Search

Reinforcement learning is a machine learning approach based on behavioral psychology. It is focused on learning agents that can acquire knowledge and learn to carry out new tasks by interacting with the environment. However, a problem occurs when reinforcement learning is used in critical contexts where the users of the system need to have more information and reliability for the actions executed by an agent. In this regard, explainable reinforcement learning seeks to provide to an agent in training with methods in order to explain its behavior in such a way that users with no experience in machine learning could understand the agent's behavior. One of these is the memory-based explainable reinforcement learning method that is used to compute probabilities of success for each state-action pair using an episodic memory. In this work, we propose to make use of the memory-based explainable reinforcement learning method in a hierarchical environment composed of sub-tasks that need to be first addressed to solve a more complex task. The end goal is to verify if it is possible to provide to the agent the ability to explain its actions in the global task as well as in the sub-tasks. The results obtained showed that it is possible to use the memory-based method in hierarchical environments with high-level tasks and compute the probabilities of success to be used as a basis for explaining the agent's behavior.

Emergency vehicles have been employed in rescue operations and supportive services, attending victims and managing critical situations in smart cities. Such vehicles, notably ambulances, fire trucks, police cars and transit agents vehicles, may be tracked and monitored in some applications for different functions. When such emergency vehicles are not equipped with GPS receivers, cameras can be used to view emergency signs printed on them, allowing indirect identification of emergency vehicles, although many complexities have to be considered when performing visual sensors-based tracking and monitoring. In this context, this paper proposes a mathematical model focused on the evaluation of the coverage efficiency of a group of visual sensors over moving vehicles, aimed at visual coverage of emergency signs. For that, vehicles, emergency signs and visual sensors are mathematically modelled in this paper, with coverage interactions among these elements being computed based on proposed geometry equations and algorithms. Doing so, the effectiveness of the positioning and configurations of visual sensors can be evaluated without requiring actual deployment, potentially reducing costs when assessing visual monitoring systems in this scenario.

Software developers can use diverse techniques and tools to reduce the number of vulnerabilities, but the effectiveness of existing solutions in real projects is questionable. For example, Static Analysis Tools (SATs) report potential vulnerabilities by analyzing code patterns, and Software Metrics (SMs) can be used to predict vulnerabilities based on high-level characteristics of the code. In theory, both approaches can be applied from the early stages of the development process, but it is well known that they fail to detect critical vulnerabilities and raise a large number of false alarms. This paper studies the hypothesis of using Machine Learning (ML) to combine alerts from SATs with SMs to predict vulnerabilities in a large software project (under development for many years). In practice, we use four ML algorithms, alerts from two SATs, and a large number of SMs to predict whether a source code file is vulnerable or not (binary classification) and to predict the vulnerability category (multiclass classification). Results show that one can achieve either high precision or high recall, but not both at the same time. To understand the reason, we analyze and compare snippets of source code, demonstrating that vulnerable and non-vulnerable files share similar characteristics, making it hard to distinguish vulnerable from non-vulnerable code based on SAT alerts and SMs.

In spatial-temporal data analysis, location data and its evolution through time are investigated with the goal of uncovering important information to provide novel insights. These insights, for example, may involve congestion identification in transportation, mobility patterns in urban computing, and storm prediction in weather forecasting. Clustering, one data analysis technique, groups spatial-temporal data based on location. Current spatial-temporal data analysis techniques fail to investigate relationships between spatial-temporal clusters, such as splitting from a cluster and merging with another one because of a change of properties over time. These relationships could hold valuable information about the existence of a cluster and its interactions with other clusters and trajectories. In this paper, we introduce a framework to identify, process, and analyze relationships between clusters of spatial-temporal data (e.g. enter, merge, or split). We describe its architecture and components, as well as a proposed clustering technique, the different approaches for distance calculation, and how we calculate cluster similarity of temporally separated clusters. The result of these operations are used in the identification of cluster relationships over space and time. The analysis of these relationships helps uncover hidden values that could support novel approaches to more effective decision-making. We evaluate our framework with two case studies, based on truck and human trajectories.

Wireless sensor networks comprising nodes equipped with cameras have become common in many scenarios, providing valuable visual data for some relevant services such as localization, tracking, patterns identification and emergencies detection. In this context, algorithms and optimization approaches have been designed to perform different types of quality assessment or performance enhancement tasks, addressing challenging issues such as networking, compression, availability, reliability, security, energy efficiency and virtually any subject related to the operational challenges of those networks. However, the dynamics of coverage failures have not been properly modelled in visual sensor networks, resulting in unrealistic perceptions when optimizing or assessing quality in most visual sensing scenarios. Particularly, the Field of View of visual sensors will be affected by occlusion caused by obstacles in the monitored field, which may turn such sensors inadequate for the expected monitoring services of the considered network. Therefore, this article proposes a mathematical model to assess occlusion caused by mobile obstacles such as vehicles on a road or forklifts in an industrial plant, aiming at the selection of the visual sensor nodes that will not have their coverage significantly restricted by those obstacles. Doing so, the proposed model can be exploited by any optimization or quality assessment approach in wireless visual sensor networks, providing a preprocessing method when selecting visual nodes.

Wireless visual sensor networks are commonly employed on several applications contexts such as smart cities, intelligent transportation systems and industrial management, aiming at the use of visual data from cameras to provide enhanced information and to expand the networks utilities. In these scenarios, some applications may require high-definition images when performing more specialized tasks, for example in face and text recognition, adding an important monitoring requirement when using camera-based sensors. In fact, it is important to ensure that the network is able to gather visual data with the associated required quality to each task, and such perceived quality may be processed as a function of the Field of View (FoV) of the visual sensors. In order to address this issue, new quality metrics are proposed for wireless visual sensor networks that are deployed to perform area coverage, exploiting for that different perceptions of the FoV. Those metrics are proposed along with redeployment optimization methods for visual sensor nodes aiming at the improvement of the perceived monitoring quality, which are based on greedy and evolutionary-based approaches. The proposed metrics and algorithms are expected to be more realistic than previous solutions, allowing flexible processing of variables as cameras' positions, orientations and viewing angles, providing then high flexibility on the definition of parameters and significantly contributing to the development of sensor networks based on visual sensors.

Emotions play an essential role in human behavior. It can be expressed in body language, facial expression, hand gesture, text, and many other ways. While recognizing emotions in human-human interaction is important, some individuals lack this ability. This is the case for individuals who have specific disabilities, for example, people diagnosed with autism, who may suffer from the inability to recognize emotions on faces, text, and figures. Some studies concentrate on developing approaches to assist individuals with this disability. In this regard, one can mention the studies dedicated to creating emotional agents to help in the recognition of emotions. However, there is a gap in studies that focus on the Portuguese language. This article presents an approach to bridge this gap by creating an emotional agent capable of processing voice and text in Portuguese. According to the sentiment present in words, the agent exhibits the emotions Joy, Sadness, and Neutral in a light-emitting diode matrix. Initial experiments indicate that it is possible to display these emotions based on the words' sentiments.

Spatial-temporal data analysis relates to the application of data analysis techniques to data where space and time are both relevant. Usually, the results of these techniques are used to classify or predict a phenomenon, but little attention is given to the explanation of how such phenomenon happened. For example, one may predict that a sporting event will happen at a particular location and date, but little is known about the indications that such event will happen (e.g. a higher number of vehicles on certain streets, parking lots becoming full, large number of vehicles going to supermarkets, or a sudden drop in pedestrian and vehicular traffic movement when the match starts). In this paper, we report on our ongoing work on using spatial-temporal cluster relationships to identify cluster lifecycles. These lifecycles are a series of stages through which a cluster passes during its lifetime, much like a human lifecycle of birth, growth, reproduction, and death. We focus on the identification of cluster lifecycle stages, namely start, expand, shrink, and end, and on their use to predict spatial-temporal phenomena, such as traffic congestion, human events, or animal movement.

The operation of a mobile robot relies significantly on the robustness of its hardware and software, in which the self-localization system plays a crucial role. However, keeping a mobile robot localized at all times is not trivial, as there are diverse sources of error which may cause mislocalizations, affecting the whole system with unforeseen consequences. In this work, we describe an innovative and reliable localization architecture for mobile robots grounded on adaptive Monte Carlo localization (AMCL), sensor fusion and scan-matching techniques for mobile robots in large indoor areas. To validate the system, we run two 24-hour-long surveillance demos with an in-house developed mobile robot, comparing and analyzing a standard and widely used AMCL localization architecture against the proposed system. While the standard AMCL leads to 85% of operation time with accurate localization, the robot under the proposed approach does not depict a single localization issue during the 24-hour pilot.

The adoption of sensors-based monitoring systems supported by Internet of Things technologies has opened new possibilities for data retrieving and processing in urban areas. Among such possibilities, emergencies management is expected to play an important role in how modern cities will evolve, reducing the negative impacts of critical events and improving the quality of life perceived by their inhabitants. Actually, when an emergency is detected and alerted, emergency vehicles, notably ambulances, fire trucks, police cars and transit agents vehicles, should be quickly assigned to respond to that situation, as soon as possible. In this context, we propose a dynamic algorithm to automatically assign emergency vehicles in smart city scenarios, exploiting for that a sensors-based emergency detection system to provide emergency alerts. The proposed algorithm can then be used to quickly assign a number of emergency vehicles in the first moments of an emergency, which can potentially save lives and improve existing crisis management applications in smart cities.

The number of people diagnosed with anxiety disorders has been increasing in recent years. The correct diagnosis of such disorders is not always a trivial task, sometimes forcing an individual to consult with many clinicians and performing several medical exams. Post-traumatic Stress Disorder (PTSD) is a disorder related to experienced events, which presented a certain degree of threat to an individual. When experiencing situations that refer to past events, an individual may present reactions that trigger physiological changes in the organism such as tachycardia or bradycardia. Many disorders have common symptoms, and identifying these subtleties results in greater diagnosis efficiency and effectiveness. Artificial Intelligence (AI) and Data Mining (DM) techniques have helped specialists in the diagnosis and prevention of diseases and disorders. In our research, we aim at finding new biomarkers to diagnose PTSD analyzing physiological signals with DM techniques. In this paper, we used a dataset from an experiment with civilians that were recently exposed to traumatic events related to violence. Those individuals completed a questionnaire that evaluates the impact of such events through PCL (PTSD Disorder Checklist for DSM-IV) scale. Heart rate and skin conductance signals were collected while viewing emotional and neutral stimuli images. We applied DM techniques and classification algorithms to evaluate and maximize PCL score prediction performance considering those physiological signal data. The best result was obtained with SMO algorithm with its hyperparameters values suggested by an auto-learning procedure, presenting an accuracy of 85.45% (p-value = 0.001), precision of 0.8 (p-value = 0.001), recall of 0.5714 and F-Measure of 0.6667 (p-value = 0.001).

Spatial-temporal data refers to potentially massive amounts of data gathered across both space and time. Spatial-temporal data analysis helps uncover the value that this type of data holds to domains such as transportation operations, traffic management, service demand, and trip planning. Specifically, cluster analysis groups data into sets known as clusters such that elements inside a cluster are more similar to each other than elements in other clusters. Cluster analysis has been successfully applied in domains such as transportation, ecology, medicine, and astronomy. However, current cluster analysis techniques limit themselves to static cluster analysis, thereby missing the identification of interesting insights and patterns related to the evolution of clusters over time. In this paper, we clarify the concept of dynamic clusters and support new forms of cluster analyses by introducing, describing, and formalizing cluster relationships that represent important events, such as split or merge, that a cluster may go through from its start to its end. These relationships provide a foundation for investigating cluster evolution and providing novel insights for better operational and business decision making.

In this work we address a problem of practical relevance -the delivery of X.509 v3 digital certificates to end users on personal computers and smartphones- and we seek to describe the design, construction and evaluation of an artifact with which it is possible to solve it. Seven meta-requirements were defined, which have been instantiated into an artifact, which in turn has been empirically evaluated. The evaluation of the artifact has demonstrated its effectiveness based on actual use and perceived ease of use. The results of this work contribute to positioning personal computers and smartphones as alternative cryptographic modules to smart cards for some usage scenarios.

Low bone mineral density can lead to weak and fragile bones that lead to problems of osteoporosis and fractures in people, early detection can help their treatment. This research compares five data mining algorithms to predict bone weakness in students between 5 and 18 years of age. The methodology used for data processing is CRISP-DM. The accuracy of the algorithms applied in the referenced works with the results obtained with the WEKA data mining tool is discussed. After making the comparison, it was determined that the JRip algorithm was more precise.

Wireless visual sensor networks (WVSN) bring a more comprehensive perception of monitored environments, leading to an increase adoption of such networks as a promising solution for a wide range of applications. Among many examples, highlight industrial applications related to the industry 4.0 paradigm, which increasingly require more data from manufacturing systems. Those sensor-based applications are in many cases safety-critical, requiring dependability guarantees mainly related with reliability and availability, that should be maintained during the whole network operation. Although several approaches have provided network deployment with dependability guarantees, sometimes the monitored environment or the application configurations can change during the network operation, which can violate the dependability requirements and demand network redeployment in order to keep those guarantees. In this paper we propose a novel algorithm to redeploy WVSN guided by the optimization of the application dependability, considering changes on cameras’ orientations. A methodology is defined to support dependability analysis. We compare the results of the proposed algorithm with previous algorithms found in literature. The achieved results show that the proposed algorithm is useful and efficient to provide network redeployment, keeping or improving the application dependability.