Abstract
Many of today’s domains of application of Machine Learning (ML) are dynamic in the sense that data and their patterns change over time. This has a significant impact in the ML lifecycle and operations, requiring frequent model (re-)training, or other strategies to deal with outdated models and data. This need for dynamic and responsive solutions also has an impact on the use of computational resources and, consequently, on sustainability indicators. This paper proposes an approach in line with the concept of Frugal AI, whose main aim is to minimize the resources and time spent on training models by re-using models from a pool of past models, when appropriate. Specifically, we present and validate a methodology for similarity-based model selection in data streaming environments with concept drift. Rather than training a new model for each new block of data, this methodology considers a pool with only a subset of the models and, for each new block of data, will select the best model from the pool. The best model is determined based on the distance between its training data and the current block of data. Distance is calculated based on a set of meta-features that characterizes the data, and on the Bray-Curtis distance. We show that it is possible to reuse previous models using this methodology, leading to potentially significant saving of resources and time, while maintaining predictive quality.

Abstract
This research contributes to the field of Human-Robot Collaboration (HRC) within dynamic and unstructured environments by extending the previously proposed Fuzzy State-Long Short-Term Memory (FS-LSTM) architecture to handle the uncertainty and irregularity inherent in real-world sensor data. Recognising the challenges posed by low-cost sensors, which are highly susceptible to environmental conditions and often fail to provide regular periodic readings, this paper introduces additional pre-processing blocks. These include two indirect Kalman filters and an additional LSTM network, which together enhance the input variables for the fuzzification process. The enhanced FS-LSTM approach is evaluated using real-world data, demonstrating its effectiveness in extracting meaningful information and accurately recognising human activities. This work underscores the potential of robotics in addressing global challenges, particularly in labour-intensive and hazardous tasks. By improving the integration of humans and robots in unstructured environments, this research contributes to the broader exploration of robotics in new societal applications, fostering connections and collaborations across diverse fields.

Abstract
Background: The coffee industry is one of the most important world supply chains, with an estimated consumption of two billion cups daily, making it the most consumed beverage worldwide. Coffee beans are primarily grown in tropical countries, with Brazil accounting for almost 50% of the production. The objective of this study is to examine the Brazilian trade between 2018 and 2022, focusing on state producers, logistical corridors, and importer countries. Methods: The methodology approach revolves around a quantitative method using Social Network Analysis measures. Results: The results reveal a massive concentration in local production (99.5%-Minas Gerais), port movements (99.9%-Santos, Itaguai, and Rio de Janeiro), and country buyers (80.9%-the United States, United Kingdon, and Japan). Conclusions: The study concludes that the Brazilian green coffee supply chain relies on a fragile and overloaded logistical network. Due to that, this study indicates that the stakeholders and decision-makers involved must consider this high concentration of production in some areas and companies. They must also address the bottlenecks in logistical corridors and the fierce competition involved in acquiring and processing Brazilian coffee production because these factors can drastically affect the revenue of the companies operating in this sector.

Abstract
Humans are good at understanding subjective or vague statements which, however, are hard to express in classical logic. Fuzzy logic is an evolution of classical logic that can cope with vague terms by handling degrees of truth and not just the crisp values true and false. Logic is the formal basis of computing, enabling the formal design of systems supported by tools such as model checkers and theorem provers.This paper shows how a model checker such as Alloy can evolve to handle both classical and fuzzy logic, enabling the specification of high-level quantitative relational models in the fuzzy domain. In particular, the paper showcases how QAlloy-F (a conservative, general-purpose quantitative extension to standard Alloy) can be used to tackle fuzzy problems, namely in the context of validating the design of fuzzy controllers. The evaluation of QAlloy-F against examples taken from various classes of fuzzy case studies shows the approach to be feasible.

Abstract
Formal contracts and assertions are effective methods to enhance software quality by enforcing preconditions, postconditions, and invariants. However, the adoption and impact of contracts in the context of mobile application development, particularly of Android applications, remain unexplored. We present the first large-scale empirical study on the presence and use of contracts in Android applications, written in Java or Kotlin. We consider 2,390 applications and five categories of contract elements: conditional runtime exceptions, APIs, annotations, assertions, and other. We show that most contracts are annotation-based and are concentrated in a small number of applications.

Abstract
The latest technological advances drive the emergence of countless real-time data streams fed by users, sensors, and devices. These data sources can be mined with the help of predictive and classification techniques to support decision-making in fields like e-commerce, industry or health. In particular, stream-based classification is widely used to categorise incoming samples on the fly. However, the distribution of samples per class is often imbalanced, affecting the performance and fairness of machine learning models. To overcome this drawback, this paper proposes Bplug, a balancing plug-in for stream-based classification, to minimise the bias introduced by data imbalance. First, the plugin determines the class imbalance degree and then synthesises data statistically through non-parametric kernel density estimation. The experiments, performed with real data from Wikivoyage and Metro of Porto, show that Bplug maintains inter-feature correlation and improves classification accuracy. Moreover, it works both online and offline.