### 17 January 2022: JULAIN Talk by Prof. Franca Hoffmann
* Institute for Applied Mathematics, University of Bonn (Germany)
***Using continuum limits to understand data clustering and classification**
*[[details]](JULAIN Talk Franca Hoffmann 17 January 2022)
### Helmholtz Helmholtz AI Food For Thought Seminar
* Jan Ebert, Helmholtz AI consultant team, FZJ: **JAX is for Joy, Autodiff, and Xeleration - Introduction and deep dive into a GPU-compatible NumPy and SciPy**
* When: 2 December 2021, 11am
* more information [[here](https://events.hifis.net/e/8FFTSeminar)]
### INM-1 Seminar, Talk by Dr. Eva-Maria Gerstner: "Research data management - a short introduction to the topic and presentation of central services of the Central Library"
* When: 17.02.2020 1pm
* Where: INM Seminarraum, building 15.9, room 4001
### Talk by Joseph Kambeitz: "Applications of Machine Learning & Computational Modeling in Psychiatry"
Department of Psychiatry, University Hospital Cologne
* When: Thursday, 23 January 2020, 11:00am
* Where: seminar room of the Institute of Neuroscience and Medicine, building 15.9, Room 4001b
### 1nd JULAIN Journal Club
* When: Monday 16 December 10-11am
* Where: JSC, Rotunda, building 16.4, room 301
*[What to read](JULAIN Journal Club)
### Talk by Mehdi Cherti: "Optimization of scientific workflows with machine learning"
Paris Saclay University
* When: Thursday, 5 December 2019, 2:30pm
* Where: JSC, Rotunda, building 16.4, room 301
With the advances of machine learning and the acceleration of data generation in scientific domains, bridging the gap between machine learning experts and domain scientists is becoming crucial. In particular, questions such as how to encourage effective collaboration, ensure reproducibility of experiments, make fair and transparent model evaluation are important and challenging. In this talk, I will share my experience in organizing machine learning challenges and contributing to an open source challenge platform that tries to address these important questions. I will describe some challenges (e.g., in environmental science, climate science, analytical chemistry and others) I was involved in, how the code of the participants was modularized to make collaboration easier, and give an overview about the inner workings of the back-end. I will also describe other projects that I worked on and which involved the platform, in particular a project on video object detection and transfer learning.
### Talk by [Tim Kietzmann](http://www.timkietzmann.de/): "Deep neural networks as a framework for understanding the dynamic computations of the human visual system"
Donders Institute for Brain, Cognition and Behaviour, Nijmegen
* When: Tuesday, 2. December 2019, 3:00pm
* Where: JSC, Rotunda, building 16.4, room 301
This talk will describe our recent methodological advances in understanding information processing in the human brain and artificial vision systems. A central theme of our work is the combination of neuroimaging and deep learning, a powerful computational framework for obtaining models of cortical information processing and task-performing vision systems. Operating in this interdisciplinary research area, I will cover our recent work in which we demonstrated that neural network architectures with recurrent connectivity provide better models of human visual processing (estimated via representational dynamics and behavioural measurements). This insight was made possible by a novel mechanism to directly infuse brain data into large-scale recurrent
neural networks. In addition, we have shown that recurrent connectivity in artificial vision systems leads to computational benefits. Recurrence enables systems to flexibly trade-off speed for accuracy while exhibiting overall higher object recognition performance. Together, these findings suggest that recurrence is required to capture the representational dynamics of the human visual system.
### Talk by Julia Sidorova: "Towards AI capable of solving long-standing open problems in research"
Computer Science and Engineering, Blekinge Institute of Technology (BTH), Sweden
### Talk by [Mihai Petrovici](https://www.kip.uni-heidelberg.de/~mpedro/): "Computing with physics: aspects of bio-inspired artificial intelligence"
University of Heidelberg
* When: Tuesday, 19 November 2019, 1:15pm
* Where: building 14.6U, room 241
*[[Read more](Talk_Mihai_Petrovici)]
### Talk by [Frederic Johannes Effenberger](https://www.gfz-potsdam.de/staff/frederic-effenberger/sec28/): "Deep and adversarial learning with high resolution solar images for space weather applications"
Helmholtz Zentrum Potsdam, GFZ German Research Centre for Geosciences
* When: Wednesday, 30 October 2019, 2:30pm
* Where: JSC, Rotunda, building 16.4, room 301
The Solar Dynamics Observatory (SDO) offers an unprecedented, very large dataset (TBs of raw data per day) of solar images in different optical and EUV wavelength bands, capturing solar atmospheric structures in high resolution and with excellent coverage and cadence since 2010. This dataset is thus well suited to study the application of advanced machine learning techniques that require large amounts of data for training, such as deep learning approaches. Here, we present our initial plans and results of deep learning as applied to solar images and discuss issues and pathways for future research. In particular, we address the scope for generative adversarial training and convolutional neural networks for data augmentation and space weather forecasting. Since ultimately, most of the space weather phenomena originate from solar activity, detailed solar images offer an excellent opportunity to improve on our predictive capabilities and utilize a large, high quality set of information encoded in image data.
### Talk by [Emre Neftci](http://nmi-lab.org/people-apphook/eneftci/): "Data and Power Efficient Intelligence with Neuromorphic Hardware"
* When: Tuesday, 27 August 2019, 1:00pm
* Where: Jülich Supercomputing Centre, Rotunda, building 16.4, room 301
The potential of machine learning and deep learning to advance artificial intelligence is driving a quest to build dedicated systems that accelerate such workloads at a large scale and in an autonomous fashion. A natural approach is to take inspiration from the brain by building neuromorphic hardware that emulates the biological processes of the brain using digital or mixed-signal technologies.<br>
In this talk, I will present interdisciplinary approaches anchored in machine learning theory and computational neurosciences that enable the applicability of neuromorphic technologies to real-world, human-centric tasks. In particular I’ll discuss the following related challenges and their possible solutions:<br>
(1) The models and tools of deep learning apply to neuromorphic hardware, but physical implementations of neural networks call for novel, continual and local learning algorithms;<br>
(2) Neuromorphic technologies have potential advantages over conventional computers on tasks where real-time adaptability, autonomy or energy efficiency are necessary, but applications and benchmarks benefiting from these qualities are not yet identified;<br>
(3) Challenges in memory technologies, compounded by a tradition of bottom-up approaches in the field and the lack of large-scale simulation environments block the road to major breakthroughs.<br>
The recent algorithmic results I will present solve some of these challenges and pave the way toward the co-design of brain-inspired computing systems and algorithms with a mathematical viewpoint. These solutions enable the roadmap towards building a software framework for neuromorphic hardware with a Tensorflow-like workflow and leveraging the scalable, distributed, low-latency and energy- efficient nature of neuromorphic hardware.
### Talk by [Frank Rudzicz](http://www.cs.toronto.edu/~frank/): "Explainable AI — Interpretability, brains, policies and politics"
* When: 25th July 2019, at 11:00 am
* Where: INM seminar room, building 15.9, room 4001b
In this talk, we take a path through several different approaches to explainability in machine learning. First, we talk about categories of explainability, then we discuss approaches to relevance ranking in terms of engineered features and in terms of heat maps in images through deep Taylor expansion. We then provide a use case of a recent publication on using machine learning with MEG data, and suggest that explainability in brain data has room for improvement. Time permitting, we will briefly cover how explainable AI may help to overcome regulatory and cultural issues in healthcare and therefore accelerate the use of AI methods in practice.
### Talk by [Mateusz Kozinski ](https://cvlab.epfl.ch/): "Learning to segment 3D linear structures with 2D annotations"
* When: 18th July 2019, at 1:30 pm
* Where: INM seminar room, building 15.9, room 4001b
We propose a loss function for training a Deep Neural Network (DNN) to segment volumetric data, that accommodates ground truth annotations of 2D projections of the training volumes, instead of annotations of the 3D volumes themselves. In consequence, we significantly decrease the amount of annotations needed for a given training set. We apply the proposed loss to train DNNs for segmentation of vascular and neural networks in microscopy images and demonstrate only a marginal accuracy loss associated to the significant reduction of the annotation effort. The lower labor cost of deploying DNNs, brought in by our method, can contribute to a wide adoption of these techniques for analysis of 3D images of linear structures.
### Lab Visit at JSC - SimLab Neuroscience
* When: Thursday, **9 May 2019, 2 pm**
* Where: JSC, Rotunde, building 16.4, room 301
*[Read more](LabVisits)
### Talk by Martin Brenzke: "Machine Learning and Statistical Methods for the Power Exhaust in Tokamaks"
* When: Thursday, **9. May 2019, 2:30 pm**
* Where: IEK-4, building 9.1, large seminar room 107
### Course „Introduction to the usage and programming of supercomputer resources in Jülich"
### Human Brain Project (HBP) Colloquium - Panel Discussion on Machine Learning
* When: **4 October 2018 5:15-6:00pm**
* Where: **Lecture Theatre, Central Library** (building 04.7)
* Panel members: Prof. Wolfgang Maass, Prof. Simon Eickoff, Dr. Anna Kreshuk, Dr. Timo Dickscheid, Dr. Jenia Jitsev
* The program is given at http://www.fz-juelich.de/conferences/HBP-Juelich-Colloquium/EN/Programme/_node.html
### Talk by Zeynep Ataka
* Title: **Explaining and Representing Novel Concepts With Minimal Supervision**
* When: **Fri. 05. Oct. 2018, 10:30 – 11:30am**
* Where: JSC ‘Rotunde’ (Building 16.3, Room 301)
Short Bio: Dr. Zeynep Akata is an Assistant Professor with the University of Amsterdam in the Netherlands, Scientific Manager of the Delta Lab and a Senior Researcher at the Max Planck Institute for Informatics in Germany. She holds a BSc degree from Trakya University (2008), MSc degree from RWTH Aachen (2010) and a PhD degree from University of Grenoble (2014). After completing her PhD at the INRIA Rhone Alpes with Prof. Dr. Cordelia Schmid, she worked as a post-doctoral researcher at the Max Planck Institute for Informatics with Prof. Dr. Bernt Schiele and a visiting researcher with Prof Trevor Darrell at UC Berkeley. Her research interests include machine learning that combine vision and language for the task of explainable artificial intelligence (XAI).
---
[[Home](Home)]
---
last change: 10.5.2022 sw
\ No newline at end of file
[[Home](Home)]
---
### July 25, 2024: **Helmholtz AI FFT Seminar** by HZDR
* “Towards Data-Driven Optimization of Experiments in Photon Science”, Jeffrey Kelling (HZDR)
* “A Review of Neural Networks in Computational Holography”, Ritz Aguilar (HZDR)
* “A deep-learning-based surrogate model for Monte-Carlo simulations of the linear energy transfer in primary brain tumor patients treated with proton-beam radiotherapy”, Sebastian Starke (HZDR)
### April 15, 2024: JULAIN Journal Club: Computational imaging in the Era of Large-Scale Foundation Models [[read more](https://gitlab.jsc.fz-juelich.de/MLDL_FZJ/MLDL_FZJ_Wiki/-/wikis/JULAIN-Journal-Club)]
### March 18, 2024: JULAIN Journal Club: Special edition on FAIR research software development [[read more](https://gitlab.jsc.fz-juelich.de/MLDL_FZJ/MLDL_FZJ_Wiki/-/wikis/JULAIN-Journal-Club)]
### March 7, 2024 11:00: **Helmholtz AI Food For Thought Seminar** [[link](https://events.hifis.net/event/1420/)]
### Feb 28, 2024 13:00: **INM-1 and Vogt Seminar**<br>
Talk by **Carsten Marr**, Director Institute of AI for Health, Helmholtz Munich<br>
**Diagnosing severe blood diseases with AI**<br>
Building: 15.9v, Room: 4001b<br>
Or online
### 16 June 2023: JULAIN Come Together<br>
[[More information](JULAIN Come Together 19.6.23)]
### 9 June 2022, 4pm: **JULAIN Talk by Thijs Vogels**<br>
EPFL Machine Learning & Optimization Laboratory, will talk about **Communication-efficient distributed learning and PowerSGD**<br>
[[More information](JULAIN Talk Thijs Vogels 9 June 2022)]
### 17 March 2022: JULAIN Talk by Shahab Bakhtiari
* Montreal Institute for Learning Algorithms (MILA)
***Specialized parallel pathways in brains and artificial neural networks**
### 17 January 2022: JULAIN Talk by Prof. Franca Hoffmann
* Institute for Applied Mathematics, University of Bonn (Germany)
***Using continuum limits to understand data clustering and classification**
*[[details]](JULAIN Talk Franca Hoffmann 17 January 2022)
### Helmholtz Helmholtz AI Food For Thought Seminar
* Jan Ebert, Helmholtz AI consultant team, FZJ: **JAX is for Joy, Autodiff, and Xeleration - Introduction and deep dive into a GPU-compatible NumPy and SciPy**
* When: 2 December 2021, 11am
* more information [[here](https://events.hifis.net/e/8FFTSeminar)]
### INM-1 Seminar, Talk by Dr. Eva-Maria Gerstner: "Research data management - a short introduction to the topic and presentation of central services of the Central Library"
* When: 17.02.2020 1pm
* Where: INM Seminarraum, building 15.9, room 4001
### Talk by Joseph Kambeitz: "Applications of Machine Learning & Computational Modeling in Psychiatry"
Department of Psychiatry, University Hospital Cologne
* When: Thursday, 23 January 2020, 11:00am
* Where: seminar room of the Institute of Neuroscience and Medicine, building 15.9, Room 4001b
### 1nd JULAIN Journal Club
* When: Monday 16 December 10-11am
* Where: JSC, Rotunda, building 16.4, room 301
*[What to read](JULAIN Journal Club)
### Talk by Mehdi Cherti: "Optimization of scientific workflows with machine learning"
Paris Saclay University
* When: Thursday, 5 December 2019, 2:30pm
* Where: JSC, Rotunda, building 16.4, room 301
With the advances of machine learning and the acceleration of data generation in scientific domains, bridging the gap between machine learning experts and domain scientists is becoming crucial. In particular, questions such as how to encourage effective collaboration, ensure reproducibility of experiments, make fair and transparent model evaluation are important and challenging. In this talk, I will share my experience in organizing machine learning challenges and contributing to an open source challenge platform that tries to address these important questions. I will describe some challenges (e.g., in environmental science, climate science, analytical chemistry and others) I was involved in, how the code of the participants was modularized to make collaboration easier, and give an overview about the inner workings of the back-end. I will also describe other projects that I worked on and which involved the platform, in particular a project on video object detection and transfer learning.
### Talk by [Tim Kietzmann](http://www.timkietzmann.de/): "Deep neural networks as a framework for understanding the dynamic computations of the human visual system"
Donders Institute for Brain, Cognition and Behaviour, Nijmegen
* When: Tuesday, 2. December 2019, 3:00pm
* Where: JSC, Rotunda, building 16.4, room 301
This talk will describe our recent methodological advances in understanding information processing in the human brain and artificial vision systems. A central theme of our work is the combination of neuroimaging and deep learning, a powerful computational framework for obtaining models of cortical information processing and task-performing vision systems. Operating in this interdisciplinary research area, I will cover our recent work in which we demonstrated that neural network architectures with recurrent connectivity provide better models of human visual processing (estimated via representational dynamics and behavioural measurements). This insight was made possible by a novel mechanism to directly infuse brain data into large-scale recurrent
neural networks. In addition, we have shown that recurrent connectivity in artificial vision systems leads to computational benefits. Recurrence enables systems to flexibly trade-off speed for accuracy while exhibiting overall higher object recognition performance. Together, these findings suggest that recurrence is required to capture the representational dynamics of the human visual system.
### Talk by Julia Sidorova: "Towards AI capable of solving long-standing open problems in research"
Computer Science and Engineering, Blekinge Institute of Technology (BTH), Sweden
### Talk by [Mihai Petrovici](https://www.kip.uni-heidelberg.de/~mpedro/): "Computing with physics: aspects of bio-inspired artificial intelligence"
University of Heidelberg
* When: Tuesday, 19 November 2019, 1:15pm
* Where: building 14.6U, room 241
*[[Read more](Talk_Mihai_Petrovici)]
### Talk by [Frederic Johannes Effenberger](https://www.gfz-potsdam.de/staff/frederic-effenberger/sec28/): "Deep and adversarial learning with high resolution solar images for space weather applications"
Helmholtz Zentrum Potsdam, GFZ German Research Centre for Geosciences
* When: Wednesday, 30 October 2019, 2:30pm
* Where: JSC, Rotunda, building 16.4, room 301
The Solar Dynamics Observatory (SDO) offers an unprecedented, very large dataset (TBs of raw data per day) of solar images in different optical and EUV wavelength bands, capturing solar atmospheric structures in high resolution and with excellent coverage and cadence since 2010. This dataset is thus well suited to study the application of advanced machine learning techniques that require large amounts of data for training, such as deep learning approaches. Here, we present our initial plans and results of deep learning as applied to solar images and discuss issues and pathways for future research. In particular, we address the scope for generative adversarial training and convolutional neural networks for data augmentation and space weather forecasting. Since ultimately, most of the space weather phenomena originate from solar activity, detailed solar images offer an excellent opportunity to improve on our predictive capabilities and utilize a large, high quality set of information encoded in image data.
### Talk by [Emre Neftci](http://nmi-lab.org/people-apphook/eneftci/): "Data and Power Efficient Intelligence with Neuromorphic Hardware"
* When: Tuesday, 27 August 2019, 1:00pm
* Where: Jülich Supercomputing Centre, Rotunda, building 16.4, room 301
The potential of machine learning and deep learning to advance artificial intelligence is driving a quest to build dedicated systems that accelerate such workloads at a large scale and in an autonomous fashion. A natural approach is to take inspiration from the brain by building neuromorphic hardware that emulates the biological processes of the brain using digital or mixed-signal technologies.<br>
In this talk, I will present interdisciplinary approaches anchored in machine learning theory and computational neurosciences that enable the applicability of neuromorphic technologies to real-world, human-centric tasks. In particular I’ll discuss the following related challenges and their possible solutions:<br>
(1) The models and tools of deep learning apply to neuromorphic hardware, but physical implementations of neural networks call for novel, continual and local learning algorithms;<br>
(2) Neuromorphic technologies have potential advantages over conventional computers on tasks where real-time adaptability, autonomy or energy efficiency are necessary, but applications and benchmarks benefiting from these qualities are not yet identified;<br>
(3) Challenges in memory technologies, compounded by a tradition of bottom-up approaches in the field and the lack of large-scale simulation environments block the road to major breakthroughs.<br>
The recent algorithmic results I will present solve some of these challenges and pave the way toward the co-design of brain-inspired computing systems and algorithms with a mathematical viewpoint. These solutions enable the roadmap towards building a software framework for neuromorphic hardware with a Tensorflow-like workflow and leveraging the scalable, distributed, low-latency and energy- efficient nature of neuromorphic hardware.
### Talk by [Frank Rudzicz](http://www.cs.toronto.edu/~frank/): "Explainable AI — Interpretability, brains, policies and politics"
* When: 25th July 2019, at 11:00 am
* Where: INM seminar room, building 15.9, room 4001b
In this talk, we take a path through several different approaches to explainability in machine learning. First, we talk about categories of explainability, then we discuss approaches to relevance ranking in terms of engineered features and in terms of heat maps in images through deep Taylor expansion. We then provide a use case of a recent publication on using machine learning with MEG data, and suggest that explainability in brain data has room for improvement. Time permitting, we will briefly cover how explainable AI may help to overcome regulatory and cultural issues in healthcare and therefore accelerate the use of AI methods in practice.
### Talk by [Mateusz Kozinski ](https://cvlab.epfl.ch/): "Learning to segment 3D linear structures with 2D annotations"
* When: 18th July 2019, at 1:30 pm
* Where: INM seminar room, building 15.9, room 4001b
We propose a loss function for training a Deep Neural Network (DNN) to segment volumetric data, that accommodates ground truth annotations of 2D projections of the training volumes, instead of annotations of the 3D volumes themselves. In consequence, we significantly decrease the amount of annotations needed for a given training set. We apply the proposed loss to train DNNs for segmentation of vascular and neural networks in microscopy images and demonstrate only a marginal accuracy loss associated to the significant reduction of the annotation effort. The lower labor cost of deploying DNNs, brought in by our method, can contribute to a wide adoption of these techniques for analysis of 3D images of linear structures.
### Lab Visit at JSC - SimLab Neuroscience
* When: Thursday, **9 May 2019, 2 pm**
* Where: JSC, Rotunde, building 16.4, room 301
*[Read more](LabVisits)
### Talk by Martin Brenzke: "Machine Learning and Statistical Methods for the Power Exhaust in Tokamaks"
* When: Thursday, **9. May 2019, 2:30 pm**
* Where: IEK-4, building 9.1, large seminar room 107
### Course „Introduction to the usage and programming of supercomputer resources in Jülich"
### Human Brain Project (HBP) Colloquium - Panel Discussion on Machine Learning
* When: **4 October 2018 5:15-6:00pm**
* Where: **Lecture Theatre, Central Library** (building 04.7)
* Panel members: Prof. Wolfgang Maass, Prof. Simon Eickoff, Dr. Anna Kreshuk, Dr. Timo Dickscheid, Dr. Jenia Jitsev
* The program is given at http://www.fz-juelich.de/conferences/HBP-Juelich-Colloquium/EN/Programme/_node.html
### Talk by Zeynep Ataka
* Title: **Explaining and Representing Novel Concepts With Minimal Supervision**
* When: **Fri. 05. Oct. 2018, 10:30 – 11:30am**
* Where: JSC ‘Rotunde’ (Building 16.3, Room 301)
Short Bio: Dr. Zeynep Akata is an Assistant Professor with the University of Amsterdam in the Netherlands, Scientific Manager of the Delta Lab and a Senior Researcher at the Max Planck Institute for Informatics in Germany. She holds a BSc degree from Trakya University (2008), MSc degree from RWTH Aachen (2010) and a PhD degree from University of Grenoble (2014). After completing her PhD at the INRIA Rhone Alpes with Prof. Dr. Cordelia Schmid, she worked as a post-doctoral researcher at the Max Planck Institute for Informatics with Prof. Dr. Bernt Schiele and a visiting researcher with Prof Trevor Darrell at UC Berkeley. Her research interests include machine learning that combine vision and language for the task of explainable artificial intelligence (XAI).
