Abstract
One of the most well-known bio-inspired algorithms used in optimization problems is the particle swarm optimization (PSO), which basically consists on a machine-learning technique loosely inspired by birds flocking in search of food. More specifically, it consists of a number of particles that collectively move on the search space in search of the global optimum. The Darwinian particle swarm optimization (DPSO) is an evolutionary algorithm that extends the PSO using natural selection, or survival of the fittest, to enhance the ability to escape from local optima. This paper firstly presents a survey on PSO algorithms mainly focusing on the DPSO. Afterward, a method for controlling the convergence rate of the DPSO using fractional calculus (FC) concepts is proposed. The fractional-order optimization algorithm, denoted as FO-DPSO, is tested using several well-known functions, and the relationship between the fractional-order velocity and the convergence of the algorithm is observed. Moreover, experimental results show that the FO-DPSO significantly outperforms the previously presented FO-PSO.

Abstract
Developing safe, concurrent (and parallel) software systems is a hard task in multiple aspects, particularly the sharing of information and the synchronization among multiple participants of the system. In the message passing paradigm, this is achieved by sending and receiving messages among different participants, raising a number of verification problems. For instance, exchanging messages in a wrong order may prevent the system from progressing, causing a deadlock.MPI is the most commonly used protocol for high-performance, message-based parallel programs, and the need for formal verification approaches is well acknowledged by much recent work (e.g., see [1]). © 2012 Springer-Verlag.

Abstract
The algebra of programming (AoP) is a discipline for programming from specifications using relation algebra. Specification vagueness and nondeterminism are captured by relations. (Final) implementations are functions. Probabilistic functions are half way between relations and functions: they express the propensity, or likelihood of ambiguous, multiple outputs. This paper puts forward a basis for a linear algebra of programming (LAoP) extending standard AoP towards probabilistic functions. Because of the quantitative essence of these functions, the allegory of binary relations which supports the AoP has to be extended. We show that, if one restricts to discrete probability spaces, categories of matrices provide adequate support for the extension, while preserving the pointfree reasoning style typical of the AoP.

Abstract
It is well-known that ROVs require human intervention to guarantee the success of their assignment, as well as the equipment safety. However, as its teleoperation is quite complex to perform, there is a need for assisted teleoperation. This study aims to take on this challenge by developing vision-based assisted teleoperation maneuvers, since a standard camera is present in any ROV. The proposed approach is a visual servoing solution, that allows the user to select between several standard image processing methods and is applied to a 3-DOF ROV. The most interesting characteristic of the presented system is the exclusive use of the camera data to improve the teleoperation of an underactuated ROV. It is demonstrated through the comparison and evaluation of standard implementations of different vision methods and the execution of simple maneuvers to acquire experimental results, that the teleoperation of a small ROV can be drastically improved without the need to install additional sensors.

Abstract

Abstract
Companies are moving from developing a single model for a problem (e.g., a regression model to predict general sales) to developing several models for sub-problems of the original problem (e.g., regression models to predict sales of each of its product categories). Given the similarity between the sub-problems, the process of model development should not be independent. Information should be shared between processes. Different approaches can be used for that purpose, including metalearning (MtL) and transfer learning. In this work, we use MtL to predict the performance of a model based on the performance of models that were previously developed. Given that the sub-problems are related (e.g., the schemas of the data are the same), domain knowledge is used to develop the metafeatures that characterize them. The approach is applied to the development of models to predict sales of different product categories in a retail company from Portugal.