BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Invited talk: Artificial Intelligence approaches for Monte Carlo s
 imulation in medical physics
DTSTART;VALUE=DATE-TIME:20220914T083000Z
DTEND;VALUE=DATE-TIME:20220914T091000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-849@events.ncbj.gov.pl
DESCRIPTION:Speakers: David Sarrut (Creatis Medical Imaging Research Cente
 r)\nMonte Carlo simulation of particle tracking in matter is the reference
  simulation method in the field of medical physics. It is heavily used in 
 various applications such as \n\n1. patient dose distribution estimation i
 n different therapy modalities (radiotherapy\, protontherapy or ion therap
 y) or for radio-protection investigations of ionizing radiation-based imag
 ing systems (CT\, nuclear imaging)\,\n2. development of numerous imaging d
 etectors\, in X-ray imaging (conventional CT\, dual-energy\, multi-spectra
 l\, phase contrast … )\, nuclear imaging (PET\, SPECT\, Compton Camera) 
 or even advanced specific imaging methods such as proton/ion imaging\, or 
 prompt-gamma emission distribution estimation in hadrontherapy monitoring.
 \n\nMonte Carlo simulation is a key tool both in academic research labs as
  well as industrial research and development services. Because of the very
  nature of the Monte Carlo method\, involving iterative and stochastic est
 imation of numerous probability density functions\, the computation time i
 s high. In this presentation\, we will review the recent use of Artificial
  Intelligence methods for Monte Carlo simulation in medical physics and th
 eir main associated challenges.\n\nhttps://events.ncbj.gov.pl/event/141/co
 ntributions/849/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/849/
END:VEVENT
BEGIN:VEVENT
SUMMARY:EuroCC technical tutorial on LUMI European Pre-Exascale Supercompu
 ter - part 2
DTSTART;VALUE=DATE-TIME:20220915T142000Z
DTEND;VALUE=DATE-TIME:20220915T160000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-862@events.ncbj.gov.pl
DESCRIPTION:Speakers: Maciej Szpindler (ACC Cyfronet AGH)\nThe LUMI is one
  of the European pre-exascale HPC systems hosted by the LUMI consortium. T
 he LUMI (Large Unified Modern Infrastructure) consortium countries are Fin
 land\, Belgium\, Czech Republic\, Denmark\, Estonia\, Iceland\, Norway\, P
 oland\, Sweden\, and Switzerland. This one-day tutorial presents a technic
 al overview of the system's hardware configuration and programming level e
 nvironment. The aim of the course is to popularize hardware design of the 
 compute nodes and network and associated programming environment. This int
 roductory material is meant to be a quick-start for those who consider acc
 ess to the LUMI resources and brief introduction to the software tools ava
 ilable and capabilities of the hardware.\n\n\n**Prerequisites**\n\nAn SSH 
 client. Appropriate accounts will be created on 14th of September after pa
 rticipants registration.\n\nhttps://events.ncbj.gov.pl/event/141/contribut
 ions/862/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/862/
END:VEVENT
BEGIN:VEVENT
SUMMARY:EuroCC technical tutorial on LUMI European Pre-Exascale Supercompu
 ter - part 1
DTSTART;VALUE=DATE-TIME:20220915T123000Z
DTEND;VALUE=DATE-TIME:20220915T140000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-861@events.ncbj.gov.pl
DESCRIPTION:Speakers: Maciej Szpindler (ACC Cyfronet AGH)\nThe LUMI is one
  of the European pre-exascale HPC systems hosted by the LUMI consortium. T
 he LUMI (Large Unified Modern Infrastructure) consortium countries are Fin
 land\, Belgium\, Czech Republic\, Denmark\, Estonia\, Iceland\, Norway\, P
 oland\, Sweden\, and Switzerland. This one-day tutorial presents a technic
 al overview of the system's hardware configuration and programming level e
 nvironment. The aim of the course is to popularize hardware design of the 
 compute nodes and network and associated programming environment. This int
 roductory material is meant to be a quick-start for those who consider acc
 ess to the LUMI resources and brief introduction to the software tools ava
 ilable and capabilities of the hardware.\n\n**Prerequisites**\n\nAn SSH cl
 ient. Appropriate accounts will be created on 14th of September after part
 icipants registration.\n\nhttps://events.ncbj.gov.pl/event/141/contributio
 ns/861/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/861/
END:VEVENT
BEGIN:VEVENT
SUMMARY:EuroCC tutorial on transfer learning in computer vision - part 2
DTSTART;VALUE=DATE-TIME:20220916T142000Z
DTEND;VALUE=DATE-TIME:20220916T160000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-864@events.ncbj.gov.pl
DESCRIPTION:Speakers: Aleksander Ogonowski (National Centre for Nuclear Re
 search)\nTransfer learning is a machine learning (ML) technique of reusing
  models with pre-trained knowledge obtained for a general ML task\, and ap
 plying it to another\, more specific ML task\, with limited training data 
 or computational resources. This hands-on training will cover the followin
 g topics in computer-vision-related problems:\n\n- Introduction to transfe
 r learning in computer vision\;\n- Image classification with feature extra
 ction - using a downloadable model with pre-trained parameters for a custo
 m classification task\;\n- Image classification with fine-tuning - update 
 parameters of a pre-trained model to get better results\;\n- Demonstration
  of handling imbalanced data set for transfer learning in image classifica
 tion\n\n**Prerequisites**\n\nFor this hands-on you will need a google acco
 unt to access Google Colab service. Due to the time constraints only the s
 implest networks will be trained during the tutorial\, larger models will 
 be left to experiment with for the participants as a home assignemnt.\n\nh
 ttps://events.ncbj.gov.pl/event/141/contributions/864/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/864/
END:VEVENT
BEGIN:VEVENT
SUMMARY:EuroCC tutorial on transfer learning in computer vision - part 1
DTSTART;VALUE=DATE-TIME:20220916T123000Z
DTEND;VALUE=DATE-TIME:20220916T140000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-863@events.ncbj.gov.pl
DESCRIPTION:Speakers: Michał Obara (National Centre for Nuclear Research)
 \nTransfer learning is a machine learning (ML) technique of reusing models
  with pre-trained knowledge obtained for a general ML task\, and applying 
 it to another\, more specific ML task\, with limited training data or comp
 utational resources. This hands-on training will cover the following topic
 s in computer-vision-related problems:\n\n- Introduction to transfer learn
 ing in computer vision\;\n- Image classification with feature extraction -
  using a downloadable model with pre-trained parameters for a custom class
 ification task\;\n- Image classification with fine-tuning - update paramet
 ers of a pre-trained model to get better results\;\n- Demonstration of han
 dling imbalanced data set for transfer learning in image classification\n\
 n**Prerequisites**\n\nFor this hands-on you will need a google account to 
 access Google Colab service. Due to the time constraints only the simplest
  networks will be trained during the tutorial\, larger models will be left
  to experiment with for the participants as a home assignemnt.\n\nhttps://
 events.ncbj.gov.pl/event/141/contributions/863/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/863/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hands-on session for quantum circuits and simulation of noisy algo
 rithms - part 2
DTSTART;VALUE=DATE-TIME:20220914T142000Z
DTEND;VALUE=DATE-TIME:20220914T160000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-860@events.ncbj.gov.pl
DESCRIPTION:Speakers: Jami Rönkkö (IQM)\nQuantum algorithms are typicall
 y expressed as a quantum logic circuits\, where the qubits of a quantum co
 mputer are sequentially operated by quantum gates. These gates are quantum
  mechanical counterparts of classical logic gates such as NOT\, XOR etc. t
 hat enable more powerful computing by using quantum superposition and enta
 nglement. \n\nThis hands-on training covers the following \n\n- introducti
 on to quantum circuits\n- how to write quantum circuits with Qiskit\, IBM'
 s open-source Python library for quantum algorithms\n- how to simulate err
 ors and environmental noise during quantum algorithms\n\n**Prerequisites**
 \n\nFor this hands-on you will need to preinstall **Python** with packages
  **qiskit** and **Jupyter notebooks** for viewing and running the notebook
 s of the session. Instructions for installing qiskit can be found here:  h
 ttps://qiskit.org/documentation/getting_started.html and as a video here: 
 https://www.youtube.com/watch?v=M4EkW4VwhcI (only first 4.5 minutes are re
 levant).\n\nInstalling conda (as instructed in the above links) is not com
 pulsory\, but might help things run smoothly.  This hands-on serves as an 
 introduction to quantum computing and does not require previous experience
 .\n\nhttps://events.ncbj.gov.pl/event/141/contributions/860/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/860/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Hands-on session for quantum circuits and simulation of noisy algo
 rithms - part 1
DTSTART;VALUE=DATE-TIME:20220914T123000Z
DTEND;VALUE=DATE-TIME:20220914T140000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-859@events.ncbj.gov.pl
DESCRIPTION:Speakers: Jami Rönkkö (IQM)\nQuantum algorithms are typicall
 y expressed as a quantum logic circuits\, where the qubits of a quantum co
 mputer are sequentially operated by quantum gates. These gates are quantum
  mechanical counterparts of classical logic gates such as NOT\, XOR etc. t
 hat enable more powerful computing by using quantum superposition and enta
 nglement. \n\nThis hands-on training covers the following \n\n- introducti
 on to quantum circuits\n- how to write quantum circuits with Qiskit\, IBM'
 s open-source Python library for quantum algorithms\n- how to simulate err
 ors and environmental noise during quantum algorithms\n\n**Prerequisites**
 \n\nFor this hands-on you will need to preinstall **Python** with packages
  **qiskit** and **Jupyter notebooks** for viewing and running the notebook
 s of the session. Instructions for installing qiskit can be found here:  h
 ttps://qiskit.org/documentation/getting_started.html and as a video here: 
 https://www.youtube.com/watch?v=M4EkW4VwhcI (only first 4.5 minutes are re
 levant).\n\nInstalling conda (as instructed in the above links) is not com
 pulsory\, but might help things run smoothly.  This hands-on serves as an 
 introduction to quantum computing and does not require previous experience
 .\n\nhttps://events.ncbj.gov.pl/event/141/contributions/859/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/859/
END:VEVENT
BEGIN:VEVENT
SUMMARY:The Artificial Neural Network Model for the simulation of the airb
 orne toxin in the urbanized area
DTSTART;VALUE=DATE-TIME:20220913T091000Z
DTEND;VALUE=DATE-TIME:20220913T094000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-832@events.ncbj.gov.pl
DESCRIPTION:Speakers: Anna Wawrzynczak (National Centre for Nuclear Resear
 ch)\, Monika Berendt-Marchel (Siedlce University of Natural Sciences and H
 umanities)\nProviding a real-time working system to localize the dangerous
  contaminant source is one of the main challenges for the city’s emergen
 cy response groups. Unfortunately\, all proposed frameworks capable of est
 imating the contamination source localization based on recorded by the sen
 sors network the substance concentrations cannot work in real-time. The re
 ason is the significant computational time required by the applied dispers
 ion models.The solution might be an application of the trained Artificial 
 Neural Network (ANN) instead of the dispersion model in the reconstruction
  algorithm. To be used\, the ANN must learn to simulate airborne contamina
 nt transport. Training the ANN is computationally expensive\, but once tra
 ined\, the ANN would be a high-speed tool enabling the estimation of the c
 ontaminant concentration distribution.\n\nThis paper presents the results 
 of training the ANN to predict the time evolution of the dispersion of the
  airborne contaminant over a city domain. The spatial distribution of the 
 contaminant is the multidimensional function dependent on the weather cond
 itions (wind direction and speed)\, coordinates of the contamination sourc
 es\, the release rate\, and its duration. Wwe try to answer what topology 
 should be ANN to forecast the contaminant strength correctly at the given 
 point of the urbanized area at a given time.\n\nhttps://events.ncbj.gov.pl
 /event/141/contributions/832/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/832/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Welcome address
DTSTART;VALUE=DATE-TIME:20220913T080000Z
DTEND;VALUE=DATE-TIME:20220913T081500Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-950@events.ncbj.gov.pl
DESCRIPTION:Speakers: Krzysztof Kurek (National Centre for Nuclear Researc
 h)\nA welcome addess from Professor Krzysztof Kurek Director General of Na
 tional Centre for Nuclear Research\n\nhttps://events.ncbj.gov.pl/event/141
 /contributions/950/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/950/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Performance of GAN-based augmentation for deep learning COVID-19 i
 mage classification
DTSTART;VALUE=DATE-TIME:20220914T094000Z
DTEND;VALUE=DATE-TIME:20220914T101000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-948@events.ncbj.gov.pl
DESCRIPTION:Speakers: Oleksandr Fedoruk (National Centre for Nuclear Resea
 rch)\nOne of the biggest challenges in the deep learning application to th
 e \nmedical imaging domain is the availability of training data. A promisi
 ng \navenue to mitigate this problem is the usage of Generative Adversaria
 l \nNetworks (GAN) to generate images to increase the size of training dat
 a \nsets. A GAN is a class of unsupervised learning methods in which two \
 nnetworks (generator and discriminator) are joined by a feedback loop to \
 ncompete with each other. In this process the generator gradually learns \
 nhow to better deceive the discriminator\, on the other hand\, the \ndiscr
 iminator gets constantly better at detecting synthetic images.\n\nWe will 
 present the results of the transfer learning-based \nclassification of COV
 ID-19 chest X-ray images. The performance of \nseveral deep convolutional 
 neural network models is compared. Data \naugmentation is a typical method
 ology used in machine learning when \nconfronted with limited data set. We
  study the impact on the detection \nperformance of classical image augmen
 tations i.e. rotations\, cropping\, \nand brightness changes. Furthermore\
 , we compare classical image \naugmentation with GAN-based augmentation. A
  StyleGAN2-ADA model of \nGenerative Adversarial Networks is trained on th
 e limited COVID-19 chest \nX-ray image set.\nAfter assessing the quality o
 f generated images they are used to \nincrease the training data set\, and
  to improve the balance between classes.\n\nhttps://events.ncbj.gov.pl/eve
 nt/141/contributions/948/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/948/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Coincidence classification in the large field-of view J-PET scanne
 rs  with machine learning methods
DTSTART;VALUE=DATE-TIME:20220914T091000Z
DTEND;VALUE=DATE-TIME:20220914T094000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-949@events.ncbj.gov.pl
DESCRIPTION:Speakers: Konrad Klimaszewski (National Centre for Nuclear Res
 earch)\nIn PET medical imaging\, the reconstruction of the spatial distrib
 ution \nof the radiotracer in patient’s body is based on the photon pair
 s \ngrouped into time coincidences. Due to the limited resolution the sele
 cted\ncoincidences contain a fraction of events with a photon scattered in
  the \npatient or detector material and photons accidentally registered in
  a coincidence.\nScatters and accidentals deteriorate the final image qual
 ity.\nFor a total-body scanner\, the background level becomes a challenge.
  \nFirst\, the accidentals statistics increase roughly quadratic with the 
 \nscanner axial length. Second\, the multiply scattered photons fraction i
 s \nmore pronounced. Morover in J-PET scanner the signal registration is b
 ased on \nthe Compton scattering process\, which makes the inter-detector 
 scatters\nharder to discriminate.\n\nWe apply supervised learning models t
 o estimate the background \ncontribution. In particular\, boosted decision
  trees and deep learning \nneural networks are considered. The training an
 d test samples are based \non GATE Monte Carlo simulations. Selection of o
 ptimal feature set and feature \ntransformations is performed. Performance
 s of XGBoost\, AdaBoost and \nselected NN classifiers are compared with cu
 t-based selection criteria. \nConsidered models are compared based on effi
 ciency metrics. Finally\, \npreliminary comparison of reconstructed image 
 quality is provided.\n\nhttps://events.ncbj.gov.pl/event/141/contributions
 /949/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/949/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine learning solutions for cluster reconstruction in planar ca
 lorimeters
DTSTART;VALUE=DATE-TIME:20220916T094000Z
DTEND;VALUE=DATE-TIME:20220916T101000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-872@events.ncbj.gov.pl
DESCRIPTION:Speakers: Michał Mazurek (NCBJ)\nRun 3 of the Large Hadron Co
 llider (LHC) of the data-taking period poses unprecedented challenges to t
 he computing models used in the high-energy physics experiments of the LHC
  accelerator. Only in the LHCb experiment\, the luminosity has increased b
 y a factor of five. Recent results show that deep learning solutions techn
 iques can significantly improve the performance of the cluster reconstruct
 ion in calorimeters when high occupancy is expected. In this talk\, we wil
 l review selected results of the LHC experiments and\, in particular\, foc
 us on the investigated convolutional (CNN) and graph neural network (GNN) 
 solutions for planar\, LHCb-inspired calorimeters with hybrid granularitie
 s.\n\nhttps://events.ncbj.gov.pl/event/141/contributions/872/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/872/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Machine Learning as Applied Technology
DTSTART;VALUE=DATE-TIME:20220913T083000Z
DTEND;VALUE=DATE-TIME:20220913T091000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-865@events.ncbj.gov.pl
DESCRIPTION:Speakers: Dietmar Millinger (Twingz Development GmbH & GREX IT
  services GmbH)\nMachine learning has evolved from an academic toolkit to 
 a new computing technology that is already being used in technical and tec
 hnology-related areas. The progress has been very impressive. While curren
 t academic research focuses on the improvement of core machine learning me
 thods\, the successful application of machine learning in real-life projec
 ts still requires constant monitoring and manual tuning.\n\nThis talk will
  highlight some of these lessons already learned\, as well as some missing
  skills and points to consider in\nmachine learning projects.\n\nhttps://e
 vents.ncbj.gov.pl/event/141/contributions/865/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/865/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Statistically Learning the Next Standard Model from 
 LHC Data
DTSTART;VALUE=DATE-TIME:20220916T090000Z
DTEND;VALUE=DATE-TIME:20220916T094000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-871@events.ncbj.gov.pl
DESCRIPTION:Speakers: Wolfgang Waltenberger (HEPHY)\nDespite the large amo
 unt of data generated by the Large Hadron Collider\n(LHC)\nso far\, search
 es for new physics have not yet provided any clear evidence of beyond the 
 Standard Model (BSM) physics. Most of these experimental searches focus on
  exclusive channels\, looking for excesses in specific final states.\nHowe
 ver\, new physics could manifest as a dispersed signal over many channels.
 \nIt therefore becomes increasingly relevant to attempt a more global appr
 oach to finding out where BSM physics may hide. To this end\, we developed
  a novel statistical learning algorithm that is capable of identifying pot
 ential dispersed signals in the slew of published LHC analyses. Aiming to 
 minimize theoretical bias\, our approach is not constrained to a specific 
 BSM scenario.\nInstead\, the algorithm is tasked with building candidate "
 proto-models"\, precursor theories to the Next Standard Model (NSM)\, from
  small excesses in the data\, while at the same time remaining consistent 
 with negative results on new physics.\n\nIn this talk\, we explain the con
 cept as well as technical details of the statistical learning procedure. W
 e also present proof of concept results obtained when running the algorith
 m over our database that contains the results of 100 searches conducted at
  the LHC. Finally\, we sketch out our vision of how the NSM could then be 
 constructed from such protomodels.\n\nhttps://events.ncbj.gov.pl/event/141
 /contributions/871/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/871/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Machine learning applications in astrophysics
DTSTART;VALUE=DATE-TIME:20220916T070000Z
DTEND;VALUE=DATE-TIME:20220916T074000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-870@events.ncbj.gov.pl
DESCRIPTION:Speakers: Agnieszka Pollo (National Centre for Nuclear Researc
 h AND Jagiellonian University)\nIn the era of astronomical "big data"\, wi
 th the amount of observational \ndata ever-increasing and about to increas
 e by orders of magnitudes \nduring the next decade\, machine learning has 
 become not only a commodity \nbut also a necessity. At the same time\, the
  application of machine \nlearning methods to astrophysical problems yield
 s many specific \nchallenges. One of them is related to the fact that whil
 e the data to \nwhich we want to apply these methods are often big\, the a
 vailable \ntraining samples are small. Moreover\, they are often not reall
 y \nrepresentative\, in a way that may be difficult to quantify\, which fa
 ces \nus with a variety of extrapolation problems. More challenges are rel
 ated \nto the interpretability of the results\, given the limited informat
 ion we \ncan access. I will try to discuss the aims\, difficulties and att
 empts to \novercome them\, making use\, among other things\, of examples f
 rom the \nresearch made in our extragalactic astrophysics group in NCBJ an
 d UJ.\n\nhttps://events.ncbj.gov.pl/event/141/contributions/870/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/870/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Can AI make us see beyond the visible: Toward CE mar
 ked deep learning software for medical image analysis
DTSTART;VALUE=DATE-TIME:20220914T070000Z
DTEND;VALUE=DATE-TIME:20220914T074000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-866@events.ncbj.gov.pl
DESCRIPTION:Speakers: Jakub Nalepa (KPLabs\, Silesian University of Techno
 logy)\nWe have witnessed the unprecedented success of deep learning in vir
 tually all areas of science and industry\, with medical image analysis not
  being an exception here. Although there are a plethora of deep learning-p
 owered techniques that established the state of the art in the field\, e.g
 .\, in the context of automatic delineation of human organs and tumors in 
 various image modalities\, deploying such methods in clinical settings is 
 a challenging process. In this talk\, we will show how deep learning\, pot
 entially coupled with computational fluid dynamics\, can help uncover impo
 rtant clinical information to diagnose and monitor of the coronary artery 
 disease from CCTA\, or to analyze brain tumors from MRI. Also\, we will di
 scuss our approach for building Sens.AI – a CE marked deep learning prod
 uct for automated brain tumor analysis. We will show how to design thoroug
 h evidence-based verification and validation procedures for such technique
 s in scenarios\, in which collecting large\, heterogeneous\, and high-qual
 ity ground truth is time-consuming\, user-dependent and error prone.\n\nht
 tps://events.ncbj.gov.pl/event/141/contributions/866/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/866/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Neural Network inference on FPGA-based platforms
DTSTART;VALUE=DATE-TIME:20220915T094000Z
DTEND;VALUE=DATE-TIME:20220915T102000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-869@events.ncbj.gov.pl
DESCRIPTION:Speakers: Grzegorz Korcyl (Jagiellonian University)\nField Pro
 grammable Gate Arrays (FPGAs) offer unique features for High-Performance C
 omputing such as natural parallelism\, streamlined processing\, and dynami
 c reconfiguration creating a relatively new concept of adaptive computing.
  \nModern device capabilities\, high-level development techniques\, and ma
 rket adoption make them a powerful and interesting component for HPC hardw
 are platforms.\nIn this talk\, I will present a technology overview and cu
 rrent techniques for implementing Neural Networks on FPGA-based platforms.
 \n\nhttps://events.ncbj.gov.pl/event/141/contributions/869/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/869/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: LUMI: Europe’s most powerful supercomputer
DTSTART;VALUE=DATE-TIME:20220915T090000Z
DTEND;VALUE=DATE-TIME:20220915T094000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-858@events.ncbj.gov.pl
DESCRIPTION:Speakers: Tomasz Malkiewicz (CSC / NeIC)\nLUMI is the first pr
 e-exascale supercomputer of the EuroHPC Joint Undertaking and is now Europ
 e’s most powerful supercomputer.  Finland\, together with 9 other countr
 ies from the Nordics and central Europe\, collaboratively hosts one of the
 se systems in Kajaani\, Finland. The vast consortium of countries with an 
 established tradition in scientific computing and strong national computin
 g centers is a key asset for the successful infrastructure.\n\nThe LUMI su
 percomputer is also one of the most advanced platforms for artificial inte
 lligence. It links together computational capacity\, artificial intelligen
 ce methods (especially deep learning)\, traditional wide-scale simulation 
 and the utilization of large masses of data to simultaneously solve a sing
 le challenge.  LUMI serves as well as a platform for the development of qu
 antum technology. Quantum computers need supercomputers alongside them to 
 harness their capacity to the right targets as a part of the research proc
 ess. LUMI has so far been linked successfully with two quantum computers: 
 the Swedish QAL 9000 and the Finnish Helmi.\n\nIn this talk we will discus
 s the LUMI infrastructure and its great value and potential for the resear
 ch community.\n\nhttps://events.ncbj.gov.pl/event/141/contributions/858/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/858/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine learning for modeling mortality with respect to smog and a
 mbient air temperature.
DTSTART;VALUE=DATE-TIME:20220914T074000Z
DTEND;VALUE=DATE-TIME:20220914T081000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-857@events.ncbj.gov.pl
DESCRIPTION:Speakers: Tomasz Fruboes (National Centre for Nuclear Research
 )\nPoor air quality and its negative impact on health is currently one of 
 the civilizational problems in Poland. The aim of this study was an attemp
 t to verify and examine\, on the basis of data on the number and causes of
  deaths registered in Bielański Hospital in Warsaw\, the increase in the 
 number of deaths in Poland in January 2017 recorded by Statistics Poland.\
 n\n We analysed the data on the number and causes of deaths in the hospita
 l from 2013 to 2018 using the methods of searching for anomalies and build
 ing models of the number of deaths depending on ambient temperature and ai
 r pollution levels. \n\n We found that the increase in the number of death
 s observed in the hospital in January 2017 was caused by respiratory syste
 m-related deaths. A model utilizing air temperature is not enough to expla
 in the increase\, but adding PM10 air pollution levels to the temperature 
 model was sufficient to achieve this. Such a model attributes 8.3% of all 
 deaths observed in January 2017 to air pollution.\n\nhttps://events.ncbj.g
 ov.pl/event/141/contributions/857/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/857/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Towards the use of quantum computers in radiotherapy
DTSTART;VALUE=DATE-TIME:20220914T101000Z
DTEND;VALUE=DATE-TIME:20220914T104000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-856@events.ncbj.gov.pl
DESCRIPTION:Speakers: Samuele  Cavinato  (Università degli studi di Padov
 a)\nRadiotherapy aims at treating patients with cancer using ionising radi
 ation. However\, a key step is the optimization of the treatment. This is 
 done using an inverse-planning approach where the treatment goals are enco
 ded into a cost-function to minimize.  The latter can be either non-convex
  or non-smooth with several local minima.\n\nQuantum computers may efficie
 ntly solve this problem thanks to their inborn parallelisation ability. Th
 erefore\, in the last two years\, our group focused on the development of 
 new optimization strategies based mainly on Tensor Network Methods where\n
 \nthe classical optimization problem is mapped into an ising-type Hamilton
 ian whose ground state corresponds to the best solution to the initial pro
 blem and the optimization variables are represented in terms of qubits.\nO
 ur preliminary results show that this approach is compatible with any type
  of function and can perform at least comparably as classical optimization
  algorithms on the test functions considered.\n\nhttps://events.ncbj.gov.p
 l/event/141/contributions/856/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/856/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Comparison Study of Transformation Methods of Pet Raw Data Into Im
 ages for Classification Using Convolutional Neural Networks
DTSTART;VALUE=DATE-TIME:20220915T074000Z
DTEND;VALUE=DATE-TIME:20220915T081000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-855@events.ncbj.gov.pl
DESCRIPTION:Speakers: Paweł Konieczka (National Centre for Nuclear Resear
 ch)\nConvolutional Neural Networks (CNNs) have been effectively applied in
  many studies where crucial information about the data is embedded in the 
 order of features (e.g. images). However\, most tabular data – such as r
 aw Positron Emission Tomography (PET) data – do not assume a spatial cor
 relation between features\, and hence are unsuitable for CNNs classificati
 on. In order to use the power of CNNs (including GPU utilization) for clas
 sification purposes of non-image data\, a transformation method of 1-D vec
 tor into image has to be applied. A method comparison of transforming tabu
 lar data into input images for CNN classification will be presented. Self-
 organizing map and DeepInsight method were used in this study.\n\nhttps://
 events.ncbj.gov.pl/event/141/contributions/855/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/855/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine learning driven analysis of the calibration data for upgra
 ded LHCb Velo.
DTSTART;VALUE=DATE-TIME:20220916T081000Z
DTEND;VALUE=DATE-TIME:20220916T084000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-854@events.ncbj.gov.pl
DESCRIPTION:Speakers: Maciej Majewski (AGH University of Science and Techn
 ology)\nThe LHCb is one of the four main experiments discovering cutting-e
 dge physics at the Large Hadron Collider. The effects of the harsh conditi
 ons of constant radiation require maximum thought and care. This includes 
 research for new applications and novel algorithms that will help to under
 stand and predict the behaviour of the vertex locator detector at LHCb. Ou
 r studies include methods based on calibration data from 2012-2018 and ins
 ights for the new (pixel-based) Velo detector in the upcoming data-taking 
 runs at LHC. Those methods will be introduced to the detector monitoring s
 oftware ecosystem.\n\nhttps://events.ncbj.gov.pl/event/141/contributions/8
 54/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/854/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Multispectral Satellite Data Analysis Using Support 
 Vector Machines With Quantum Kernels
DTSTART;VALUE=DATE-TIME:20220913T134000Z
DTEND;VALUE=DATE-TIME:20220913T142000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-850@events.ncbj.gov.pl
DESCRIPTION:Speakers: Artur Miroszewski (Jagiellonian University)\nSupport
  vector machines (SVMs) are a well-established classifier effectively depl
 oyed in an array of pattern recognition and classification tasks. In this 
 work\, we consider extending classic SVMs with quantum kernels and applyin
 g them to satellite data analysis.\n\nThe design and implementation of SVM
 s with quantum kernels (hybrid SVMs) is presented. It consists of the Quan
 tum Kernel Estimation (QKE) procedure combined with a classic SVM training
  routine. The pixel data are mapped to the Hilbert space using ZZ-feature 
 maps acting on the parameterized ansatz state. The parameters are optimize
 d to maximize the kernel target alignment.\n\nWe approach the problem of c
 loud detection in satellite image data\, which is one of the pivotal steps
  in both on-the-ground and on-board satellite image analysis processing ch
 ains. The experiments performed over the benchmark Landsat-8 multispectral
  dataset revealed that the simulated hybrid SVM successfully classifies sa
 tellite images with accuracy on par with classical SVMs.\n\nhttps://events
 .ncbj.gov.pl/event/141/contributions/850/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/850/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: The Potential of Quantum Information in Machine Lear
 ning applied to Quantum Physics and Medicine
DTSTART;VALUE=DATE-TIME:20220913T120000Z
DTEND;VALUE=DATE-TIME:20220913T124000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-848@events.ncbj.gov.pl
DESCRIPTION:Speakers: Beatrix Hiesmayr (University of Vienna)\nFirstly\, I
  will enlighten how different quantum information is compared to classical
  information and how this can affect machine learning algorithms. Then I d
 iscuss different attempts to obtain quantum neural networks and their perf
 ormances. In the last part I will focus on applications of machine learnin
 g to problems in quantum physics and medicine.\n\nhttps://events.ncbj.gov.
 pl/event/141/contributions/848/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/848/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Invited talk: Deep learning image reconstruction for positron emis
 sion tomography (PET): present status and future perspectives
DTSTART;VALUE=DATE-TIME:20220915T070000Z
DTEND;VALUE=DATE-TIME:20220915T074000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-847@events.ncbj.gov.pl
DESCRIPTION:Speakers: Andrew Reader (King's College London)\nImage reconst
 ruction for positron emission tomography (PET) has been developed over man
 y decades\, starting out with filtered backprojection methods\, with advan
 ces coming from improved modelling of the data statistics and improved mod
 elling of the overall physics of the data acquisition / imaging process. H
 owever\, high noise and limited spatial resolution have remained major iss
 ues in PET\, and conventional state-of-the-art methods have exploited othe
 r medical imaging modalities (such as MRI) in order to assist in denoising
  and enhancing the spatial resolution for PET. Nonetheless\, there is a dr
 ive towards not only improving image quality\, but also to reducing the in
 jected radiation dose and reducing scanning times. While the arrival of ne
 w PET scanners\, such as total body PET (TB PET)\, is helping\, there is s
 till a need to improve the reconstruction of PET images in terms of qualit
 y and speed.\n\nDeep learning methods are forming the new frontier of rese
 arch for PET image reconstruction. They can learn the imaging physics and 
 its inverse\, learn the noise and also exploit databases of high-quality r
 eference examples\, to provide improvements in image quality. There are fo
 ur main approaches: direct full data-driven learning of reconstruction ope
 rators\, direct methods which incorporate known imaging physics\, methods 
 which integrate deep learning into existing iterative reconstruction algor
 ithms (unrolled reconstruction) and methods which exploit deep learning as
  a means of representing the images to reconstruct (e.g. the deep image pr
 ior). This talk will cover a review of these methods\, their advantages an
 d disadvantages. The outlook of current and future directions for deep lea
 rning in PET reconstruction will then be considered\, such as self-supervi
 sion\, and quantifying uncertainty.\n\nhttps://events.ncbj.gov.pl/event/14
 1/contributions/847/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/847/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Finding Strong Gravitational Lenses with Self-Attention
DTSTART;VALUE=DATE-TIME:20220916T074000Z
DTEND;VALUE=DATE-TIME:20220916T081000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-835@events.ncbj.gov.pl
DESCRIPTION:Speakers: Hareesh Thuruthipilly (National Center for Nuclear R
 esearch (NCBJ))\nThe upcoming large-scale surveys like LSST are expected t
 o find approximately $10^5$ strong gravitational lenses by analysing data 
 of many orders of magnitude larger than those in contemporary astronomical
  surveys. In this scenario\, non-automated techniques will be highly chall
 enging and time-consuming\, even if they are possible at all. We propose a
  new automated architecture based on the principle of self-attention to fi
 nd strong gravitational lenses and its advantages over convolution neural 
 networks are investigated. From our study\, we showed that self-attention-
 based models have clear advantages compared to simpler CNNs. They have hig
 hly competing performance in comparison to the current state-of-art CNN mo
 dels. Moreover\, introducing the encoder layers can also tackle the over-f
 itting problem present in the CNNs by acting as effective filters. In addi
 tion\, we have also identified some new strong lens candidates from the Ki
 lo Degree Survey (KiDS) using this new architecture.\n\nhttps://events.ncb
 j.gov.pl/event/141/contributions/835/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/835/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Machine Learning methods for simulating particle response in the Z
 ero Degree Calorimeter at the ALICE experiment\, CERN
DTSTART;VALUE=DATE-TIME:20220916T101000Z
DTEND;VALUE=DATE-TIME:20220916T104000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-826@events.ncbj.gov.pl
DESCRIPTION:Speakers: Jan Dubiński (Warsaw University of Technology)\nCur
 rently\, over 50% of the computing power at CERN’s GRID (>500 000 CPUs i
 n 170 centres)  is used to run High Energy Physics simulations. The recent
  updates at the Large Hadron Collider (LHC) create the need for developing
  more efficient simulation methods.  In particular\, there exist a demand 
 for a fast simulation of the neutron Zero Degree Calorimeter\, where exist
 ing Monte Carlo-based methods impose a significant computational burden.\n
 \n\nWe propose an alternative approach to the problem that leverages machi
 ne learning. Our solution utilises neural network classifiers and generati
 ve models to directly simulate the response of the calorimeter. In particu
 lar\, we examine the performance of variational autoencoders and generativ
 e adversarial networks\, expanding the GAN architecture by an additional r
 egularisation network and a simple\, yet effective postprocessing step.\n\
 n\nOur approach increases the simulation speed by 2 orders of magnitude wh
 ile maintaining the high fidelity of the simulation.\n\nhttps://events.ncb
 j.gov.pl/event/141/contributions/826/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/826/
END:VEVENT
BEGIN:VEVENT
SUMMARY:EuroCC - National Competence Center for HPC
DTSTART;VALUE=DATE-TIME:20220915T102000Z
DTEND;VALUE=DATE-TIME:20220915T105000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-828@events.ncbj.gov.pl
DESCRIPTION:Speakers: Marek Magrys (ACC Cyfronet AGH)\nThe Polish National
  Competence Center for HPC built within the EuroCC project will be describ
 ed during the talk. EuroCC is a pan-European project funded by the EuroHPC
  JU\, a joint initiative between the EU\, European countries and private p
 artners to develop a World Class Supercomputing Ecosystem in Europe. Polis
 h NCC is formed on the base of PLGrid Consortium members\, which include a
 ll six Polish HPC centres:  Cyfronet\, Cyfronet\, CI TASK\, ICM UW\, NCBJ\
 , PSNC\, WCNS. A short overview of the project and the current status will
  be provided\, as well as a list of currently available services and compe
 tencies provided by the NCC partners.\n\nhttps://events.ncbj.gov.pl/event/
 141/contributions/828/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/828/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Generative models for intelligent medical data analysis
DTSTART;VALUE=DATE-TIME:20220915T081000Z
DTEND;VALUE=DATE-TIME:20220915T084000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-829@events.ncbj.gov.pl
DESCRIPTION:Speakers: Kamila Kalecińska (AGH University of Science and Te
 chnology)\nBig variety of medical data types and their complex structure m
 ay be a challenge for data scientists. The process of creating the data is
  usually time-consuming\, while access to medical facilities databases is 
 limited due to privacy issues.\nGenerative models can be of great help in 
 the process of data augmentation. The presentation will contain the idea\,
  current status and results of generative models training (AutoEncoders\, 
 GANs) in order to build a tool for generating medical data: 3D medical DIC
 OM images representing the patient's geometry as well as phase space files
  necessary in the process of simulating the radiotherapy dose deposited in
  the phantom.\n\nhttps://events.ncbj.gov.pl/event/141/contributions/829/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/829/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Quantum simulations of loop quantum gravity
DTSTART;VALUE=DATE-TIME:20220913T124000Z
DTEND;VALUE=DATE-TIME:20220913T131000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-833@events.ncbj.gov.pl
DESCRIPTION:Speakers: Grzegorz Czelusta (Jagiellonian University)\nOne of 
 the possible applications of quantum computers in the near future are simu
 lations of physics. An example are quantum gravitational systems associate
 d with the Planck scale physics. Such systems are expected to be of the ma
 ny-body type\, which justifies utility of quantum computations in the anal
 ysis of their complex quantum behaviour. In this talk\, loop quantum gravi
 ty - a leading candidate for the theory of quantum gravitational interacti
 ons - is considered. In this case\, quantum geometry of space is represent
 ed by the so-called spin networks\, i.e. graphs with nodes associated with
  the "atoms of space". A construction of quantum circuits which generate s
 tates of spin networks will be presented. Furthermore\, a quantum algorith
 ms which enable projection of states on physical subspace of Hilbert space
  and determination of amplitudes of transitions between different states o
 f spin network are proposed. Results of implementation of the approach on 
 IBM superconducting quantum computers will be presented. Obtained results 
 provide building blocks for quantum simulations of complex spin networks\,
  which can give insight into the Planck scale physics in the near future.\
 n\nhttps://events.ncbj.gov.pl/event/141/contributions/833/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/833/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Biased AI image generations models and misconceptions about health
  conditions
DTSTART;VALUE=DATE-TIME:20220913T094000Z
DTEND;VALUE=DATE-TIME:20220913T101000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-836@events.ncbj.gov.pl
DESCRIPTION:Speakers: Adam Zadrożny (National Centre for Nuclear Research
 )\, Marianna Zadrożna ( )\nIn recent years we could observe more text-to-
 image models\, most notably DALLE\, DALLE-2\, and DALLE mini. Those models
  allow generating images based on user prompts. They were trained on datas
 ets of images and captions crawled from the web. But those datasets contai
 n some biases\, especially those present in media. However DALLE and DALLE
 -2 are debiased models against at least race and gender\, but DALLE-mini i
 s not which makes him possible to pick up biases from the dataset. Some of
  those biases might be linked to misconceptions in society. Like prompt 'a
 utistic child' will be 9 out of 9 cases a boy in preschool age\, but 'auti
 stic girl' will be presented as in secondary school years. \n\nIn the talk
 \, we would like to show how not debiased models text-to-image could serve
  as a tool to study misconceptions about health that exists in society.\n\
 nhttps://events.ncbj.gov.pl/event/141/contributions/836/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/836/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Welcome from the organizers
DTSTART;VALUE=DATE-TIME:20220913T081500Z
DTEND;VALUE=DATE-TIME:20220913T083000Z
DTSTAMP;VALUE=DATE-TIME:20260309T052635Z
UID:indico-contribution-141-843@events.ncbj.gov.pl
DESCRIPTION:Speakers: Wojciech Krzemien (National Centre for Nuclear Resea
 rch)\nShort communication from the workshop organizers.\n\nhttps://events.
 ncbj.gov.pl/event/141/contributions/843/
LOCATION:
URL:https://events.ncbj.gov.pl/event/141/contributions/843/
END:VEVENT
END:VCALENDAR