ABSTRACT. The aim of this talk will be to give an overview of visual domain adaptation methods. First, I will briefly recall historical shallow methods. Then, I will discuss different manners to exploit deep convolutional architectures in visual domain adaptation, focusing on image categorization. In the last part of my talk, I will overview recent trends in domain adaptation, including deep discriminative models, adversarial and encoder-decoder based models, network parameter adaptation methods, semi-supervised and curriculum learning based models. I will present methods proposed in the literature for image classification, semantic segmentation, object detection and others.

ABSTRACT. In this tutorial, we will discuss different aspects of scaling up deep learning models for image and video processing and the existing bottlenecks in terms of optimisation, memory footprint, and latency. We will present works that address these challenges by regularising the deep models, by compressing or sparsifying models, or by enabling parallelism during training and inference. The tutorial will start with basic notions of convnets for image and video processing and will present recent techniques for stabilising training or making these models more efficient.

ABSTRACT. Setting up an infrastructure for application deployment is a non-trivial task. We usually simplify the task by using the Infrastructure as Code (IaC) approach. Through IaC we set-up software defined infrastructure, able to run applications. Several tools and platforms have been developed to describe the system and to implement the actual deployment of the application (e.g., Puppet, Chef, Ansible, Terraform, etc.). When the infrastructure is heterogeneous (e.g., combination of Cloud and HPC) the challenges to set up a collaborative infrastructure is even harder, as the paradigms and environments differ significantly. Cloud infrastructure is focusing on servers, events, functions, while the HPC infrastructure is procedural and takes the application and merely executes it, usually on bare metal. As expected, the applications deployed on the respective systems differ in their dynamics and longevity. Whilst deployment of applications on the Cloud can be solved with the aforementioned tools, the HPC systems are in this respect in their infancy – the application is usually scheduled to be executed on a pre-defined set of processors and from that point on, the scripts merely gather input data, execute the application and then organise the output data. Merging the two approaches is thus currently rather hard and requires an explicit boundary between the Cloud and HPC parts of the application, posing a significant issue for the overall modelling and thus set-up of the system with the IaC approach. SODALITE, an H2020 project, is targeting simplification of the application deployment complexity while retaining or improving application performance on targeted HPC heterogeneous and cloud systems. The application deployment is abstracted through modelling of application’s component relationships, policies and performance. The application is deployed using appropriate container technologies, matching the targeted heterogeneous HPC and cloud-based platforms. The starting point is the definition of the system and the application within an AI-supported IDE, using a straightforward, TOSCA-similar language. The smart IDE backed with the Graph DB knowledge base supports the user with the suggestions on how the system and application could (or should) be modelled. In the next general step, this definition is executed through an orchestrator, resulting in an execution of the application within the software-defined environment. In cases where the source code is available, it is optimised for the targeted infrastructure before execution. This step ensures that the application performance is not lost due to abstraction. However, in all cases, the execution of the application is monitored, as SODALITE is using machine learning and control-theory approaches to improve runtime performance.SODALITE is currently in the middle of its development thus not all of its functionalities are available. Whilst we first addressed the typical private Cloud infrastructures (e.g. OpenStack) and HPC (e.g. Torque job scheduler) using containers to encapsulate the applications, there is still work to be done to address public Clouds (e.g. AWS) and other HPC schedulers (e.g. Slurm). The IDE is functional, however it still requires further improvements and enhancements. Finally, not all optimisation approaches are developed at the moment.The aim of the proposed approach is to flatten the learning curve for Ops enabling them to concentrate on domain problems, resulting in lower overall costs of development and application lifecycle management.

ABSTRACT. The fixed point problem raise more issues: the existence of fixed point, the uniqueness of fixed point, and exact or approximative method/methods of finding the fixed point. A natural question raise now is whether turn single valued fixed point theory for nonexpansive mapping in multivalued mapping. There are some challenges.

ABSTRACT. In this paper, based on the work [Asadollah Aghajani, Josef Banas, Navid Sabzali: Some generalizations of Darbo`s fixed point theorem and applications, Bull. Belg. Math. Soc. Simon Stevin 20 (2013), 345–358] we provide some generalizations of Darbo fixed point theorem.

ABSTRACT. Abstract—In this paper we introduce a new class of mappings, called enriched quasi nonexpansive mappings. We prove a convergence theorems for the Mann iteration used to approximate fixed points of enriched quasi nonexpansive mappings in a real Hilbert space.

ABSTRACT. In this paper, we prove common fixed point theorems under an enriched type contraction condition for two single- valued mappings satisfying a weak commutativity condition in Banach spaces. We also prove that the unique common fixed point of this mappings can be approximated by means of Krasnoselskij iteration for which strong convergence is obtained.

ABSTRACT. The aim of this paper is to introduce a large class of mappings, called enriched Suzuki mappings, that includes all Suzuki mappings and some nonexpansive mappings.

ABSTRACT. Systems for symbolic mathematical computation have risen to wide-spread acceptance over the past 50 years, from the early interactive systems of Reduce and Macsyma, to the popular general systems such as Maple and Mathematica, and open source and specialized tools. For the most part, these systems have not strayed far from the original model of a teletype conversation between a user and system, with the addition of batch computing, typeset formula display and graphical rendering of objects. Over the same period, other software tools have evolved to embrace media convergence, dynamic environments and multi-party interaction. In this talk we explore how these technologies can be applied and extended for symbolic mathematics. In particular, we will touch on the priority of mathematical collaboration, the deepening relationships among document analysis, computer algebra and proof assistants, and the suitability of specific technologies such as TypeScript, WebAssembly and Node for symbolic mathematical systems implementation. We will conclude with some thoughts on how a next generation of symbolic mathematics tools can impact education, industry and every-day life.

ABSTRACT. We present new simplifications of the Lambert W function.

ABSTRACT. This work introduces the CMK knot coloring software system. CMK is a command-line tool written in SWI-Prolog that computes colorings of three-dimensional knots by finite quandles. The original purpose was to classify knots according to computational properties. It features a predicate that computes knot quandle presentations from braid words. The authors describe the key algorithms. Errors in five braid representations within the Mathematica KnotData collection are revealed through CMK.

ABSTRACT. This paper is part of our ongoing research and collaboration on understanding the relations between CAD algorithms, equational constraints and curtains. In our previous work we provided an algorithm that would decompose the hypersurface given by an equational constraint, to produce a sign-invariant quasi-CAD (even in the presence of curtains). That method however fails to take full advantage of multiple equational constraints. Here we provide further clarification to \cite{McCallum2001} and discuss the close relationship between order invariance and lex-least invariance.

ABSTRACT. We present our work in progress on creating na- of Nemo (which we’ll not use) is a wrapper for some number tive functions in the Julia programming language to compute theoretic C-libraries like flint and arb, it is easy to use it with canonical forms of matrices over arbitrary fields.

ABSTRACT. Nowadays, the Copernicus services are providing an unprecedented amount of data able to characterize our climate, atmosphere, marine and land environment. The information encapsulated in those datasets is invaluable for experts and scientist all around the world. However, despite the potential, due to high technical requirements, the raw data is almost impossible to be used by the general public. The platform we are proposing, called WeCTOU, has the aim to fill some of this gap, by providing the public with climate and environmental information for touristic locations in Romania. WeCTOU is an independent service developed and operated by Meteo Romania, being funded by European Commission and the European Centre for Medium-Range Weather Forecasts (ECMWF) through C3S (Copernicus Climate Change Service). Through a simple and intuitive web interface, WeCTOU offers the users interactive access to a number of indicies, grouped in three classes: 1) climate-based indices for time period 2000-2018 (e.g. frostbite risk, respiratory comfort, weather for staying outdoor, etc.); 2) near-real time indicies (e.g. green cover, snow cover, sea water temperature); 3) forecast indicies (e.g. pollen allergy risk, sun burn risk, air quality). The application is build entirely with standard compliant free and open source software applications like GDAL, rasterio and GRASS GIS for the data processing workflows, Leaflet and Charts.js for the web front-end, GeoServer and PostGIS for the back-end data services.

ABSTRACT. The large amount of collected data in the field of Earth Observation has created the need for an automatization in processing and extraction information from it. Thus, deep learning (DL) techniques have gained popularity among the remote sensing community. Agriculture is one of the domains where DL can improve the current state-of-the-art. In this paper, we focus on the task of crop type classification, a key task in the process of assesing the agricultural market and yield. To this purpose, we introduce a large-scale dataset, based on publicly available data (data from satellite Sentinel-2 and annotations from Land Parcel Identification System), to be used for further research in this field.

ABSTRACT. This paper considers a problem of recognition of agricultural vegetation state from aerial photographs at various spatial resolutions. A semantic segmentation based on convolutional neural networks is used as a basis for recognition. Two variants of High-Resolution network architecture are described and used. These neural networks were trained and applied to aerial images of agricultural fields. In our experiments accuracy of four land classes recognition (soil, healthy vegetation, diseased vegetation and other objects) was about 93-94%

ABSTRACT. The Arctic is warming much faster than the global average and most of the areas located at high latitude experienced rapid changes due to permafrost degradation. The advantages of using remote sensing data to reveal landscape changes in permafrost areas has been widely acknowledged. While medium and coarse resolution free data cover large areas, and can give an insight of large scale changes, more subtle changes cannot be identified. The analysis focus on three study sites, two are located in West Siberia and one in the European side of Russia between 65-70N latitude, in different permafrost type (continuous, discontinuous, sporadic), and cover around 100 sq km each. We used optic VHR satellite images from QuickBird, GeoEye and WorldView at 0.5-2.5 m spatial resolution and object-based image analysis and statistics to assess the spatio-temporal changes in landcover types, number and extent of water bodies, fluvial dynamics between 2003 and 2017. The results revealed that in all sites there is an increase in number and area of surface water and an expansion of shrubs. The results are in agreement with other findings in similar areas from the Arctic. Acknowledgement This work was supported by a grant of the Romanian National Authority for Scientific Research and Innovation, CCDI-UEFISCDI, project number ERANET-RUS-PLUS-SODEEP, within PNCD III.