Cellular technologies are widely used in the ICT domain, that paving the way towards new opportunities for education and practical experimentation. It also tackles software-related engineering areas, from Digital Processing (DSP) to Software Defined Networks and Network Virtualization applications. Due to the various constraints such as limited access to hardware components, high costs of such equipment, and spectrum regulations that must not be violated, practical education in telecommunications, electrical engineering, and electronics usually lacks the opportunity to pursue experimentation on the realistic cellular deployments. Therefore, in this paper we present a network testbed that allows students to experiment with a fully functional 4G LTE system with no radio. This testbed system mimics the realistic 4G LTE deployment, supporting students towards acquiring valuable knowledge in the field of cellular networks. It is low-cost due the fact there is no need for radio components or Software Defined Radio (SDR) devices, with no limitation on frequency utilization and regulations. The testbed provides seamless scalability for education classes as it can be deployed on top of any machine with general-purpose processor, installing the whole system within two networked PCs or in a fully virtualized environment. We present the testbed framework, as well as the hands-on practices on incorporating such low-cost realistic testbed into the education within various engineering fields.

A practical compound of education in computer science and electrical engineering, driven by increased availability and maturity of many emerging technologies, should be enriched by various laboratory resources in order to synchronize the paces between technology advancements and education. In particular, advancements in containerization as a virtualization technique pave the way towards allowing students to deploy their project applications with a lightweight resource footprint on top of the cloud. Being backed by a valuable feedback from 45 Bachelor students, in this paper we present the best practices on how virtualization can be leveraged to create a scalable environment for on-demand remote experimentation with distributed systems.

Storage of large quantities of industrial by-products can pose a serious environmental problem. There is a growing need to recycle these industrial by-products, including red mud, in the construction industry, which is one of the largest consumers of raw materials. A preliminary study of the potential of red mud as a raw material for concrete is presented in this paper. Chemical composition of red mud, determined by X-ray fluorescence and particle size distribution by laser diffraction, is tested as part of its initial evaluation. The reactivity of red mud is evaluated by the R3 test method with isothermal calorimetry. The compressive strength test is carried out on a mortar sample in which 30% by weight of cement is replaced by red mud. Preliminary tests indicate that red mud can be used as raw material in alternative binders.

Recent work has shown that large text-based neural language models, trained with conventional supervised learning objectives, acquire a surprising propensity for few- and one-shot learning. Here, we show that an embodied agent situated in a simulated 3D world, and endowed with a novel dual-coding external memory, can exhibit similar one-shot word learning when trained with conventional reinforcement learning algorithms. After a single introduction to a novel object via continuous visual perception and a language prompt ("This is a dax"), the agent can re-identify the object and manipulate it as instructed ("Put the dax on the bed"). In doing so, it seamlessly integrates short-term, within-episode knowledge of the appropriate referent for the word "dax" with long-term lexical and motor knowledge acquired across episodes (i.e. "bed" and "putting"). We find that, under certain training conditions and with a particular memory writing mechanism, the agent's one-shot word-object binding generalizes to novel exemplars within the same ShapeNet category, and is effective in settings with unfamiliar numbers of objects. We further show how dual-coding memory can be exploited as a signal for intrinsic motivation, stimulating the agent to seek names for objects that may be useful for later executing instructions. Together, the results demonstrate that deep neural networks can exploit meta-learning, episodic memory and an explicitly multi-modal environment to account for 'fast-mapping', a fundamental pillar of human cognitive development and a potentially transformative capacity for agents that interact with human users.

This work is an assessment of changes in magnetic properties of magnetic nanoparticles in various biological environments. To mimic variations in biological conditions, we have evaluated magnetic performance of Synomag® nanoparticles in two experiments: the effect of viscosity by varying the ratio of glycerol/water mixture and the effect of immobilization after blocking the Brownian relaxation by freeze-drying (to mimic uptake in macrophages). The magnetic response was measured with the Superparamagnetic quantifier. Synomag® exhibits a slight decrease (7.9%) of magnetic response under increased viscosity from η1=0.95 to η6=259.71 mPa.s, and a dramatic magnetic signal drop (78.2%) after freeze-drying. Synomag® nanoparticles are less sensitive to viscosity due  to an additional  relaxation mechanism of disordered spins within the nanoflowers. However, the magnetic performance has been reduced due to the blocking of Brownian relaxation after immobilization.   Int. J. Mag. Part. Imag. 6(2), Suppl. 1, 2020, Article ID: 2009022, DOI: 10.18416/IJMPI.2020.2009022

This research presents the epileptic focus region localization during epileptic seizures by applying different signal processing and ensemble machine learning techniques in intracranial recordings of electroencephalogram (EEG). Multi-scale Principal Component Analysis (MSPCA) is used for denoising EEG signals and the autoregressive (AR) algorithm will extract useful features from the EEG signal. The performances of the ensemble machine learning methods are measured with accuracy, F-measure, and the area under the receiver operating characteristic (ROC) curve (AUC). EEG-based focus area localization with the proposed methods reaches 98.9% accuracy using the Rotation Forest classifier. Therefore, our results suggest that ensemble machine learning methods can be applied to differentiate the EEG signals from epileptogenic brain areas and signals recorded from non-epileptogenic brain regions with high accuracy.

U ovome ćemo radu nastojati opisati umetnute nerestriktivne relativne rečenice kao tipične parentetičke strukture. Opis će podrazumijevati njihovu sintaksičku i semantičku interpretaciju, ali i naznake o njihovoj pragmatičkoj ulozi i zasnivat će se na analizi primjera pronađenih u korpusu tekstova koji pripadaju različitim stilovima bosanskoga jezika. Rad je podijeljen u tri dijela. U prvom ćemo dijelu definirati parentetike i dati pregled dosadašnjih pristupa u analiziranju te pojave. U drugome dijelu rada fokusirat ćemo se na definiranje i analizu umetnutih nerestriktivnih relativnih rečenica, a u trećem će se dijelu izdvojiti najznačajnija zapažanja.

Abstract Objectives The aim of the study was to determine carotid intima-media thickness (CIMT) values in patients who developed and did not develop preeclampsia (PE), and to determine whether CIMT values could be predictors of PE development. Methods The study included pregnant women who were examined by regular ultrasound examination at the Materno-Infantil Presidente Vargas Hospital (HMIPV) in Porto Alegre, Brazil, from April 2016 to September 2017. The examinations were performed every three months. Patients were divided into two groups. The first group included patients diagnosed with PE (n=21) and second group included patients who did not have PE (n=199). A high frequency ultrasound device (12 MHz) with a semi-automatic method was used to estimate CIMT. Results CIMT was significantly higher in pregnant women with PE than in women without PE (55±0.11 vs. 0.44±0.06, respectively; p<0.001). Using a cut-off value of 0.51 mm, CIMT had a specificity of 77.9% and sensitivity of 81% in the diagnosis of PE. With CIMT ≥0.6 mm, the probability of a patient developing PE was 44.4%; with CIMT >0.42 mm, the probability was only 4.2%. Conclusions An increase in CIMT was associated with the onset of PE. CIMT values were significantly higher in patients who develop PE.