Abstract
Image dehazing tries to solve an undesired loss of visibility in outdoor images due to the presence of fog. Recently, machine-learning techniques have shown great dehazing ability. However, in order to be trained, they require training sets with pairs of foggy images and their clean counterparts, or a depth-map. In this paper, we propose to learn the appearance of fog from weakly-labeled data. Specifically, we only require a single label per-image stating if it contains fog or not. Based on the Multiple-Instance Learning framework, we propose a model that can learn from image-level labels to predict if an image contains haze reasoning at a local level. Fog detection performance of the proposed method compares favorably with two popular techniques, and the attention maps generated by the model demonstrate that it effectively learns to disregard sky regions as indicative of the presence of fog, a common pitfall of current image dehazing techniques.

Abstract
A single synchronous controller (SSC) technique is proposed in this paper for control of interfaced converters under high penetration of renewable energy resources (RER) into the power grid. The proposed SSC is based on a new dynamic model concerning to the power grid stability (PGS) and modeled based on all features of a synchronous generator (SG), which can properly improve the performance of the power grid in those scenarios in which a large-scale penetration of RERs is considered. Different transfer functions are achieved to assess the high performance of the proposed control technique. Simulation results are presented to demonstrate the superiority of the proposed SSC in the control of the power electronic-based synchronous generator under high penetration of RERs into the power grid.

Abstract
This paper presents an approach to the shared task HaSpeeDe within Evalita 2018. We followed a standard machine learning procedure with training, validation, and testing phases. We considered word embedding as features and deep learning for classification. We tested the effect of merging two datasets in the classification of messages from Facebook and Twitter. We concluded that using data for training and testing from the same social network was a requirement to achieve a good performance. Moreover, adding data from a different social network allowed to improve the results, indicating that more generalized models can be an advantage.

Abstract
This note revisits layered logics from a coalgebraic point of view, and proposes a naturality condition to express the typical hierarchical requirement under which all abstract transitions should be traceable in more specialised layers. © Springer International Publishing AG 2018.

Abstract
High level vehicle automation systems are currently being studied to attenuate highway traffic and energy consumption by applying the concept of platooning, which has gained increased attention due to progresses in the next generation of mobile communication (5G). The introduction of more complex automation systems originates, however, fault entry points that hinders the system safety and resilience. This paper presents an initial control architecture for the electric vehicles platoon from a fault-tolerant control perspective. To achieve a fault-tolerant platooning structure an over-actuated electric vehicle topology is proposed which may allow the implementation of different redundancies. Furthermore, some of the major challenges in the platooning network control system (NCS) are presented and the techniques to overcome these issues are explored.

Abstract
Pricing and capacity decisions in car rental companies are characterized by high flexibility and interdependence. When planning a selling season, tackling these two types of decisions in an integrated way has a significant impact. This paper tackles the integration of capacity and pricing problems for car rental companies. These problems include decisions on fleet size and mix, acquisitions and removals, fleet deployment and repositioning, as well as pricing strategies for the different rental requests. A novel mathematical model is proposed, which considers the specific dynamics of rentals on the relationship between inventory and pricing as well as realistic requirements from the flexible car rental business, such as upgrades. Moreover, a solution procedure that is able to solve real-sized instances within a reasonable time frame is developed. The solution procedure is a matheuristic based on the decomposition of the model, guided by a biased random-key genetic algorithm (BRKGA) boosted by heuristically generated initial solutions. The positive impact on profit, of integrating capacity and pricing decisions versus a hierarchical/sequential approach, is validated.