Search

Countering the fake news phenomenon has become one of the most important challenges for democratic societies, governments and non-profit organizations, as well as for the researchers coming from several domains. This is not a local problem and demands a holistic approach to analyzing heterogeneous data and storing the results. The research problem we face in this paper is the proposition of an innovative distributed architecture to tackle the above-mentioned problems. The architecture uses state-of-the-art technologies with a focus on efficiency, scalability and also openness, so that community-created components and digital content analyzers could be added. Moreover, we prove the usability of the prototype on Kaggle Fake News dataset. In particular, we consider different configurations of the proposed deep neural network and present the results reflecting the effectiveness, scalability and transferability of the proposed solution. [ABSTRACT FROM AUTHOR]

Recent progress in the area of modern technologies confirms that information is not only a commodity but can also become a tool for competition and rivalry among governments and corporations, or can be applied by ill-willed people to use it in their hate speech practices. The impact of information is overpowering and can lead to many socially undesirable phenomena, such as panic or political instability. To eliminate the threats of fake news publishing, modern computer security systems need flexible and intelligent tools. The design of models meeting the above-mentioned criteria is enabled by artificial intelligence and, above all, by the state-of-the-art neural network architectures, applied in NLP tasks. The BERT neural network belongs to this type of architectures. This paper presents Transformer-based hybrid architectures applied to create models for detecting fake news. [ABSTRACT FROM AUTHOR]

In this paper parametrized analysis of CPW (Coplanar Waveguide) applied in photonics is presented. The analysis is based on analytical formulas, which were delivered and modified from conformal mapping technique, and also after review of available literature from analytical models for the conventional CPW. Adjustment of the already reported CPW formulas was performed to properly model thin films CPW structures in PIC (Photonic Integrated Circuit). Thus, a closed expressions were proposed to simulate basic parameters, like: per unit length (p.u.l.) R (resistance), L (inductance), C (capacitance), G (conductance) and propagation quantity like characteristic impedance Zo versus frequency. The CPW line dedicated for the PIC application, is based on InP (Indium Phosphide) substrate and with air or BCB (Benzocyclobutene) surroundings of the metallization on the top. The modified formulas are curvefitting approximations, with new fitting coefficients. Essential modifications are for p.u.l. C and p.u.l. L. Usefulness and effectiveness of the modified method was verified through results comparison with simulations in full-wave Electronic Design Automation software, with the circuit simulator QUCS and with real measurements, conducted with the use of PNA (Performance Network Analyzer). The paper novelty are new coefficients for analytical formulas, implementation of modified formulas in the octave code and model parametrization for geometrical parameters. [ABSTRACT FROM AUTHOR]