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.

In today's rapidly evolving industry, new challenges constantly arise that must be addressed quickly to keep up with the growing demand. Industry 5.0 aims to eliminate ob-stacles in production lines and integrate intelligent systems into manufacturing without excluding human operators from their daily tasks. In this context, developing software for controlling and managing data-driven evidence in the production cells becomes essential when considering performance, reliability, and practical aspects, as well as the financial returns it can bring to organizations. This work presents a brief overview of such an approach, proposing the development of a software architecture that manages industrial quality inspection in automotive parts using an Internet of Things publisher/subscriber model for controlling and synchronizing sub-processes. The results obtained demonstrate the benefits of implementing a control and evidence management system, offering production operators online monitoring of failures in manufactured parts and effective visualization of evidence of failures. This helps in decision-making and integrates the operator (user) with the system, which aligns with Industry 5.0 principles. This contribution is supported by creating a database of quality inspection evidence that can be used to train machine learning models to detect failures more accurately in the future.

This paper describes a digital-twin-based approach to dimensional control designed to assist quality inspection of automotive air-conditioning rotors in the vehicle manufacturing industry. It focuses on dimensional quality control within the field of dimensional metrology, aiming to accurately measure the dimensions of parts, subassemblies, and complete equipment systems. The proposed approach has created a prototype, collecting measurement data from a test sample to generate validation results for error profiling. The main goal is to develop a method for improving the precision of sensors used for this type of quality inspection.

This article reviews computer vision technologies for detecting and tracking objects in industrial cargo handling activities. We have proposed a dataset and a methodology for identifying people, containers, cages, equipment, boxes, and piping, in real-time operation. Our experimental results demonstrate that our artificial neural network model effectively detects and segments objects in non-sparse environments using an annotated industrial image dataset, achieving average precision up to 95% for most classes, including 93% of test instances. This improved perception capability enhances operators' decision-making and accident prevention.

The Choquet integral (CI) is an averaging aggregation function that has been used, e.g., in the fuzzy reasoning method (FRM) of fuzzy rule-based classification systems (FRBCSs) and in multicriteria decision making in order to take into account the interactions among data/criteria. Several generalizations of the CI have been proposed in the literature in order to improve the performance of FRBCSs and also to provide more flexibility in the different models by relaxing both the monotonicity requirement and averaging conditions of aggregation functions. An important generalization is the$C_{F}$-integrals, which are preaggregation functions that may present interesting nonaveraging behavior depending on the function$F$adopted in the construction and, in this case, offering competitive results in classification. Recently, the concept of d-Choquet integrals was introduced as a generalization of the CI by restricted dissimilarity functions (RDFs), improving the usability of CIs, as when comparing inputs by the usual difference may not be viable. The objective of this article is to introduce the concept of$dC_{F}$-integrals, which is a generalization of$C_{F}$-integrals by RDFs. The aim is to analyze whether the usage of$dC_{F}$-integrals in the FRM of FRBCSs represents a good alternative toward the standard$C_{F}$-integrals that just consider the difference as a dissimilarity measure. For that, we consider six RDFs combined with five fuzzy measures, applied with more than 20 functions$F$. The analysis of the results is based on statistical tests, demonstrating their efficiency. Additionally, comparing the applicability of$dC_{F}$-integrals versus$C_{F}$-integrals, the range of the good generalizations of the former is much larger than that of the latter.

Several generalizations of the Choquet integral have been applied in the Fuzzy Reasoning Method (FRM) of Fuzzy Rule-Based Classification Systems (FRBCS’s) to improve its performance. Additionally, to achieve that goal, researchers have searched for new ways to provide more flexibility to those generalizations, by restricting the requirements of the functions being used in their constructions and relaxing the monotonicity of the integral. This is the case of CT-integrals, CC-integrals, CF-integrals, CF1F2-integrals and dCF-integrals, which obtained good performance in classification algorithms, more specifically, in the fuzzy association rule-based classification method for high-dimensional problems (FARC-HD). Thereafter, with the introduction of Choquet integrals based on restricted dissimilarity functions (RDFs) in place of the standard difference, a new generalization was made possible: the d-XChoquet (d-XC) integrals, which are ordered directional increasing functions and, depending on the adopted RDF, may also be a pre-aggregation function. Those integrals were applied in multi-criteria decision making problems and also in a motor-imagery brain computer interface framework. In the present paper, we introduce a new FRM based on the d-XC integral family, analyzing its performance by applying it to 33 different datasets from the literature.

Restricted dissimilarity functions (RDFs) were introduced to overcome problems resulting from the adoption of the standard difference. Based on those RDFs, Bustinceet al.introduced a generalization of the Choquet integral (CI), called d-Choquet integral, where the authors replaced standard differences with RDFs, providing interesting theoretical results. Motivated by such worthy properties, joint with the excellent performance in applications of other generalizations of the CI (using its expanded form, mainly), this article introduces a generalization of the expanded form of the standard Choquet integral (X-CI) based on RDFs, which we named d-XC integrals. We present not only relevant theoretical results but also two examples of applications. We apply d-XC integrals in two problems in decision making, namely a supplier selection problem (which is a multicriteria decision-making problem) and a classification problem in signal processing, based on motor-imagery brain-computer interface (MI-BCI). We found that two d-XC integrals provided better results when compared to the original CI in the supplier selection problem. Besides that, one of the d-XC integrals performed better than any previous MI-BCI results obtained with this framework in the considered signal processing problem.

This work aims to evaluate explainable classification methods for the detection of fish species from hydroacoustic data acquired by echo sounders at a region near the coastline of south and southeastern Brazil. Decision trees and fuzzy rule-based methods were adopted. The fitted models were evaluated by quality measures based on the performance of the classifiers and also by an expert which analyzed the usefulness of the rules on describing the schools. The models learned by the algorithms performed well for the available data and were able to represent the documented behavior of the species considered in the studied region, according to the literature.

In Multi-Criteria Decision Making, one of the most used algorithm designed to deal with decision making is the Technical Order by Preference to Ideal Solution (TOPSIS), which is based on finding a solution that is close to the best possible solution and distant from the worst possible solution. The Group Modular Choquet Random TOPSIS (GMC-RTOPSIS) is a generalization of the TOPSIS method capable of dealing with multiple and heterogeneous data types and interaction among criteria by means of the discrete Choquet integral. On the other hand, CT-integrals are a generalization of the Choquet integral using t-norms, which are more flexible than the standard Choquet integral. CT-integrals are pre-aggregation functions, which means that we do not require them to be monotonic in the whole domain, just in some specific directions, that is, they are directionally monotonic. Due to the excellent performance of CT-integrals in classification and multimodal fuzzy fusion decision problems, the objective of this paper is to generalize the GMC-RTOPSIS by using CT-integrals and to analyze the results provided by the use of five different t-norms in an example of a decision making problem.

It is common sense the relevance of overlap and grouping functions, specially in applications in which associativity is not required. In this work, based on previous investigations concerning classes of implication functions derived from overlap functions O, grouping functions G and fuzzy negations N (namely, (G, N)-implications, RO-implications and QL-implications constructed from (O, G, N)), a deep study on D-implications constructed from grouping functions G is carried out. Such fuzzy implications, also known as Dishkant implications, are obtained from D-operations derived from (O, G, N), which are used for the generalization of the implication p → q ≡ q(¬p∧¬q) of orthomodular lattices. We investigate under which conditions such D-operations are fuzzy implication functions, presenting a general form for obtaining D-implications. We also provide a comparative study of D-implications derived from grouping functions and other classes of fuzzy implications constructed from N, O and G, analysing the intersections among such classes.

Fuzzy Rule Based Classification Systems are an useful tool to deal with classification problems. In these systems, one fundamental point is the manner of how the available information about the problem is aggregated. The mechanism responsible to perform the aggregation is the Fuzzy Reasoning Method (FRM). Recently, some FRMs using generalizations of the Choquet integral to perform the aggregation were proposed in the literature. Since these generalizations are defined considering a specific fuzzy measure, in this paper we apply different fuzzy measures in the generalizations that presented the best performance in each study in the literature. We analyze how the performance is affected according to each fuzzy measure.

In recent years several data classification techniques have been proposed.However, it is not a trivial task to choose the most appropriate classifier for deal with a particular problem and set it up properly.In addition, there is no optimal algorithm to solve all prediction problems.In order to improve the result of the classification process, the stacking strategy combines the knowledge acquired by individual learning algorithms aiming to discover new patterns not yet identified.Stacking combines the outputs of base classifiers, induced by several learning algorithms using the same dataset, by means of a meta-classifier.The main goal of this paper is to evaluate the effects of classifier diversity on the accuracy of stacking.We have performed a lot of experiments which results show the impact of multiple diversity measures on the gain of stacking, considering many real datasets extracted from UCI machine learning repository and three synthetic twodimensional datasets.The results revealed connections between some measures and the gain of stacking, but they imply a weak or moderate relationship that suggest predicting the improvement on the best base classifier accuracy using diversity measures is inappropriate.