Abstract
The sedimentary processes play a major role in every aquatic ecosystem, however, there are few automated options for in-situ monitoring of sediment displacement in the streambed of waterways. We present an automated optical instrument for in-situ continuous monitoring of sediment deposition and erosion of the streambed that requires no calibration. With a production cost of 32euro, power consumption of 300 mu A in sleep mode, and capacity to monitor the bedform of a waterway, the sensor was developed to evaluate the sediment dynamics of coastal areas with a wide spatial and temporal resolution. The novel device is intended to be buried in the sand and uses 32 infrared channels to monitor the streambed sediment height. For testing purposes, a maximum measuring length of 160 mm and 5 mm resolution was chosen, but these values are scalable. Sensors can be built with different ranges and precision according to the needs of the fieldwork. A laboratory experiment was conducted to demonstrate the working principle of the instrument and its behaviour regarding the turbidity originated by suspended sediment and the settling and deposition of the suspended particles. The device was deployed for 119 days in an estuarine area and was able to detect patterns in the sediment deposition and resuspension during the tidal cycles. Also, abnormal events occurred during the experiment as floods and algae blooms. During these events, the sensor was able to record exceptional erosion and sediment deposition rates. The reported automated instrument can be broadly used in sedimentary studies or management and planning of fluvial and maritime infrastructures to provide real-time information about the changes in the bedform of the watersheds.

Abstract
Assessment and tracking of activities in non-traditional contexts, such as immersive environments, is a complex and timeconsuming process for instructors. This limits the widespread adoption of immersive environments, since the consequences of those constraints are lack of awareness for orchestration of learning, and lack of elements for assessment. The Inven!RA architecture proposes tackling this problem by collecting status and outcome analytics from multiple immersive activities into a single learning plan, where they are mapped to learning objectives. This enables the creation of learning dashboards for instructors and students, to support their awareness and assessment, enabling learning orchestration and self-regulation of learning. We present an implementation of the Inven!RA architecture in a platform linked to a remote computer networking laboratory, exemplifying how the architecture can achieve its purported goals.

Abstract
Complex optimization problems are often associated to large search spaces and consequent prohibitive execution times in finding the optimal results. This is especially relevant when dealing with dynamic real problems, such as those in the field of power and energy systems. Solving this type of problems requires new models that are able to find near-optimal solutions in acceptable times, such as metaheuristic optimization algorithms. The performance of these algorithms is, however, hugely dependent on their correct tuning, including their configuration and parametrization. This is an arduous task, usually done through exhaustive experimentation. This paper contributes to overcome this challenge by proposing the application of sequential model algorithm configuration using Bayesian optimization with Gaussian process and Monte Carlo Markov Chain for the automatic configuration of a genetic algorithm. Results from the application of this model to an electricity market participation optimization problem show that the genetic algorithm automatic configuration enables identifying the ideal tuning of the model, reaching better results when compared to a manual configuration, in similar execution times.

Abstract
AbstractOne of the main obstacles to the successful treatment of cancer is the phenomenon of drug resistance. A common strategy to overcome resistance is the use of combination therapies. However, the space of possibilities is huge and efficient search strategies are required. Machine Learning (ML) can be a useful tool for the discovery of novel, clinically relevant anti-cancer drug combinations. In particular, deep learning (DL) has become a popular choice for modeling drug combination effects. Here, we set out to examine the impact of different methodological choices on the performance of multimodal DL-based drug synergy prediction methods, including the use of different input data types, preprocessing steps and model architectures. Focusing on the NCI ALMANAC dataset, we found that feature selection based on prior biological knowledge has a positive impact on performance. Drug features appeared to be more predictive of drug response. Molecular fingerprint-based drug representations performed slightly better than learned representations, and gene expression data of cancer or drug response-specific genes also improved performance. In general, fully connected feature-encoding subnetworks outperformed other architectures, with DL outperforming other ML methods. Using a state-of-the-art interpretability method, we showed that DL models can learn to associate drug and cell line features with drug response in a biologically meaningful way. The strategies explored in this study will help to improve the development of computational methods for the rational design of effective drug combinations for cancer therapy.Author summaryCancer therapies often fail because tumor cells become resistant to treatment. One way to overcome resistance is by treating patients with a combination of two or more drugs. Some combinations may be more effective than when considering individual drug effects, a phenomenon called drug synergy. Computational drug synergy prediction methods can help to identify new, clinically relevant drug combinations. In this study, we developed several deep learning models for drug synergy prediction. We examined the effect of using different types of deep learning architectures, and different ways of representing drugs and cancer cell lines. We explored the use of biological prior knowledge to select relevant cell line features, and also tested data-driven feature reduction methods. We tested both precomputed drug features and deep learning methods that can directly learn features from raw representations of molecules. We also evaluated whether including genomic features, in addition to gene expression data, improves the predictive performance of the models. Through these experiments, we were able to identify strategies that will help guide the development of new deep learning models for drug synergy prediction in the future.

Abstract
Although machine learning (ML) is a field that has been the subject of research for decades, a large number of applications with high computational power have recently emerged. Usually, we only focus on solving machine learning problems without considering how much energy has been consumed by the different frameworks used for such applications. This study aims to provide a comparison among four widely used frameworks such as Tensorflow, Keras, Pytorch, and Scikit-learn in terms of many aspects, including energy efficiency, memory usage, execution time, and accuracy. We monitor the performance of such frameworks using different well-known machine learning benchmark problems. Our results show interesting findings, such as slower and faster frameworks consuming less or more energy, higher or lower memory usage, etc. We show how to use our results to provide machine learning developers with information to decide which framework to use for their applications when energy efficiency is a concern.

Abstract