Abstract
Management of network resources in advanced IoT applications is a challenging topic due to their distributed nature from the Edge to the Cloud, and the heavy demand of real-time data from many sources to take action in the deployment. FANETs (Flying Ad-hoc Networks) are a clear example of heterogeneous multi-modal use cases, which require strict quality in the network communications, as well as the coordination of the computing capabilities, in order to operate correctly the final service. In this paper, we present a Virtual Network Embedding (VNE) framework designed for the allocation of dataflow applications, composed of nano-services that produce or consume data, in a wireless infrastructure, such as an airborne network. To address the problem, an anypath-based heuristic algorithm that considers the quality demand of the communication between nano-services is proposed, coined as Quality-Revenue Paired Anypath Dataflow VNE (QRPAD-VNE). We also provide a simulation environment for the evaluation of its performance according to the virtual network (VN) request load in the system. Finally, we show the suitability of a multi-parameter framework in conjunction with anypath routing in order to have better performance results that guarantee minimum quality in the wireless communications.

Abstract
Performing experiments with Internet-of-Things edge devices is not always a trivial task, as large physical testbeds or complex simulators are often needed, leading to low reproducibility and several difficulties in crafting complex scenarios and tweaking parameters. Most available simulators try to simulate as close to reality as possible. While we agree that this kind of high-fidelity simulation might be necessary for some scenarios, we argue that a low-fidelity easy-to-change simulator may be a good solution when rapid prototyping orchestration strategies and algorithms. In this work, we introduce SIMoT, a low-fidelity orchestrator simulator created to achieve shorter feedback loops when testing different orchestration strategies for task allocation in edge devices. We then transferred the simulator-validated algorithms to both physical and virtual testbeds, where it was possible to assert that the simulator results correlate strongly with the observations on those testbeds.

Abstract
Event extraction is an Information Retrieval task that commonly consists of identifying the central word for the event (trigger) and the event's arguments. This task has been extensively studied for English but lags behind for Portuguese, partly due to the lack of task-specific annotated corpora. This paper proposes a framework in which two separated BERT-based models were fine-tuned to identify and classify events in Portuguese documents. We decompose this task into two sub-tasks. Firstly, we use a token classification model to detect event triggers. To extract event arguments, we train a Question Answering model that queries the triggers about their corresponding event argument roles. Given the lack of event annotated corpora in Portuguese, we translated the original version of the ACE-2005 dataset (a reference in the field) into Portuguese, producing a new corpus for Portuguese event extraction. To accomplish this, we developed an automatic translation pipeline. Our framework obtains F1 marks of 64.4 for trigger classification and 46.7 for argument classification setting, thus a new state of the art reference for these tasks in Portuguese.

Abstract
The application of machine learning algorithms to predict the mechanism of action (MoA) of drugs can be highly valuable and enable the discovery of new uses for known molecules. The developed methods are usually evaluated with small subsets of MoAs with large support, leading to deceptively good generalization. However, these datasets may not accurately represent a practical use, due to the limited number of target MoAs. Accurate predictions for these rare drugs are important for drug discovery and should be a point of focus. In this work, we explore different training strategies to improve the performance of a well established deep learning model for rare drug MoA prediction. We explored transfer learning by first learning a model for common MoAs, and then using it to initialize the learning of another model for rarer MoAs. We also investigated the use of a cascaded methodology, in which results from an initial model are used as additional inputs to the model for rare MoAs. Finally, we proposed and tested an extension of Mixup data augmentation for multilabel classification. The baseline model showed an AUC of 73.2% for common MoAs and 62.4% for rarer classes. From the investigated methods, Mixup alone failed to improve the performance of a baseline classifier. Nonetheless, the other proposed methods outperformed the baseline for rare classes. Transfer Learning was preferred in predicting classes with less than 10 training samples, while the cascaded classifiers (with Mixup) showed better predictions for MoAs with more than 10 samples. However, the performance for rarer MoAs still lags behind the performance for frequent MoAs and is not sufficient for the reliable prediction of rare MoAs.

Abstract

Abstract
Accurate and practical load modeling plays a critical role in the power system studies including stability, control, and protection. Recently, wide-area measurement systems (WAMSs) are utilized to model the static and dynamic behavior of the load consumption pattern in real-time, simultaneously. In this article, a WAMS-based load modeling method is established based on a multi-residual deep learning structure. To do so, a comprehensive and efficient load model founded on combination of impedance-current-power and induction motor (IM) is constructed at the first step. Then, a deep learning-based framework is developed to understand the time-varying and complex behavior of the composite load model (CLM). To do so, a residual convolutional neural network (ResCNN) is developed to capture the spatial features of the load at different location of the large-scale power system. Then, gated recurrent unit (GRU) is used to fully understand the temporal features from highly variant time-domain signals. It is essential to provide a balance between fast and slow variant parameters. Thus, the designed structure is implemented in a parallel manner to fulfill the balance and moreover, weighted fusion method is used to estimate the parameters, as well. Consequently, an error-based loss function is reformulated to improve the training process as well as robustness in the noisy conditions. The numerical experiments on IEEE 68-bus and Iranian 95-bus systems verify the effectiveness and robustness of the proposed load modeling approach. Furthermore, a comparative study with some relevant methods demonstrates the superiority of the proposed structure. The obtained results in the worst-case scenario show error lower than 0.055% considering noisy condition and at least 50% improvement comparing the several state-of-art methods.