Abstract

Abstract
The ever-growing nature of user generated data in online systems poses obvious challenges on how we process such data. Typically, this issue is regarded as a scalability problem and has been mainly addressed with distributed algorithms able to train on massive amounts of data in short time windows. However, data is inevitably adding up at high speeds. Eventually one needs to discard or archive some of it. Moreover, the dynamic nature of data in user modeling and recommender systems, such as change of user preferences, and the continuous introduction of new users and items make it increasingly difficult to maintain up-to-date, accurate recommendation models. The objective of this workshop is to bring together researchers and practitioners interested in incremental and adaptive approaches to stream-based user modeling, recommendation and personalization, including algorithms, evaluation issues, incremental content and context mining, privacy and transparency, temporal recommendation or software frameworks for continuous learning.

Abstract
In this paper a real-time laboratorial experiment is presented, intended to validate a 'grey-box' equivalent model for medium voltage active distribution networks with high presence of converter-connected generation, considering the latest European grid codes requirements, in response to severe faults at the transmission network side. A hybrid setup was implemented at INESC TEC's laboratory (Porto, Portugal), relying on a real-time digital simulator to provide the interface between simulation and physical assets available at the laboratory, in a power-hardware-in-the-loop configuration. The study considered the laboratory's internal network to be operating (virtually) as a medium voltage distribution network with converter-connected generation (fault ride through compliant), connected to a fully-detailed transmission network model. The aggregated reactive power response of the laboratory's network was fitted by the dynamic equivalent model, recurring to an evolutionary particle swarm optimization algorithm. The methodology adopted, testing conditions and respective results are presented. © 2019 IEEE.

Abstract
Fs-laser micromachining is a high precision fabrication technique that can be used to write novel three-dimensional structures, depending on the nature of light-matter interaction. In fused silica, the material modification can lead to (i) an increase of the refractive index around the focal volume, resulting in the formation of optical circuits, or (ii) an enhancement of the etch rate of the laser-affected zones relative to the pristine material, leading to a selective and anisotropic etching reaction that enables fabrication of microfluidic systems. Here, both effects are combined to fabricate a Fabry-Pérot interferometer, where optical waveguides and microfluidic channels are integrated monolithically in a fused silica chip. By filling the channel with a magnetic fluid whose refractive index changes with an external magnetic field, the device can be used as a magnetic field sensor. A linear sensitivity of -0.12 nm/mT is obtained in the 5.0±0.5 to 33.0±0.5 mT range, with the field being applied parallel to the light propagation direction.

Abstract
Neural field models, formalized by integro-differential equations, describe the large-scale spatio-temporal dynamics of neuronal populations [1]. They have been used in the past as a framework for modeling a wide range of brain functions, including multi-item working memory [2]. Neural field equations support spatially localized regions of high activity (or bumps) that are initially triggered by brief sensory inputs and subsequently become self-sustained by recurrent interactions within the neural population. We apply a special class of oscillatory coupling functions and analyze how the shape and spatial extension of multi-bump solutions change as the spatial ranges of excitation and inhibition within the field are varied [3]. More precisely, we use numerical continuation to find and follow solutions of neural field equations as the parameter controlling the distance between consecutive zeros of the coupling function is varied [4]. Important for a working memory application (e.g. [5]), we investigate how changes in this parameter affect the shape of bump solutions and therefore the maximum number of bumps that may exist in a given finite interval.

Abstract
This study analyzed the accuracy of a BGL predictive model (BGL-PM) for type 1 diabetes mellitus patients (T1DM) in a real-world environment. The study population consisted of 10 individuals with T1DM, half of them were female, age 33 (SD:11.2), BMI of 26.1 (4.2) and 60% were under carbohydrate-count treatment. After consent, patients underwent a medical evaluation and registered their daily activities using a smartphone application (GlucoTrends) for 28 days, with BGL and heart rate continuously monitored. BGL-PM was developed using a Deep Learning architecture, based on Recurrent Neural Networks. Models were trained for each patient using different training sets sizes (7, 14, 21 days). Prediction accuracy was evaluated by Mean Absolute Percentage Error (MAPE) on the last 5 days for different Prediction Horizons (PH): 30, 60, 120, 180 and 360 minutes, comparing full day and nocturnal period. The model predicted BGL with relevant accuracy for the dataset with 21 training days up to 60 minutes in both periods: full day (median MAPE 22.5%) and nocturnal (14.3%) (Figure). The BGL-PM was able to provide useful BGL predictions, especially during the night period, which can be improved by increasing the training period. Consequently, this BGL-PM poses as a complementary tool for the prevention of acute complications such as hypoglycemia and hyperglycemia in the management of DM. Disclosure M. Foss-Freitas: None. G.S. Moreira: Stock/Shareholder; Self; GlucoGear Tecnologia. V.P. Antloga: Stock/Shareholder; Self; GlucoGear Tecnologia. C.R. Neto: Research Support; Self; University of Sao Paulo. E.M. Rodrigues: Consultant; Self; GlucoGear Tecnologia. M.F. da Costa: Research Support; Self; GlucoGear. A.P. dos Santos: None. Y.K. Matsumoto: Board Member; Self; GlucoGear. Stock/Shareholder; Self; GlucoGear. Other Relationship; Self; GlucoGear.