The work by UL focuses on the optimization of ray-tracing techniques, commonly used in simulation of nuclear fusion phenomena, especially estimation of radiation on plasma facing surfaces, signal calculation of diagnostic systems to arrive at synthetic diagnostic signal.
Download the entire article:
Author: Matic Brank
In recent years, artificial intelligence (AI) has dramatically changed various industries, including tourism. The integration of advanced technologies such as high-performance computing (HPC) and machine learning has enabled tourism companies to improve user experience, optimize processes and adapt marketing strategies.
The tourism industry faces many challenges that can be addressed by advanced technologies. One of the main problems is processing the huge amounts of data generated by tourism providers such as hotels, airlines and travel agencies. Managing this data requires powerful systems that can analyze and predict customer behavior and optimize operational processes in real time. In addition, the tourism sector faces the need to improve the user experience at all customer touch points, from travel planning to post-service feedback. The key is for companies to understand customer preferences and behavior, which is no easy task in such a dynamic environment.
Download the entire article:
Authors: Domen Verber, Jani Dugonik
Many breakthroughs in speed and accuracy of single image super-resolution (SISR) have been achieved. One of the biggest challenges is how to recover finer texture details when super-resolution is applied at large upscaling factors. A typical solution to SISR involves using a convolutional neural network (CNN), however new approaches using a generative adversarial network (GAN) are now also popular. The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. In this work, we present an evaluation of SRResNet and SRGAN. SRResNet is a deep residual network and SRGAN is a generative adversarial network for image super-resolution (SR). SRResNet is able to recover reasonable quality photo-realistic textures from heavily downsampled images. SRGAN is capable of inferring photo-realistic natural images with high upscaling factors. This is achieved using a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes the solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. The content loss is motivated by perceptual similarity instead of similarity in pixel space.
Download the entire article:
Author: Sebastien Strban
Object counting is crucial in applications ranging from cell growth analysis, to crowd estimation in video surveillance systems, to product counting on production lines. Classical methods are not applicable in these cases because they require large annotated training sets and struggle in dense scenarios. On the other hand, the methods developed at the Visual Cognitive Systems Laboratory at the Faculty of Computer Science and Information Science, University of Ljubljana, develop accurate counting with just a few training examples and represent a methodological leap in computer vision.
Download the entire article:
The digital transformation of organizations represents a necessary process for achieving competitiveness in both the economic and public sectors, thereby contributing to the development of the entire society. To encourage and accelerate this process, the European Commission has adopted two multi-annual work programs for the Digital Europe Program (DIGITAL) to support digital transformation and strengthen cybersecurity.
In the framework of the call for proposals for European Digital Innovation Hubs (EDIH), published by the European Commission in 2021, two Slovenian candidates were selected, the DIGI-SI consortium and SRC-EDIH, while 4PDIH obtained a Certificate of Excellence.
Download the entire article:
Optimizing a container ship's hull for both calm water and added resistance in waves is a complex technical challenge. In calm water, the focus is on reducing resistance to enhance fuel efficiency and speed. This often involves streamlining the hull shape to minimize resistance. However, in wave conditions, additional factors like wave-induced motions, wave reflection and refraction come into play. This demands an iterative approach by employing CFD simulations and HPC resources to balance streamlined shapes for calm water efficiency hull geometry features that reduce added wave resistance.
Download the entire article:
In the development of new medicinal compounds, research laboratories employ robotic equipment for high-throughput screening (HTS), which enables automated experimental in vitro biological evaluation of various compounds with a capacity of several hundred thousand compounds [1]. With the rapid advancement of computer science, this relatively expensive and complex process is complemented and being replaced by computational tools for molecular modelling. Similar to experiments conducted in vivo or in vitro, we can also refer to experiments conducted in silico, which literally means in silicon, essentially referring to computer simulation.
Download the entire article:
