ABSTRACT Understanding financial behaviour, particularly in the stock market, has attracted significant interest in recent years due to advancements in artificial intelligence and its impact on the global economy. The field of stock market prediction, which explores the interaction between finance and computer science to create predictive models, aims to forecast the behaviour of various securities in the financial market. One of the most well‐known and widely used techniques is Deep Learning, which employs different deep neural network structures for learning nonlinear models. In this study, we used open data from some of the largest companies in Brazil—Petrobras (PETR4), Itaúsa (ITSA4), and Vale (VALE3)—provided by BovDB, a historical dataset containing the stock prices of all companies listed on the Brazilian stock exchange (B3) from 2000 to 2020. As part of the preprocessing, a price correction factor was applied to neutralise the effects of market events on stock behaviour, enabling the recurrent neural network (RNN) model to process this information better. The results showed that using this correction factor significantly improves predictions, reducing abrupt behaviours in stock prices and decreasing the model's prediction error. For instance, the prediction error in VALE3 stock was reduced by < 10% compared with uncorrected data. These findings highlight the potential of using an event correction factor in stock data processed by an RNN, facilitating its training and providing more reliable forecasts.

Managing big data traffic plays an important role in contemporary communication and is essential for efficiently handling an unprecedented volume of information. In the globalized context of the internet, the ability to measure and predict this traffic is a strategically valuable resource that requires a deep understanding of historical data. This article proposes a predictor based on a generalization of the Choquet integral, which aggregates data, reducing the complexity and dimensionality of traffic predictions. The approach is assessed using real data, demonstrating that the Choquet integral achieves higher accuracy with the appropriate$\alpha$parameter. Considering the worst-case scenario in terms of wasted time, and given that the overall algorithm achieved a satisfactory error rate, we can conclude that the least efficient algorithm was the brute force search. In comparison, binary search and random binary search demonstrated a time efficiency improvement of 56.75 % and 59.49 %, respectively. Among the integrals evaluated, the Choquet (a) integral yielded the smallest errors.

Rapid technological advancement has revolutionized the acquisition, processing, and storage of personal data, with notable data breaches from significant corporations emphasizing the value of data and the need for enhanced privacy protection. This has led to a global focus on individual privacy by enacting privacy-centric laws. The healthcare sector, known for its data sensitivity, presents distinct challenges necessitating stringent privacy protocols. The healthcare sector, known for its data sensitivity, presents distinct challenges necessitating stringent privacy protocols. Thus, there is a critical need for robust data privacy measures, including (pseudo)anonymization, to address this shift. This paper introduces a tailored multilevel (pseudo)anonymization architecture designed for healthcare data, capable of ensuring secure data handling and precise anonymization across various sources, even in a (pseudo)anonymized state. The proposed architecture was developed as a proof of concept and underwent thorough evaluation through a series of experiments. The outcomes are encouraging by showcasing effectiveness in achieving accurate anonymization, secure data linkage, and supporting re-identification when essential for individual security.

Network traffic engineering aims at the network quality, optimizing routes and detecting network attacks. In this context, traffic prediction is an essential tool to capture the underlying behavior of a network. Therefore, this work proposes a modularization architecture for volumetric prediction models, allowing switching between models and setups at runtime in controllers of Software Defined Networks (SDN), dealing with short time series and delivering the data already processed for the prediction. The proposed architecture compares the results from four traditional predictors based on short-range time dependency.

Industrial Internet-of-Things (IoT) massively deploys intelligent computing in industrial production and manufacturing environments seeking automation, reliability, and control. Machine Learning models provide intelligent decisions to drive manufacturing systems to the next level of productivity, efficiency, and safety. One of the critical challenges that must be faced is the deployment of Machine Learning models at the network edge to detect data anomalies caused by Industrial IoT hardware failures, since industrial IoT devices are prone to errors and failures. These anomalies can harm the industrial IoT system by producing false alarms, consuming network resources, and affecting productivity. Because of that, it is critical to rely on low latency and high precision detection systems to verify the data received from industrial IoT devices. In light of this, we assessed key performance indicators of five machine learning models running at edge computing, to provide in-depth discussions. The performance results were obtained from an oil refinery scenario using a real industrial IoT dataset. The performance was measured in terms of (a) Accuracy, (b) Precision, (c) Recall, (d) F1 score, (e) Training time, and (f) Response time.

Internet-of-Things for Health (IoTH) is an emerging application of Internet-of-Things which is changing the traditional health system in a profound manner. IoTH environments must detect individual hardware error, malfunctions, or tampering attacks, since these occurrences result in erroneous alarms that undermine the capacity to respond to true positive risk situations. Besides, devices running malfunctions and tampered nodes can lose their capability to forward the network traffic, which can cause loss of communication. For health monitoring, traffic loss can be critical, since it can delay the medical support. To address these problems, this paper proposes Resilient Information for E-health Reporting (RIER), which is designed to provide a set of local decisions before starting to monitor an event in the IoTH environment. It is able to detect true positive events and also classifying them as critical or non-critical. Data is double-communicated via two instances of IPv6 Routing Protocol for Low Power and Lossy Networks (RPL) in order to successfully deliver the notifications of critical events even if some node in the path is compromised. The conducted simulations demonstrate that RIER can achieve more than 90% of notification delivery rate.

Cloud computing is the latest trend in business for providing software, platforms and services over the Internet. However, a widespread adoption of this paradigm has been hampered by the lack of security mechanisms. In view of this, the aim of this work is to propose a new approach for detecting anomalies in cloud network traffic. The anomaly detection mechanism works on the basis of a Support Vector Machine (SVM). The key requirement for improving the accuracy of the SVM model, in the context of cloud, is to reduce the total amount of data. In light of this, we put forward the Poisson Moving Average predictor which is the core of the feature extraction approach and is able to handle the vast amount of information generated over time. In addition, two case studies are employed to validate the effectiveness of the mechanism on the basis of real datasets. Compared with other approaches, our solution exhibits the best performance in terms of detection and false alarm rates.

There is an intuitive notion that the costs associated with project support actions, currently deemed too high and increasing, are directly related to the effort spent during their development and test phases. Despite the importance of systematically characterizing and understanding this relationship, little has been done in this realm mainly due to the lack of proper tooling for both sharing information between IT project phases and learning from past experiences. To tackle this issue, in this paper we propose a solution that, leveraging existing IT project lifecycle data, is able to predict support costs. The solution has been evaluated through a case study based on the ISBSG dataset, producing correct estimates for more than 80% of the assessed scenarios.

The reuse of knowledge acquired by operators to diagnose failures in Information Technology (IT) infrastructures has potential to decrease the recurrence of failures and, consequently, reduce possible losses and maintenance costs. Nevertheless, existing solutions to support failure diagnosis lack of flexibility to adapt to a constantly changing IT environment. As a result, diagnostic is performed in an ad hoc and static fashion, which hampers the reuse of knowledge to solve similar failures affecting different elements of an IT infrastructure. To bridge this gap, in this paper we propose an extension of Common Information Model (CIM), supported by a conceptual solution for the identification of the root causes of problems, adaptable to changes in the target infrastructure and applicable to similar failures. Experiments carried out considering typical failures during the deployment of IT changes provide evidence about the efficacy of the proposed solution.

Planning is a fundamental sub-process of the overarching Information Technology (IT) change management process, proposed by the Information Technology Infrastructure Library to help organizations to deploy and maintain IT services in an effective and efficient way. A major issue behind IT change planning and of special importance for the alignment of changes with business objectives/constraints - the adequate projection of which human resources to assign to change activities - has not been properly addressed in previous investigations. To fill this gap, in this paper we propose and analyze novel strategies for planning the assignment of human resources to change activities. These strategies explore different ways to prioritize humans to activities (i.e., from the most to the less efficient or proficient humans), and to rank/cluster the activities that should be analyzed first. The novel strategies have been experimentally evaluated through ChangeAdvisor, a prototypical implementation of a decision support system that helps IT administrators in the task of understanding the trade-offs between alternative change designs.

Information Technology (IT) products and services provided by modern organizations are designed in projects that often involve large amount of resources (e.g., humans, hardware, and software). It is essential that organizations enforce rational practices for project management, in order to successfully conclude projects and avoid waste of substantial resources. In this context, Risk Management is fundamental to guarantee the accomplishment of project's objectives by dealing with adverse and favorable events. Although important, risk assessment in IT projects is usually performed by stakeholders in interviews and brainstorms which may be a very time/resource-consuming task. Therefore, in this paper, we introduce a solution to automate the risk assessment process, based on the history of previously conducted projects. Furthermore, comprehensive and interactive risk reports are proposed in order to ease the analysis of automatically generated reports. The results show that our solution is not only useful to speed the risk assessment process, but also to assist the decision making of project managers by organizing risk information according to the project structure.