Journal of the Brazilian Computer Society

Evaluating Methods for Violence Classification and Firearm Detection in Indoor CCTV Environment

2024-04-02T20:55:49+00:00

The adoption of security systems based on computer vision for violence detection has the potential to significantly improve safety in various public and private properties. However, developing these systems can be extremely challenging.We can choose to use classification models to identify violence in images or also use object detection models to identify firearms, which may indicate robbery. Additionally, when developing such systems focused on private environments, we encounter specific challenges, such as obtaining appropriate datasets to train the algorithms. Many publicly available datasets for violence detection consist of outdoor images, with elements such as streets and cars, which do not adequately reflect the nuances and unique characteristics of private properties. In this work, we evaluate both learned and handcrafted features to classify videos as 'violence' or 'non-violence' across a variety of datasets, including a new dataset composed exclusively of closed-circuit television (CCTV) images. Additionally, we propose a new dataset for firearm detection in CCTV images and conduct some experiments using YoloV8. In this way, we hope to provide a clearer insight into the possible decisions when developing a security system for indoor environments.

Analyzing equitable access in a remote socio-enactive setting: A case study

2024-04-15T13:16:15+00:00

Ubiquitous Computing as proposed by Mark Weiser in 1991 creates a huge diversity of interaction possibilities through, e.g., sensors, actuators, physiological data, wearables, to name but a few. Providing equitable access to everyone in these unpredictable interaction environments is already a challenge. Additionally, the Covid-19 pandemic brought new challenges to these scenarios by adding remote interactions. Ubiquitous computing remote environments should provide equitable access to everyone, including people with disabilities. To better understand such issues and contribute to their solution, we studied and evaluated access in the Aquarela Virtual workshop using the UbiAccess evaluation instrument. This workshop offered a remote ubiquitous computing environment within the context of a kindergarten school. We used thematic analysis to understand the interaction needs of an Autistic Spectrum Disorder (ASD) child during the workshop. This paper extends previous research with additional data and its analysis - in particular, concerning information provided by the teacher responsible for the specific classroom, and data on the ASD child contributing to: 1) Reflections about remote ubiquitous environments and equitable access; 2) Benefits of ubiquitous computing interactions as regards the ASD child; and 3) Coverage of the UbiAccess evaluation instrument.

An Information Theoretic Learning Artificial Immune Network for Alternative Clustering

2024-06-17T14:39:22+00:00

Clustering is an unsupervised task employed when there is no prior knowledge about the structure and information contained in the data. Nowadays the amount of information and the dimensionality of data increased. Due to this, several datasets contain samples that can be clustered in different ways, presenting different partitions. Classical algorithms tend to obtain a single partition per execution and also require information like the number of clusters. Immuno-inspired algorithms were developed to reduce some of these drawbacks. They can find alternative solutions without knowing the number of clusters, but high dimensionality reduces their performance leading to low convergence rates. Information Theoretic Learning (ITL) uses statistical information of the data regardless of prior knowledge of the structure of these data and the dimensionality involved. Applied in several papers for clustering, ITL-based algorithms tend to present good performance for this task. This paper presents an immuno-inspired ITL-based algorithm (ITL-aiNet) capable of finding and maintaining high-quality and diverse solutions for datasets regardless of their dimensionality and structure. Real-world image and document datasets of varying dimensions were used in the experiments, allowing different ways of clustering. The results were evaluated using external indices. The proposed approach was capable of maintaining high-quality and diverse solutions, compared to other strategies found in the literature. The indices used to measure the quality and diversity of solutions indicated that the algorithm is capable of finding and maintaining good solutions. Solutions that have greater diversity than other algorithms in some datasets and higher quality in others.

Advancing Electric Engineering Education through Immersive Virtual Reality: Deep Learning and Evolutionary Algorithms for Image Stitching and Rectification in Virtual Lab Environments

2024-04-24T13:48:30+00:00

Virtual Reality (VR) technology has emerged as a transformative tool in education, offering immersive and interactive experiences that enhance learning outcomes. This paper delves into the application of image stitching and rectification techniques to create a VR lab environment, specifically tailored for electrical engineering education. The importance of VR technology in education is explored, highlighting its role in promoting active learning and providing experiential learning opportunities. The primary emphasis of this Paper lies in the smooth incorporation of image stitching algorithms for the creation of panoramic perspectives, along with the implementation of rectification techniques to correct irregular borders within the stitched images. By utilizing Convolutional Neural Networks (CNNs) and Genetic Algorithms (GAs), the proposed approach optimizes the rectification process, resulting in visually cohesive representations. Demonstrating the utilization of the VR lab across a range of situations, such as examining power transfer and creating control panels for water pumps in irrigation initiatives, the immersive setting enables students to delve into intricate systems. The performance of the proposed method was evaluated using various metrics, including mean squared error, peak signal to noise ratio (PSNR), structural similarity index (SSIM), and Fréchet inception distance (FID). the combination of deep learning algorithm specifically (CNN) and optimization algorithm specifically (Genetic algorithm (GA)) led to an increase in the accuracy of the rectified images where the average PSNR reached 23.98, SSIM was 0.8066, and FID was 18.72. Regarding the users’ opinion about the generated environment by stitching and rectifying images, participants demonstrated consistent positive sentiments, with mean scores ranging from 3.65 to 4.03, all above the scale midpoint, and moderate variability indicated by standard deviation values ranging from 1.070 to 1.251, suggesting general favorability with some variation in responses. This experience empowers the users to gain insights and cultivate essential problemsolving abilities at a heightened level. Collaborative learning is facilitated, enabling students to engage in collaborative projects regardless of their physical location. Through the synthesis of image processing techniques and VR technology, this research contributes to the enrichment of educational experiences and the advancement of electrical engineering education.

On the use of Change History Data to Enhance Class Change-Proneness Prediction Models

2024-07-30T15:09:51+00:00

As software evolves, new artifacts are created, modified, or removed. One of these main artifacts generated in the development of object-oriented software is the class. Classes have a very dynamic life cycle that can result in additional costs to the project. One way to mitigate this is to detect, in the early stages of the development, classes that are prone to change. Some approaches in the literature adopt Machine Learning (ML) algorithms to predict the change-proneness of a class. However, most of these approaches do not consider the temporal dependency between training instances, i.e., they consider that the instances are independent. To overcome such a limitation, this study presents an approach for predicting change-proneness based on the class change history. The approach adopts the sliding window method and is evaluated to obtain six kinds of models, which are derived by using, as predictors, different sets of metrics: structural, evolutionary, and smell-based. The evaluation uses five systems, four ML algorithms, and also explores some resample techniques to deal with imbalanced data. Regardless of the kind of model analyzed and the algorithm used, our approach overcomes the traditional one in 378 (~80) cases, out of 420, considering all systems, kinds of models, indicators, and algorithms. Moreover, the results show that our approach presents the best performance when the set of evolutionary metrics is used as predictors. There is no improvement when smell-based metrics are added. The Random Forest algorithm with the resampling technique ADA reaches the best performance among the ML algorithms evaluated.

A benchmark for Portuguese zero-shot stance detection

2024-06-24T12:55:30+00:00

Stance detection is the task of inferring for/against attitudes towards a particular target from text. As targets are in principle unlimited, however, research in the field has moved from so-called in-domain classification (which assume the availability of a sufficient number of stances towards the intended target for training purposes) to more realistic zero-shot scenarios. However, regardless of which - or how much - training data is taken into account, most existing zero-shot approaches are devoted to the English language, in stark opposition to alternatives devoted to Portuguese. As a means to overcome some of these difficulties, this article presents a benchmark (hereby understood as the combination of a dataset, baseline systems and their results) for zero-shot Portuguese stance detection that is, to the best of our knowledge, the first of it kind. More specifically, we adapt a number of existing models available for the English language to Portuguese, and introduce novel approaches to the task based on more recent prompt engineering methods and off-task labelling, achieving SOTA results that are, in some cases, even superior to in-domain classification.

Exploring Normalization for High Convergence on Federated Learning for Drones

2024-07-08T20:14:39+00:00

The usage of mobile devices like drones has been increasing in various fields, ranging from package delivery to emergency services and environmental monitoring. Intelligent services increasingly use the processing power of these devices in conjunction with techniques such as Federated Learning (FL), which allows machine learning to be carried out in a decentralized way using data accessed by clients or devices. However, in normal operations, the data accessed by clients is distributed heterogeneously among themselves, negatively impacting learning results. This article discusses the normalization in Federated Learning local training to mitigate results obtained in heterogeneous distributions. In this context, we propose Federated Learning with Weight Standardization on Convolutional Neural Networks (FedWS) and evaluate it with Batch Normalization, Layer Normalization, and Group Normalization in experiments with heterogeneous data distributions. The experiments demonstrated that FedWS achieved higher accuracy results ranging from 3% to 6% and reduced the computational and communication costs between 25% and 40%, being more suitable for use in devices with computational resource limitations.

BWBEV: A Bitwise Query Processing Algorithm for Approximate Prefix Search

2024-08-06T13:21:58+00:00

We tackle the challenge of conducting an approximate prefix search within datasets of strings. We explore using a bit-parallelism technique to compute the edit distance between distinct strings and illustrate its adaptation for an approximate prefix search procedure referred to as BWBEV. This technique employs a unary representation of edit vectors alongside bitwise operations to efficiently update these vectors during the edit distance computation. We also show how to apply our new bit-parallelism technique strategy to online edit distance computation between strings without index structure. Our experiments with BWBEV applied to approximate prefix search for a query autocompletion task revealed a substantial acceleration of over 36% when contrasted against state-of-the-art methods.

Potential Adverse Effects Caused by Gamification Elements in mHealth Apps

2024-07-30T21:55:19+00:00

Health practices using mobile devices (known as mobile health, or mHealth) have increased in recent years. To increase users’ motivation to use it, mHealth researchers and developers have started to use gamification. However, no studies were found on the negative aspects of gamification in health. This work aims to determine the potential adverse effects of gamification in mHealth applications and analyze how different types of gamification elements can cause these effects. To this end, an analysis of 70 gamified mHealth applications extracted from the Google Play Store was performed to identify which gamification elements are commonly applied in mHealth. Then, to identify the negative aspects of gamification in mHealth and how the gamification elements can be related to them, a systematic review of the literature was carried out, where 37 studies were selected. Finally, a study was carried out with users of gamified mHealth applications to analyze in practice what was found in the previous steps. The results of these studies show which gamification elements are the most used in mHealth applications, the potential adverse effects found, and a list of which gamification elements can cause these effects. The results of this study provide insights into the risks associated with gamification in mHealth applications and can make developers, designers and professionals aware of these harmful issues.

Exploring the Influence of Software Evolution on Mobile App Accessibility: Insights from User Reviews

2024-08-22T18:39:10+00:00

Software systems constantly evolve to accommodate stakeholders' and environments' requirements that change over time. In this process, the frequent modifications can increase software complexity and negatively impact its global quality when conducted in an unstructured way. The evolutive nature of mobile environments led researchers to investigate how mobile app evolution impacts complexity, security, resource consumption, maintainability, usability, and accessibility. In particular, there has been limited research on the impact of app updates on mobile accessibility: most studies focused on tracking the number of accessibility violations found by automated tools across successive versions of a small set of applications. In a previous work, we made a contribution to this field by identifying accessibility reviews associated with app updates and prompting ChatGPT-4 to provide an overview of the main accessibility issues and enhancements perceived by users in the new releases of a mobile app. In this manuscript, we extend our previous work by adopting manual content analysis to delve deeper into our research questions and by adding new research questions associated with the identification of reviews linked to app updates, user characteristics, WCAG principles and guidelines, and user demands reported in accessibility reviews. Our results show that the accessibility barriers reported by users are mostly linked to the WCAG 2.2 Perceivable principle, and the Distinguishable and Adaptable guidelines, which includes poor color scheme, small font size, unlabeled elements, and lack of customization options. Accordingly, the consequences of the lack of accessibility is mainly connected to the difficult users experience to perceive elements of the interface (e.g. difficult to read and distinguish content, watch videos) and to use screen readers, in addition to feel discriminated against. The most common demand developers and organizations receive is to bring back some accessible feature or to fix accessibility bugs.

A Robust Client Selection Mechanism for Federated Learning Environments

2024-04-12T17:34:17+00:00

There is a exponential growth of data usage, specially due to the proliferation of connected applications with personalized models for different applications. In this context, Federated Learning (FL) emerges as a promising solution to enable collaborative model training while preserving the privacy and autonomy of participating clients. In a typical FL scenario, clients exhibit significant heterogeneity in terms of data distribution and hardware configurations. In this way, randomly sampling clients in each training round may not fully exploit the local updates from heterogeneous clients, resulting in lower model accuracy, slower convergence rate, degraded fairness, etc. In addition, malicious users could disseminate incorrect weights, which may decrease the accuracy of aggregated models and increase the time for convergence in FL. In this article, we introduce Resilience-aware Client Selection Mechanism for non-IID data and malicious clients in FL environment, called RICA. The proposed mechanism employs data size and entropy as criteria for client selection. In addition, RICA relies Centroid-Based Kernel Alignment (CKA) to identify and exclude potentially malicious clients. Our evaluation shows an improvement of 125% in Accuracy values in a scenario of malicious clients, which means the RICA+CKA demonstrates a more stable and resilient approach, reaching 90% accuracy in a few rounds compared to the default average approach, reached only around 30%. Therefore, results of the behavior of RICA+CKA in different datasets show the evaluation of different numbers of clients reaching around 90% while the other approach does not pass the 50% Accuracy.

A baropodometric analysis of postural therapy supported by immersive virtual reality

2024-09-02T17:23:33+00:00

Virtual Reality (VR) can support postural therapies through proprioceptive assistance and engaging interactive features. Plantar pressure measurements are critical for accurate and individualized diagnosis of gait and posture. In a previous work, we developed an immersive VR environment to support postural therapy and conducted a user study. This study expands on previous research and aims to investigate the impact of postural therapy supported by immersive VR technology, compared to a traditional method. Our analysis utilizes unexplored baropodometric data collected during supervised therapy sessions in our prior work. Based on our current understanding, this study represents the pioneering exploration of combining baropodometric analysis with immersive VR technology to support postural therapy. In the experiment, healthy students (n = 22) with an average age of 12 years were randomly assigned to either an experimental or a control group. The experimental group underwent the therapeutic intervention using immersive VR, while the control group followed the same protocol without VR assistance. Baropodometric data were collected before and after the sessions and analyzed using the Wilcoxon test. Both groups showed postural improvements after the intervention, particularly in reducing foot pressure imbalances between the left and right feet; however, a statistical analysis indicated that VR-assisted therapy demonstrated more noticeable improvements in baropodometric parameters regarding weight and rearfoot weight distributions, as well as maximum foot pressure. Our findings reinforce the potential of integrating VR technology and baropodometric parameters to enhance therapeutic outcomes. This combination can serve as a complementary tool in clinical and research contexts, with the aim of improving diagnosis and treatment in postural therapy.

Effects of Nonsensical Responses in Virtual Human Simulations on Clinicians’ Empathic Communication and Emotional Responses

2024-09-02T17:24:29+00:00

In this manuscript, we report on research that explores the application of virtual human patients to train clinicians on empathic communication skills. During training, clinicians received empathy scores as they interacted with two virtual humans portraying suicidal ideation, who at times provided nonsensical responses. We video-recorded clinicians' interactions with virtual humans and analyzed their facial expressions, as well as their verbal responses. In phase I of our study, we analyzed clinicians' facial expressions during three key moments: after a sensical response from a virtual human (baseline), following the last nonsensical response of the interaction, and after a sensical response that followed the last nonsensical response. In phase I, facial expressions were grouped into Negative (anger, disgust, sadness, and fear) and Positive (happiness, neutral, and surprise) facial affective behaviors. We observed that nonsensical responses from virtual humans can negatively affect clinicians' positive and negative facial affective behaviors. We found a significant increase in the percentage of time clinicians express negative facial affective behaviors immediately following nonsensical responses. In phase II, we recruited additional clinician-participants and investigated how different proportions of nonsensical responses affect clinicians' facial expressions of individual basic emotions (instead of groups of positive and negative facial expressions), as well as whether nonsensical responses moderate the association between expressions of basic emotions and empathy scores obtained by clinicians during training. We observed a statistically significant positive interaction between proportions of nonsensical responses and angry facial expressions in predicting average empathy scores. That is, the relationship between anger and empathy scores was significant at low and mean levels of nonsensical responses, but not at high levels. These results suggest that at low and mean levels nonsensical responses negatively impact clinicians' performance, hindering their ability to acquire empathy skills. We discuss the impacts of technological limitations during virtual human interactions, particularly nonsensical responses, and the importance of controlling for such issues.

Virtual Reality System Controlled by Embedded Artificial Intelligence for Supporting Phobia Treatment

2024-09-02T17:26:33+00:00

In recent years, the health area has received technological contributions that provide support for diagnostic practices, monitoring, and treatment of different disorders and diseases, mainly combining various techniques of Artificial Intelligence, Virtual Reality, and Mobile Computing. There are many challenges to integrating these technologies and providing solutions that consider the automation of processes, the simplification of interaction between professionals and patients, the low price of equipment, the individualization of use, mobility, and the use of Artificial Intelligence strategies. Aiming to overcome the limitations of two previous works, which applied technological combinations in the desensitization of stress and phobias, this work aims to develop a technological combination that integrates an autonomous and low-cost virtual environment, with multi-agent control and natural language communication support, to be used in the Treatment by Exposure in Virtual Environments - VRET in the area of Clinical Psychology, more specifically related to Anxiety Disorders. Low-cost virtual reality glasses were used, with visualization on a smartphone. The prototype, called PhobIA 3DS, is controlled by multi-agents that have modules for capturing physiological signals (heart rate); uses natural language to obtain the level of anxiety perceived by the patient; considers these two pieces of information in a Fuzzy system, which, in turn, generates a response on the calculated level of anxiety; and controls and changes the display of specific scenarios for each level of anxiety. Finally, the system was evaluated by a group of 6 experienced psychologists to verify aspects of the interface, relevance, and usability. The data obtained by the evaluation showed positive results and good prospects for using the system in real activities. As a contribution, this work created an integration of AI technologies in an ESP32 microcontroller connected to a smartphone and attached to low-cost goggles. This combination of technics opens perspectives for adopting affordable technologies in phobia treatments.

Building soft skills through a role-play based approach for Requirements Engineering remote education

2023-02-13T17:57:29+00:00

Teaching Requirements Engineering requires adopting pedagogical techniques to develop students' technical skills for identifying users' needs and designing software solutions. Additionally, since requirements engineering involves group work, students must cultivate social skills such as communication, empathy, and conflict resolution. In remote learning scenarios, developing these skills becomes more challenging due to limited interaction. To address these needs, this paper proposes adapting a project-based collaborative learning approach for remote education that combines Role-Play and Send-a-Problem learning techniques. In this approach, students collaborate on software projects in teams, assuming two roles: customers and software developers. We evaluated this approach during two cycles of Action Research, conducted remotely in two disciplines of a Software Engineering undergraduate course involving advanced and beginner students, respectively. In the advanced students' class, we observed that the methodology enhanced communication skills, analytical reasoning, conflict resolution, and empathy. To validate these results, we conducted a new study with beginning students, achieving positive outcomes despite the need for more support in team communication. As our primary contribution, we provide guidelines for implementing this collaborative learning approach online.

Mapping RDEVSNL-based Definitions of Constrained Network Models to Routed DEVS Simulation Models

2023-03-21T19:34:06+00:00

The Routed DEVS (RDEVS) formalism has been introduced recently to provide a reasonable formalization for the simulation of routing processes over Discrete Event System Specification (DEVS) models. Due to its novelty, new software tools are required to improve the Modeling and Simulation (MS) tasks related to the RDEVS formalism. This paper presents the mapping between constrained network models obtained from textual specifications of routing processes and RDEVS simulation models implemented in Java. RDEVSNL context-free grammar (previously defined) is used to support the textual specification of a routing process as a constrained network model. Such grammar is based on a metamodel that defines the syntactical elements. This metamodel is used in this paper as a middleware that allows mapping constrained network model concepts with RDEVS simulation models. From such a constrained network model template, RDEVS Java implementations are obtained. The proposal is part of a work-in-progress intended to develop MS software tools for the RDEVS formalism using well-known abstractions to get the computational models through conceptual mapping. Using these tools, modelers can specify simulation models without needing to codify any routing implementation. The main benefits are i) reduction of implementation times and ii) satisfactory simulation model correctness regarding the RDEVS formalism.

Adopting Human-data Interaction Guidelines and Participatory Practices for Supporting Inexperienced Designers in Information Visualization Applications

2023-09-27T18:07:38+00:00

Nowadays, voluminous data support may influence decision-making. People with varied profiles need to interact with data to gain valuable insights. There is a need for software tools to support the understanding and management of information to favor Human-Data Interaction (HDI) with a richer user experience. This study explores the combination of HDI design guidelines and participatory approaches to improve user experience in data interaction. We defined a design process to support the activities and adapted participatory practices to facilitate HDI design. We conducted workshops with inexperienced designers developing information visualization applications for common-sense domains. They generated and analyzed several application prototypes. Results suggest that design guidelines help generate HDI-based prototypes with a good user experience.

Advancing Chatbot Conversations: A Review of Knowledge Update Approaches

2023-07-10T11:37:52+00:00

Conversational systems like chatbots have emerged as powerful tools for automating interactive tasks traditionally confined to human involvement. Fundamental to chatbot functionality is their knowledge base, the foundation of their reasoning processes. A pivotal challenge resides in chatbots' innate incapacity to seamlessly integrate changes within their knowledge base, thereby hindering their ability to provide real-time responses. The increasing literature attention dedicated to effective knowledge base updates, which we term content update, underscores the significance of this topic. This work provides an overview of content update methodologies in the context of conversational agents. We delve into the state-of-the-art approaches for natural language understanding, such as language models and alike, which are essential for turning data into knowledge. Additionally, we discuss turning point strategies and primary resources, such as deep learning, which are crucial for supporting language models. As our principal contribution, we review and discuss the core techniques underpinning information extraction as well as knowledge base representation and update in the context of conversational agents.

Technologies to Support Adaptable Game Design: A Systematic Mapping Study

2023-10-08T20:02:37+00:00

Designing games accessible and inclusive to a broader audience requires considering social, technical, and cultural aspects that recognize the diversity of the target audience. One strategy to foster inclusion is through adaptability, allowing games to be adapted based on the needs of their users. In this context, a Systematic Mapping Study was conducted to identify and document existing studies focusing on the adaptation of digital games. Aiming at creating a structured and wide panorama of the scientific literature, this paper draws on the 5W2H Framework to answer the following research questions: (1) What game characteristics are being adapted? (2) Why is it being adapted? (3) For whom is it being adapted? (4) When does the adaptation occur? (5) Who is carrying out the adaptation? (6) What technologies are present? (7) Where are the technologies used? (8) How were the technologies evaluated? (9) What was evaluated? (10) Who evaluated? and (11) What theories and methodologies support the study? Beyond presenting the current state-of-the-art in digital game adaptation, the mapping study highlights a prevailing trend in research on adaptive systems. Notably, Patients and Health Professionals emerge as the primary target audience, although many studies did not specify their target audience. The results also indicate that the evaluation of technologies commonly employs Proof of Concept, Experiment, and Case Studies, with Patients and Students frequently considered as the target audience, albeit as secondary participants in the adaptation and evaluation processes. Finally, the findings reveal that most papers lack explicit grounding in a theoretical or methodological foundation, suggesting the need for further investigation.

A Deep Learning Model for the Assessment of the Visual Aesthetics of Mobile User Interfaces

2023-09-04T19:39:34+00:00

Visual aesthetics is one of the first aspects that users experience when looking at graphical user interfaces (GUIs), contributing to the perceived usability and credibility of a software system. It can also be an essential success factor in contexts where graphical elements play an important role in attracting users, such as choosing a mobile app from an app store. Therefore, visual aesthetics assessments are crucial in interface design, but traditional methods, involving target user representatives assessing each GUI individually, are costly and time-consuming. In this context, machine learning models have been demonstrated to be promising in automating the assessment of GUIs of web-based software systems. Yet, solutions for the assessment of mobile GUIs using machine learning are still unknown. Here we introduce a deep learning model to assess the visual aesthetics of mobile Android applications designed with App Inventor. We used a supervised learning approach to train and compare models using three different architectures. The highest performing model, a Resnet50, achieved a mean squared error of .022. The assessments of new GUIs showed an excellent correlation with human ratings (ρ = .9), and the Bland Altman plot analysis revealed 95% agreement with their labels. These results indicate the model’s effectiveness in automating the visual aesthetics assessment of GUIs of mobile apps.

Promoting Children's Participation in a Participatory Design Process in a Rural School: A new role needed?

2023-09-04T19:35:21+00:00

In Brazil, Education of Rural Areas is a model which started with social movements and became a public policy with the aim of improving participation of people of rural areas in making decisions about the model of their education. Schools in that context need to improve access to scientific and technological knowledge, but ensuring that previous values and knowledge of the students about agroecology and sustainability are considered. Even though some studies focus on digital inclusion and teacher training to use technologies in rural schools, very few address the development of digital technologies by students themselves, in this context. Participatory Design (PD) is a method often used to develop technological artifacts that could help address this gap. Of particular interest in the context of Education of Rural Areas, PD includes a valid preoccupation with power balance between designers and target users in decision-making. However, this power balance is still hard to attain, even more when design involves vulnerable groups, like children. In this sense, models and frameworks of children's participation can give a more solid theoretical framing for PD with children. In this paper, we present a theoretical model for supporting PD with children which was drawn from theories of children's participation, and refined through its application for qualitatively analysing a design process of digital educational artifacts with children from rural schools in Brazil. We highlight children’s autonomy in the creation of artifacts within a process managed by adults, which we interpret as a new role of children in PD which emerged from the educational context, that is children as artifact designers. The model proposed can be used for researchers and designers to plan or to analyse children's participation in PD interventions, helping them to employ methods that promote their autonomy and participation.

Multiobjective message scheduling for Hybrid Synchronization in Distributed Simulations

2023-10-27T14:47:40+00:00

One of the essential aspects of distributed simulations is to order events according to a causal consistency model. Traditionally, implementing causal consistency can be made using a conservative or optimistic approach. However, traditional techniques are costly in processing time to ensure causality. A promising approach to order events is a hybrid synchronization approach, where processes can change dynamically between optimistic and conservative approaches. Unfortunately, synchronizing processes running a hybrid synchronization is a complex problem. In this work, we discuss a multi-objective scheduling of hybrid synchronization messages problem. Beyond that, we propose using a scheduling algorithm to reach an equilibrium between processing and causality violations and describe how to integrate the algorithm in an existing distributed simulator. The algorithm uses two memoization phases, making the scheduling suitable for a dynamic environment. Finally, to demonstrate the feasibility of our scheduling approach, we implemented it in an existing distributed simulation architecture. Analysis based on the experiments demonstrates the behavior of the simulation regarding the number of discarding/processed messages and work performed.

An optimization-based framework for personal scheduling during pandemic events

2023-11-20T16:03:51+00:00

In recent years, companies have faced the challenge of adapting to new guidelines and strategies aimed at preventing and reducing the transmission of COVID-19 within the workplace. An essential aspect of this adaptation is effectively managing the workday schedule to minimize social contact.
This paper introduces a comprehensive optimization framework designed to automate the planning of employee schedules during pandemic events. Our framework utilizes integer linear programming to establish a set of general constraints that can accommodate various types of distancing restrictions and cater to different objective functions.
To employ the framework, a company simply needs to instantiate a subset of these constraints along with an objective function based on its specific priorities. We conducted tests on our scheduling framework within three distinct real-life companies, yielding promising results. Our approach successfully increased the number of in-person workers by 15%, all while adhering to the social distancing restrictions mandated by these companies. Furthermore, the solutions generated by our method were implemented and validated within these organizations.

Generative neural networks for providing pseudo-measurements in electric power distribution systems

2023-10-13T01:28:19+00:00

The success of automation and control functions envisioned for smart distribution networks depends on reliable real-time network supervision. This task is performed by the distribution state estimator, responsible for processing a set of measurements received by the supervisory control and data acquisition (SCADA) system. In smart grids, the advanced measurement infrastructure (AMI) allows to collect regular readings of consumer voltage and power measurements—this can complement the few measurements (coming from the SCADA system) usually available for monitoring the distribution network and benefit the state estimation process. However, due to communication bottlenecks, such measurements are available only on an hourly basis. In order to circumvent the lack of real-time measurements this paper investigates the application of different neural network models—AutoEncoder, Contractive AutoEncoder, and Variational AutoEncoder—and proposes a methodology to generate AMI pseudo-measurements to complement SCADA measurements when only the latter are available for processing. Simulations performed with a 34-bus distribution system illustrate the proposed methodology, and the results obtained confirm its potential for pseudo-measurement provision.

Recognition of Brazilian vertical traffic signs and lights from a car using Single Shot Multi box Detector

2023-12-11T10:28:51+00:00

This work presents an automated system for recognizing Brazilian vertical traffic signs and lights using artificial intelligence. The main objective of the system is to contribute to road safety by alerting drivers to potential risks such as speeding, alcohol consumption, and cell phone use, which could lead to severe accidents. The system’s core contribution lies in its ability to accurately recognize various traffic signs and lights, providing crucial warnings to drivers. To achieve this, the system utilizes a light version of the single shot multi box detector as its detection algorithm and experiments with three Mobilenet versions as base networks. The optimal Mobilenet version is selected based on a mean average precision higher than 80%, which guarantees reliable detection results. The dataset used for training and evaluation comprises images extracted from YouTube traffic videos, each annotated to create the necessary labels for training. Through this extensive experimentation, the system demonstrates its efficacy in achieving accurate and efficient detection. The results of the experiments are compared with other existing approaches and our work significantly advances the field by providing a tailored dataset, an optimized model, and also valuable insights into traffic sign and light recognition, collectively contributing to the improvement of road safety.

Catastrophic Forgetting in Deep Learning: A Comprehensive Taxonomy

2024-01-03T19:48:11+00:00

Deep Learning models have achieved remarkable performance in tasks such as image classification or generation, often surpassing human accuracy. However, they can struggle to learn new tasks and update their knowledge without access to previous data, leading to a significant loss of accuracy known as Catastrophic Forgetting (CF). This phenomenon was first observed by McCloskey and Cohen in 1989 and remains an active research topic. Incremental learning without forgetting is widely recognized as a crucial aspect in building better AI systems, as it allows models to adapt to new tasks without losing the ability to perform previously learned ones. This article surveys recent studies that tackle CF in modern Deep Learning models that use gradient descent as their learning algorithm. Although several solutions have been proposed, a definitive solution or consensus on assessing CF is yet to be established. The article provides a comprehensive review of recent solutions, proposes a taxonomy to organize them, and identifies research gaps in this area.

IoT and 5G Networks: A Discussion of SDN, NFV and Information Security

2024-01-10T20:14:56+00:00

Having an infrastructure capable of exchanging data at high speed is an efficient way to drive the evolution and development of new applications and existing services. The 5G technology has emerged as a trusted source to meet the increased demand of Internet of Things (IoT) devices connected to the network, in addition to enabling Internet connectivity at high broadband speeds. Another important feature of 5G is to allow the use of Software Defined Network (SDN) and Network Function Virtualization (NFV), mechanisms responsible for performing network configurations through software, as well as the control and management of devices using the configuration network functions or device virtualization. The concern with information security in the 5G network is increasing, as cybercriminals try to access important data that is transported over the network, since the demand for connected IoT devices will be greater, allowing for several possibilities of attacks. The understanding of possible threats and attacks is necessary, so that new measures are taken against cybercrimes presented in the 5G and IoT networks. This paper aims to elucidate some conceptions of what 5G technology is and the use of IoT in this network, contextualizing the SDN and NFV paradigms to allow the configuration of the functionality and management of the network by software. In addition, concerns are reported about possible information security attacks that can occur in 5G networks.

IoT Peritoneal Dialysis: an approach exploring remote patient monitoring

2024-02-26T13:12:59+00:00

It is estimated that 5.4 million people will undergo Renal Replacement Therapy by 2030. Peritoneal dialysis seems to be the most widespread form of home treatment for these patients, but it faces problems related to its adherence. Remote monitoring has the potential to increase treatment adherence. This work aims to design an approach that integrates: (i) a platform for the acquisition of vital signs and other parameters of a patient on peritoneal dialysis; (ii) an environment where customizable rules build Situation Science and, when necessary, send notifications to the medical team; and (iii) a signal and image visualization interface that can be accessed remotely.

LAGOON: Achieving bounded individual fairness through classification frequency equalization

2024-04-04T17:29:26+00:00

One of the main concerns about using machine learning models for classification is algorithmic discrimination. Several works define different meanings of fairness to avoid or mitigate unfair classifications against minorities. The achievement of algorithmic fairness implies modifying training data, model operation, or outputs. Hence, the fair algorithm may modify the original classification. Generally, fairness means not discriminating against a person or a group. In a utopia, a system would classify every person or minority as privileged, which may decrease the utility of classification. We define λ-fairness, a relaxation of individual fairness designed to achieve fairness while maintaining utility with configurable parameters. We also propose a post-processing method that uses frequency equalization to achieve fairness in machine learning models by generalizing the outputs into frequencies. We used this flexible approach on LAGOON, an algorithm that achieves λ-fairness using frequency equalization. For experiments, we employ three benchmarks with different contexts to evaluate the quality of our approach. We compared our results to two baselines that aim to achieve fairness and minimize utility loss.

Dependability Evaluation of a Smart Poultry Monitoring System with Disaster Recovery Mechanism

2024-04-24T01:39:21+00:00

The Internet of Things (IoT) has changed how poultry farming is carried out, offering various advantages to farmers. One notable benefit is the real-time monitoring of bird breeding tasks, ensuring the well-being of the animals. Farmers can enhance their operations through task automation by incorporating an edge server for local sensor data processing. Tasks automation enables farmers to make informed decisions, improving production efficiency, bird quality, and agribusiness profits. However, poultry farming faces challenges, with disaster recovery a critical concern. Potential events like fires, power outages, or equipment failures can significantly impact birds and production. Consequently, continuous monitoring of birds is vital, and any disruptions must be minimized to uphold system integrity. This study introduces Stochastic Petri Nets (SPN) models to evaluate the availability and reliability of an intelligent bird breeding system. The system integrates a disaster recovery solution for uninterrupted operations. Furthermore, a sensitivity analysis is conducted on the components of the smart poultry system to pinpoint the most relevant one to the system's availability in the proposed architecture. This analysis can aid system architects in developing distributed architectures, considering points of failure and recovery measures. The study results demonstrate the system's high availability and reliability, enabling farmers to make informed decisions and improve the overall productivity of their farms.

Global Localization using OpenStreetMap and Elevation Offsets

2024-05-07T19:40:21+00:00

Localization is a critical component in autonomous vehicle navigation stacks. While GNSS-only localization cannot be fully reliable and available all the time, localization based on 3D high-definition (HD) maps have to be robust to world changes, which is still a challenging issue. Added to that, in general, HD maps are expensive and difficult to construct and maintain.
In this paper, we propose a particle filter-based 2D global pose estimation method that can use the crowdsourced OpenStreetMap (OSM) API, a digital surface map, or both. The main contributions of the proposed approach are: that it is lightweight, does not require the vehicle to map the environment, does not require a GPU (can be used with low-power computing resources), is agnostic to the odometry source, and achieved relatively low position and orientation errors for this localization modality using the KITTI dataset sequences. The proposed method's implementation is open source and is available with the experimental results on our GitHub page.

When boosting gender stereotypes increases flow experience and reduces self-handicapping in gamified tutoring systems

2024-03-02T05:50:50+00:00

The threat of stereotypes affects various psychological mechanisms, including affective/subjective, cognitive, and motivational ones, and can be present in gamified online educational environments in various ways. In this study, we aimed to investigate whether gender stereotypes in gamified virtual environments could affect the flow experience, self-handicapping behavior, and performance of Brazilian students. To achieve this, we experimented with 147 participants (60 males and 87 females) who were high school and higher education students from public and private institutions in the state of Alagoas, located in the northeastern region of Brazil. We randomly allocated the participants to three distinct virtual environments: a neutral environment, a male-stereotyped environment, and a female-stereotyped environment. We introduced the stereotype threat condition when the participant was in an environment that did not correspond to their gender. In contrast, the boost condition occurred when the environment corresponded to their gender. The results of this study indicate that the presence of gender stereotypes can influence both the flow experience and the self-handicapping behavior of Brazilian students in gamified virtual environments. We observed statistically significant differences that suggest that the implementation of stereotypes can influence the relationship between variables.

An extension to Kendall’s Tau metric to evaluate dissimilarities between data series

2022-11-02T14:53:52+00:00

Data analysis is performed to examine, interpret, and extract information from data series, and it includes applying various methods and techniques to understand patterns and compare data. An approach to compare data is to use rank metrics that help identify how distinct two data series are when compared to each other according to patterns, formats, criteria, and dimensions in both data series. Among these metrics, Kendall’s Tau metric stands out, as it is robust and inexpensive, widely used in analyzing sequences and genomes, to detect errors in flash memories, and to compare distributions and top-k ranked values. However, a challenge arises when comparing lists with different lengths or when lists do not share the same elements. This happens, for example, when lists are defined by top-k elements, commonly called k-list. In this case, there is no guarantee that two k-lists share the same set of elements. Traditional metrics like Kendall’s Tau are designed to quantify differences only between shared elements in lists. Recognizing this limitation, a possible solution is to apply the metric to the shared elements of the lists. Another solution, named the generalization of Kendall’s Tau, proposed by Fagin et al., considers all elements in two lists. However, this generalization of Kendall Tau is a semi-metric, as it does not satisfy the triangular inequality. To solve this problem, we propose the Extended Kendall Tau (EKT) metric that meets all the conditions of a metric and simultaneously considers the distinct elements of the compared lists. The proposed metric was evaluated by applying conventional Kendall’s Tau and the extended Kendall’s Tau over 40 text files divided into five different languages (eight files per language). We compared KT and EKT measures within the ”same language” and across ”other language” files for the two scenarios. The results revealed that both methods could accurately identify the differences between the groups of texts of the ”same language” and ”other language”. However, the numerical results show that EKT is able to more significantly highlight the difference between groups of texts of different languages.

Extracting Features from Text Flows based on Semantic Similarity for Text Classification: an Approach Inspired by Audio Analysis

2024-03-06T12:27:34+00:00

Text classification is a mainly investigated challenge in Natural Language Processing (NLP) research. The higher performance of a classification model depends on a representation that can extract valuable information about the texts. Aiming not to lose crucial local text information, a way to represent texts is through flows, sequences of information collected from texts. This paper proposes an approach that combines various techniques to represent texts: the representation by flows, the benefit of the word embeddings text representation associated with lexicon information via semantic similarity distances, and the extraction of features inspired by well-established audio analysis features.
In order to perform text classification, this approach splits the text into sentences and calculates a semantic similarity metric to a lexicon on an embedding vector space. The sequence of semantic similarity metrics composes the text flow. Then, the method performs the extraction of twenty-five features inspired by audio analysis (named Audio-Like Features). The features adaptation from audio analysis comes from a similitude between a text flow and a digital signal, in addition to the existing relationship between text, speech, and audio. We evaluated the method in three NLP classification tasks: Fake News Detection in English, Fake News Detection in Portuguese, and Newspaper Columns versus News Classification. The approach efficacy is compared to baselines that embed semantics in text representation: the Paragraph Vector and the BERT. The objective of the experiments was to investigate if the proposed approach could compete with the baselines methods improve their efficacy when associated with them. The experimental evaluation demonstrates that the association between the proposed and the baseline methods can enhance the baseline classification efficacy in all three scenarios. In the Fake News Detection in Portuguese task, our approach surpassed the baselines and obtained the best effectiveness (PR-AUC = 0.98).

Evaluating Thresholds for Object-Oriented Software Metrics

2024-05-20T15:50:28+00:00

Software metrics measure quantifiable or countable software characteristics. Researchers may apply them to provide better product understanding, evaluate the process effectiveness, and improve the software quality. A threshold is a value that aids the proper interpretation of software measurements; it indicates whether or not a given value represents a quality risk. Thresholds are unknown for most software metrics, inhibiting their use in a software quality assessment process. In a previous paper, we proposed a catalog with 18 object-oriented software metrics thresholds, providing a preliminary case study in proprietary software to validate them. This article evaluates these thresholds more deeply, considering significant aspects. We show a new example of threshold derivation, discussing it qualitatively. We explain these software metrics and discuss their threshold values, presenting each one’s application level, definition, formula, and implications for the software design. We conduct a study with two software systems to evaluate the capacity of our thresholds to identify software quality enhancement after a restructuring process. We assess these thresholds using two case studies, comparing the evaluation provided by the thresholds with the qualitative analysis given by manual inspections. The study results indicate that the thresholds may lead to few false-positive and false-negative occurrences, i.e., the thresholds provide a proper quantitative assessment of software quality. This study contributes with empirical evidence that the metrics’ thresholds proposed in our previous work provide a proper interpretation of software metrics and, hence, may aid the application of software metrics in practice.

Mining Comparative Opinions in Portuguese: A Lexicon-based Approach

2023-12-04T20:42:01+00:00

The constant expansion of e-commerce, recently boosted due to the coronavirus pandemic, has led to a massive increase in online shopping, made by increasingly demanding customers, who seek comments and reviews on the Web to assist in decision-making regarding the purchase of products. In these reviews, part of the opinions found are comparisons, which contrast aspects expressing a preference for an object over others. However, this information is neglected by traditional sentiment analysis techniques and it is not applicable for comparisons, since they do not directly express positive or negative sentiment. In this context, despite efforts in the English language, almost no studies have been done to develop appropriate solutions that allow the analysis of comparisons in the Portuguese language. This work presented one of the first studies on comparative opinion in Portuguese. Four main contributions are (1) A hierarchical approach for detecting comparative opinions, which consists of an initial binary step, which subdivides the regular opinions from the comparatives, to further categorize the comparatives into the five opinion groups: (1) Non-Comparative; (2) Non-Equal Gradable; (3) Equative, (4) Superlative; and (5) Non-Gradable. The results are promising, reaching 87% of Macro-F1 and 0.94 of AUC (Compute Area Under the Curve) for the binary step, and 61% of Macro-F1 in multiple classes; (2) An lexicon algorithm to detect the entity expressed as preferred in comparative sentences, reaching 94% of Macro-F1 for Superlative; (3) Two new datasets with approximately 5,000 comparative and non-comparative sentences in Portuguese; and (4) a lexicon with words and expressions frequently used to make comparisons in the Portuguese language.

The Legal Handling of Digital Accessibility: a Comparison of Evaluation and Policy Approaches in Federal-Level Cases in Brazil and the United States

2023-11-10T17:07:23+00:00

Countries in the Global South still have less well-established procedures and resources to enact public policies and surveillance of digital accessibility. This paper presents a comparative study on the legal handling of digital accessibility in one country in the Global North and one in the Global South – the United States and Brazil. The study analyzed federal-level procedures handled by government surveillance bodies concerning public and private organizations, involving 26 cases in Brazil and 124 in the United States. The results showed that most cases in Brazil were finalized based on automated accessibility evaluations and with little evidence of long-term actions for sustained accessibility. All cases analyzed from the United States resulted in settlement agreements leading to medium to long-term initiatives, which covered continuous verification of conformance to accessibility standards and, in many cases, demanded inspections by accessibility specialists and tests with users with disabilities. The paper discusses the implications of including effective accessibility evaluation methods and long-term solutions in legal cases.

VANETs’ research overview updated: past, present and future

2024-04-15T20:26:00+00:00

The automotive industry has been undergoing significant changes, from electric and autonomous cars to the implementation of technologies for efficient and safe communication between vehicles. Vehicular Ad Hoc Networks (VANETs), in synergy with the Internet of Things and Artificial Intelligence technologies, contribute to increasing the safety and efficiency of land transport systems, aiding in the reduction of environmental pollution, and providing multiple applications to users. This article offers a historical perspective of research in VANETs, analyzing about 600 articles published between 2007 and 2021 in important conferences and journals. A systematic methodology was adopted for the selection and analysis of the articles, focusing on criteria such as thematic relevance, research methodologies employed, and significant contributions to the field. The most promising areas, main tools, and methodologies used in the studies were identified. We detected trends in the topics addressed and their future perspectives. Additionally, a detailed discussion on the main research problems found and a comparison with other studies were carried out, highlighting gaps and persistent methodological flaws in research. Specific issues, such as the lack of standardization in simulation methodologies and the need for more realistic approaches, are emphasized. Finally, perspectives for future research in VANETs are explored, suggesting promising directions, such as the development of enhanced security protocols, integration with emerging cloud computing technologies, and exploration of new applications in smart urban scenarios.

Brazilian Portuguese Version of Intrinsic Motivation Inventory (IMI-Teq Br): Towards a Digital Well-Being Culture in Brazil

2024-03-28T19:36:09+00:00

Self-determination theory (SDT), a foundational psychological framework, has emerged as a pivotal lens through which to understand the dynamics of human-computer interaction (HCI) and Games User Research (GUR). Central to SDT is the conceptualization of intrinsic motivation, characterized by voluntary behaviors arising fro personal interest. An established method for assessing intrinsic motivation across various contexts is the employment of the Intrinsic Motivation Inventory (IMI). This study presents a meticulous translation and cross-cultural adaptation of the 22-item version of IMI, known as the Task Evaluation Questionnaire, into Brazilian Portuguese (IMI Teq Br). The process adhered to the comprehensive methodology outlined by Beaton, encompassing Translation, Synthesis, Back-translation, Expert Committee review, and Pre-testing phases. Statistical analyses, including Student’s T-test for independent samples, Exploratory Factor Analysis (EFA), and Confirmatory Factor Analysis (CFA), were conducted to ensure the validity and reliability of the translated instrument. Our findings corroborate the robustness of the adapted questionnaire, affirming its suitability for use within the Brazilian Portuguese-speaking context. This paper meticulously delineates the adaptation process and resultant statistical outcomes, offering insights into the significance of IMI Teq Br for engagement and motivation research grounded in SDT principles. Additionally, we thorougly discuss the challenges inherent to this context in Brazil, providing valuable considerations for future endeavors.