation values emphasized that the DBODL-SPP algorithm achieves better results than other techniques for stock price prediction
Read MoreDoi: https://doi.org/10.54216/IJAACI.050201
Vol. 7 Issue. 1 PP. 08-23, (2024)
The research focuses on an accurate workload prediction approach for auto-scaling resources in the Private Cloud using improved Time-Series models. Although many factors still result in dynamic workloads of cloud systems, an accurate forecast becomes vital for service quality and cost. The chapter discusses a Proactive Prediction Engine (PPE) framework using Auto Regressive Integrated Moving Average (ARIMA) and Recurrent Neural Network Long Short-Term, to forecast CPU utilization. Real-time datasets of OpenStack private cloud and Amazon AWS were used for experimental evaluation. The analyses show that the RNN_LSTM model performs far better than ARIMA by reducing the MAE and RMSE values by roughly 40 percent in each set. This has further reinforced that RNN_LSTM can model non-linearity and handle correlation issues in the workload data. Automated scaling of the instances with the Open Stack based on the predicted CPU load is made possible by the integration of RNN_LSTM prediction with OpenStack, supported by Terraform. This strategy reduces times of service outages and enables the efficient use of resources in the network. Regarding accuracy and automation, the proposed method can be a relevant solution for workload management for private cloud infrastructure. In this respect, the results support the implementation of deep learning-based predictive models to optimize the performance of autoscaling.
Read MoreDoi: https://doi.org/10.54216/IJAACI.070105
Vol. 7 Issue. 1 PP. 63-77, (2025)
A Decision Support System (DSS) for the recognition of mitotic nuclei (MN) on the histopathological image (HI) aids pathologists in cancer diagnoses by automating the MN detection, a key indicator of tumor proliferation and cell division. Leveraging innovative image processing and machine learning (ML) algorithms, such a system can accurately detect MN, which are crucial indicators of cell division and tumor proliferation. By automating these processes, pathologists can focus more on complicated diagnostic tasks while ensuring efficient and consistent analysis. ML approaches, comprising support vector machines (SVMs) or convolutional neural networks (CNNs) can be widely applied for the classification task. These techniques learn from annotated data to accurately discriminate between mitotic and non-MN. Incorporating these technologies into pathology workflow facilitates research efforts in oncology for improved treatment strategies, enhances diagnostic accuracy, and reduces variability among observers. This study presents an Optimal Bayesian Neural Network based Decision Support System for Mitotic Nuclei Detection (OBNN-DSSMND) technique on Histopathologic Imaging. The goal of the OBNN-DSSMND technique is to detect the mitotic and non-mitotic cells on the HIs. In the initial phase, the OBNN-DSSMND technique undergoes the bilateral filtering (BF) technique to preprocess the input images. Next, the OBNN-DSSMND technique involves a feature fusion process encompassing SqueezeNet, DenseNet, and VGG-19 models. Meanwhile, the hyperparameter selection of the DL models is performed by using the Archimedes Optimization algorithm (AOA). For mitotic nuclei detection, the OBNN-DSSMND technique applies a BNN classifier, which recognizes the presence of mitotic and non-mitotic cells on the HIs. The experimental assessment of the OBNN-DSSMND approach was examined utilizing a benchmark image dataset. The widespread simulation analysis reported that the OBNN-DSSMND technique achieves better results than other techniques.
Read MoreDoi: https://doi.org/10.54216/IJAACI.070101
Vol. 7 Issue. 1 PP. 01-16, (2025)
The importance of business procedures and web services in facilitating effective and dynamic company operations is highlighted in this section as it delves into their construction and integration. Web services are defined by their reuse and seamless integration, and they communicate and integrate using standard like XML, WSDL, UDDI, and SOAP. The importance of web service composing is emphasized throughout the section. This technique involves combining many services to handle complicated tasks and improve performance. Static (design-time), dynamic (runtime) composing approaches, together with orchestrating, and the choreography, are the main categories in the field. Using state-of-the-art methods such as BPEL (Business Process Execution Language), Petri nets, and AI-based methods, the method of composition entails three critical phases: identifying services, selection, and scheduling. To demonstrate how to deal with dependency issues, mistakes, and optimizing, this section also discusses scheduling difficulties by combining Hierarchical Task Networks (HTN) with Partial Order Planning (POP). Being compliant with QoS (Quality of Service) standards is supported by dynamically services selection, which also facilitates strong, automatic business processes. Web services have the ability to streamline Business-to-Business (B2B) interactions, improve agility, and save costs, as highlighted in this section. Companies may improve the quality of products, speed delivery, and provide individualized services by automating workflows and using dynamically composition. The study suggests cutting-edge mathematical techniques to boost performance and shows how to put them to use in practical situations. Comparing the two methods at one service, the Proposed Method completes the work in 0.16 seconds, which is 98.67% quicker than the Conventional Method's 0.3 seconds are. Because it yields quicker responses without sacrificing efficiency, the Proposed Method is more accurate. With an increase in time for execution accuracy, the suggested technique is more effective and faster at one service.
Read MoreDoi: https://doi.org/10.54216/IJAACI.070102
Vol. 7 Issue. 1 PP. 17-33, (2025)
Web services are a crucial part of large-scale software development and cross-organizational collaboration. This chapter discusses the challenges of selecting the finest internet services among the vast array of possibilities available, with an emphasis on quality of service (QoS) features. Web services must fulfil every requirement needed to provide optimal user experience and the efficient execution of corporate operations. In order to find the best services, we look at important quality of service characteristics including response speed, reliability, accessibility, and efficiency. In what follows, you will find a detailed method for selecting services. The approach consists of three steps: finding services, improving them according to QoS constraints, and grading those using weighted normalized techniques. At each stage, methods are provided to ensure an accurate and successful selection that meets the customer's needs. The proposed method seems to work, according to the results of the trials. The rating of services for several customers with varying limits, achieved using real-life data sets, demonstrates the approach of filtering and assessing to acquire optimal results. This method boosts the efficiency and usefulness of the selected services by combining functional and non-functional aspects. Finally, this part concludes by stressing the importance of quality of service in guaranteeing customer satisfaction and optimizing the delivery of services in competitive and fast-changing environments. Service 3 has the highest accuracy rate at 96.5%. Due to their low reaction times and high availability, Services 2 and 6 are in close second place. Services 4 and 7 have good availability ratings; however, they take longer to respond. Services 1 and 8 have moderate availability and high response times; hence, they get the lowest scores. When it comes to reliability and accuracy, Service 3 remains your most effective choice.
Read MoreDoi: https://doi.org/10.54216/IJAACI.070103
Vol. 7 Issue. 1 PP. 34-49, (2025)
Recently, the combination of Deep Learning (DL) methods within the Internet of Things (IoTs) has developed in the agricultural field, especially in the domain of pest management. This study considers the implementation and development of an innovative method for Insect Detection and Classification using DL within the environment of the IoTs in agriculture. The developed system advantages advanced DL approaches for analysing images captured by IoT-enabled devices, enabling real-time identification and categorization of insect pests. By continuously incorporating these technologies, these research goals to increase the efficiency and precision of pest monitoring, finally providing to sustainable agricultural technologies and increased crop yield. This study presents an Automated Insect Detection and Classification using Pelican Optimization Algorithm with Deep Learning (AIDC-POADL) technique on Internet of Enabled Agricultural Sector. The main objective of the AIDC-POADL system is to identify and categorize various types of insects exist in the agricultural field. In the primary stage, the AIDC-POADL technique involves DenseNet-121 model to learn complex features in the input images. Also, the hyperparameter choice of the DenseNet-121 algorithm developed by the POA. At last, multilayer perceptron (MLP) model can be applied to discriminate the insects into various classes. To validate the enhanced performance of the AIDC-POADL algorithm, a series of simulations are involved. The experimental outcomes stated that the AIDC-POADL technique offers enhanced recognition results over other approaches.
Read MoreDoi: https://doi.org/10.54216/IJAACI.070104
Vol. 7 Issue. 1 PP. 50-62, (2025)