Intense week in which the combination of lectures and lab exercises will bring participants (1) closer to the use of deep learning and computer vision in tasks related to medical imaging and other medical data, and (2) to high-performance computing to considerably reduce the running times of model-training processes.
From here you can download the flyer to distributed it everywhere (thanks in advance).
| Info | Detail |
|---|---|
| Organised by | The DeepHealth project, an H2020 funded project with agreement number 825111 |
| Dates | January 24-28, 2022 here you can find the schedule |
| Location | Only via Zoom (link will be provided to accepted participants) |
| Target audience | PhD and Master’s degree students in AI/ML/DS and professionals from Industry with previous knowledge on Machine Learning. Experience with Linux and shell-script is recommended to all audiences types. |
| Registration | Via EventBrite in this link |
| Topics | Deep Learning, Computer Vision, Medical Imaging and High-Performance Computing |
The DeepHealth Winter School will include theoretical sessions (master classes) about the above enumerated four topics, and lab sessions to show attendants how the software created in the DeepHealth project can be installed and used.
Lab exercises will be guided by junior and senior researchers involved in the DeepHealth project.
Here you can see and download the schedule
Monica Caballero, DeepHealth project coordinator, will do a short presentation of the DeepHealth project Monday at 9-9-30 AM, then Jon Ander Gómez, DeepHealth technical manager, will do an overview of the Winter School using the contents of this GitHub repository.
-
Doing Deep Learning with the European Distributed Deep Learing Library (EDDL), by Roberto Paredes (UPV)
- Two sessions: Monday 10-11:30 AM and Tuesday 9-10 AM slides for both sessions
-
Doing Computer Vision with the European Computer Vision Library (ECVL), by Costantino Grana (UNIMORE)
- Two sessions: Monday 12-1:30 PM and Tuesday 10-11 AM slides for monday session slides for tuesday session
-
Medical Imaging: Tuesday 11:30 AM to 1:30 PM slides for the four presentations of this session
-
Introduction to medical imaging: a constant learning experience by Marco Grangetto (UNITO)
-
From H&E to pixels: digital pathology applications for colon cancer diagnosis by Luca Bertero (UNITO) and Carlo Barbano (UNITO)
-
Neural Network-derived perfusion maps in patients with acute ischemic stroke by Federico D’Agata (UNITO) and Enzo Tartaglione (UNITO)
-
Lung cancer diagnosis by Daniele Perlo (CDSS), Riccardo Renzulli (UNITO) and Marco Grosso (CDSS)
-
-
Medical Image manipulation
-
DICOM & NifTI formats by Costantino Grana (UNIMORE): Tuesday 3-3:30 PM slides
-
Deep Learning pipeline on histopathology images: detection of prostatic tumor. Data engineering by Francesco Versaci (CRS4) and learning pipeline by Giovanni Busonera (CRS4): Tuesday 3:30-4:30 PM
-
-
High-Performance Computing Wednesday 12-1:30 PM
-
Heterogeneus Architectures in EDDL
-
GPU programming in the EDDL, by Roberto Paredes (UPV): Thursday 3-4 PM
-
Reconfigurable Architectures Support in EDDL Thursday 4-6 PM, by
- José Flich (UPV) for FPGA,
- Enzo Tartaglione (UNITO) for pruning, and
- Vicent Templier (CEA) for quantization methodologies
-
-
Presentation of other ICT-11 projects:
-
HPC & Cloud Security in the LEXIS project Friday 10:30-11:30 AM by Frédéric Donnat and Barry Butler from Outpost24
-
A theoretical part consisting of an introduction to the Zero-Trust concept with the concrete use-case of LEXIS Platform and the Zero Trust Architecture applied to HPC and Cloud infrastructure, by Frédéric Donnat. slides
-
A pragmatical part where Outpost24 will introduce the importance of a full cybersecurity assessment approach to easily improve the security posture and reduce effort to identify, assess and prioritise the remediation of security vulnerabilities, by Barry Butler. slides
-
-
CYBELE: Making HPC more accessible for Agri-food Business Friday 11:30 AM - 12:30 PM by Dr. Steven Davy, Head Of Division for Programmable Autonomous Systems at Walton Institute for Information and Communication Systems Science
-
Here you can see and download the schedule
Lab sessions are open to all registered people (i.e., even for those who registered for master classes only), i.e., all attendees can run the non-distributed experiments in their own computers if equipped with one GPU at least. User accounts to run experiments by distributing the workload on HPC infrastructures will be created only for 10 working of groups of 5 people comming from the 50 attendees registered in the lab sessions.
Before the lab exercises, as the starting point of the lab sessions, we have an introduction to the ECVL & EDDL environment for potential developers, scheduled Monday afternoon from 3 PM to 5 PM, and consisting in a presentation and brief explanation of the main features of both ECVL and EDDL libraries.
Then we will continue with the sessions to carry out the lab exercises.
- Barbara Cantalupo (UNITO), Jon Ander Gómez (UPV), Costantino Grana (UNIMORE)
- Federico Bolelli (UNIMORE), Iacopo Colonnelli (UNITO), Salvador Carrión (UPV), Álvaro López (UPV), Javier Martínez (UPV)
Here you can see and download the schedule
 |
Monica Caballero (NTTData Spain) DeepHealth Project Coordinator | |
 |
Jon Ander Gómez (UPV), DeepHealth Technical Manager |
|
Marco Aldinucci (UNITO) | |
| Federico Bolelli (UNIMORE) | ||
|
Giovanni Busonera (CRS4) | |
|
Barbara Cantalupo (UNITO) | |
|
Salvador Carrión (UPV) | |
 |
Iacopo Colonnelli (UNITO) | |
 |
José Flich (UPV) | |
 |
Costantino Grana (UNIMORE) | |
 |
Marco Grangetto (UNITO) | |
 |
Marco Grosso (CDSS) | |
|
Álvaro López (UPV) | |
|
Javier Martínez (UPV) | |
|
Miquel Moreto (BSC) | |
|
Roberto Paredes (UPV) | |
 |
Daniele Perlo (CDSS) | |
|
Luca Pireddu (CRS4) | |
|
Eduardo Quiñones (BSC) | |
|
Riccardo Renzulli (UNITO) | |
 |
Enzo Tartaglione (UNITO) | |
 |
Francesco Versaci (CRS4) |