Videoreihe – Beyond the Patterns

This is a lecture series that demonstrates all kinds of topics of science within and beyond pattern recognition.

Beyond the Patterns- Teil 1: Pitfalls of Presentations

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

In this first video, I attempt to give a couple of hints on how to make good lectures, but you can see that some things go wrong.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

 

Beyond the Patterns- Teil 2: Neurobiology of Learning

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 
In this session, we have Prof. Dr. Schulze as a guest to detail the biology and neurobiology of learning and memory.

Abtract: Brains process information from sensory organs to create an internal representation of the world. To do so, information not only has to be stored, but evaluated and selected based on previous experience. This process enables organisms to control their behavior and thereby interact with the world around them. In the talk, I will describe how biological brains fulfill this complex task. I will describe how information is stored in biological neuronal networks, and how these memory functions are influenced by experience and emotions. The differences in information processing and storage between biological and computer systems will be described, and possible implications for artificial neuronal networks will be discussed.

Short Bio: Holger Schulze studied biology at the Technical University of Darmstadt, where he graduated and did his PhD on sound processing in the auditory cortex of Mongolian gerbils. In 1996 moved to the Leibniz Institute for Neurobiology in Magdeburg, where he started his own group and worked on mechanisms of learning and memory in the auditory system. In 2003 he finished his habilitation in physiology at the Medical School of the Otto-von-Guericke University in Magdeburg. In 2007 he followed a call to Erlangen, where he is an associate professor for experimental otolaryngology. His main research topics now are the neurophysiological mechanisms of hearing, tinnitus, and sleep.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

Beyond the Patterns- Teil 3: Ambient Health Intelligence

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

In this session, we have Prof. Dr. Schuller as a guest to detail the concept of Ambient Health Intelligence.

Abstract: The vision of accompanying Artificial Intelligence providing in situ health diagnosis wherever we are has long since been depicted on the big screen. More recently, however, former science fiction is steadily leaving the fiction grounds as more and more wearables feature more and more sensors with increasing features of intelligence. Already today’s smart watches and trackers monitor our heart rate, blood oxygen level, and recently also started to “listen in” equipped with microphones. But the promise held by such mobile health does not stop at recognising heart attacks, and with the IoT, sensing can embed surrounding sensors’ information leading into the era of Ambient Health Intelligence. This talk highlights the implication on the AI side facing challenges such as analysing data from largely unknown, potentially noisy and lossy sensor signals, fusion of highly asynchronous and heterogenous information, learning from little data, “green” efficient processing, or coping with uncertainty. For protection of our health data, but shared benefit from each other’s data, further considerations touch upon adversarial attacks and federated learning. Tomorrow’s ambient health intelligence may offer real-time and earlier diagnosis, and personalised therapy for all, anytime, anywhere, but it comes with great responsibility.

Short Bio: Björn W. Schuller received his diploma, doctoral degree, habilitation, and Adjunct Teaching Professor in Machine Intelligence and Signal Processing all in EE/IT from TUM in Munich/Germany. He is Full Professor of Artificial Intelligence and the Head of GLAM at Imperial College London/UK, Full Professor and Chair of Embedded Intelligence for Health Care and Wellbeing at the University of Augsburg/Germany, co-founding CEO and current CSO of audEERING – an Audio Intelligence company based near Munich and in Berlin/Germany, and permanent Visiting Professor at HIT/China amongst other Professorships and Affiliations. Previous stays include Full Professor at the University of Passau/Germany, and Researcher at Joanneum Research in Graz/Austria, and the CNRS-LIMSI in Orsay/France. He is a Fellow of the IEEE and Golden Core Awardee of the IEEE Computer Society, Fellow of the BCS, Fellow of the ISCA, President-Emeritus of the AAAC, and Senior Member of the ACM. He (co-)authored 900+ publications (30k+ citations, h-index=85), is Field Chief Editor of Frontiers in Digital Health and was Editor in Chief of the IEEE Transactions on Affective Computing amongst manifold further commitments and service to the community. His 30+ awards include having been honoured as one of 40 extraordinary scientists under the age of 40 by the WEF in 2015. He served as Coordinator/PI in 15+ European Projects, is an ERC Starting and DFG Reinhart-Koselleck Grantee, and consultant of companies such as Barclays, GN, Huawei, or Samsung.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

Beyond the Patterns- Teil 4: Shaam Al Shayah - Refugee, Master of Medical Engineering & YouTuber

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Today, I had the great opportunity to talk to Shaam Al Shayah, an alumn of FAU, who is today successful in the industry and hosts a high-quality YouTube Channel.

Short Bio: Sha-am (the name means „glory“) is a young lady from Syria who came to Germany as a refugee and found a new home. She had to leave her home country Syria because of the war. She studied in Syria, Lebanon, and Germany switching between the fields of Mechanical engineering, Electronic engineering, and computer science but always focusing on Medical Engineering to graduate finally with a master degree as a medical engineer from the University of Erlangen-Nurnberg, achieving her dreams in being an educated working woman. She is also a Youtuber who shares her experience in Germany and a founding member of Medical Data donors. She is supporting other students in Germany to establish “Mubadara” (mean initiative) a nonprofit project to support students who are willing to start/proceed their studies in German Universities. So far it is unfortunately only in Arabic but hopefully, in the future, it can grow to English too.

Shaam’s Videos:
Udacity nano degree
FastAI Demo
Living Expenses in Germany
e-Scooters in Germany (with English subtitles)
Shaam’s Socials:
Donate%Support: https://rb.gy/qbwsxg
Instagram: https://www.instagram.com/shaam.shayah/
Facebook: https://www.facebook.com/shaam.shayah.FB 
Youtube Channel: https://rb.gy/5sn64w

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

Beyond the Patterns- Teil 5: Ivana Isgum - Deep learning for Automatic Detection of Cardiovascular Disease in CT and MR exams

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It is a great pleasure to present this invited talk by Ivana Isgum from the University of Amsterdam on her great research in the field of Machine Learning and Medical Imaging:

Title: Deep learning for automatic detection of cardiovascular disease in CT and MR exams
Prof. Dr. Ivana Išgum, UMC Amsterdam, University of Amsterdam

Abstract: Deep learning has revolutionized many fields including medical imaging. Routinely acquired cardiac images provide important information for the diagnosis of cardiac disease, and image-guided therapy and intervention. In this presentation, I will show recent development of the AI methods for automatic analysis of cardiac CT and MR exams in my group. Cardiac CT allows visualization of coronary arteries. Hence, I will present our work for a fully automatic analysis of the coronary artery morphology in CT exams. Moreover, to extend the utilization potential of the CT exams, we are developing methods for quantification of cardiac function through analysis of cardiac chambers in 4D CT. Unlike CT, MR is routinely used for the quantification of cardiac function. Therefore, I will present the methods we are developing for the automation of this process. Finally, I will briefly show how we address the interpretability of the automatic decision making, quantification of uncertainty, and other issues related to the implementation of automatic AI methods.

Biography: Ivana Išgum is University Professor of AI and Medical Imaging at the Amsterdam University Medical Center, University of Amsterdam. In fall 2018 she started as Scientific Lead of the company Quantib-U. Ivana Išgum graduated in Mathematics at the University of Zagreb, Croatia in 1999. She obtained her PhD degree at the Image Sciences Institute in 2007 with a thesis titled ‘Computer-aided detection and quantification of arterial calcifications with CT’. She was a Postdoc at the Laboratory for Clinical and Experimental Image Processing in Leiden University Medical Center, and subsequently Assistant and Associate Professor at UMC Utrecht where she was leading Quantitative Medical Image Analysis (QIA) group at the Image Sciences Institute. In 2019 Ivana was appointed Full Professor and moved with her group to University of Amsterdam. Her group is focusing on the development of algorithms for quantitative analysis of medical images to enable automatic patient risk profiling, diagnosis and prognosis using AI techniques. Ivana Išgum is the recipient of several large grants, has presented extensively in medical image conferences and published in scientific peer-reviewed journals.

Links to Ivana’s Papers:

N. Lessmann et al. IEEE Trans Med Imaging. 2018;37(2):615-625
https://arxiv.org/pdf/1711.00349.pdf

Van Velzen et al. Radiology 2020 Apr;295(1):66-79
https://pubs.rsna.org/doi/10.1148/radiol.2020191621?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed

 

Wolterink et al. IEEE Trans Med Imaging. 2017 Dec;36(12):2536-2545
https://ieeexplore.ieee.org/document/7934380

 

Van Velzen et al, SPIE Medical Imaging 2020
https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11313/2549557/Coronary-artery-calcium-scoring-can-we-do-better/10.1117/12.2549557.short

 

Bruns et al. Med Phys 2020, in press

https://arxiv.org/ftp/arxiv/papers/2008/2008.03985.pdf

 

Bruns et al. SPIE Medical Imaging 2021
Not on arXiv (yet)

 

Sander et al. Sci. Rep. 2020; in press
https://arxiv.org/pdf/2011.07025.pdf

 

Sander et al. SPIE Medical Imaging 2021; in press
https://arxiv.org/pdf/2010.13172.pdf

 

 

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

Beyond the Patterns- Teil 6: Mauricio Reyes - Medical image analysis in the era of deep learning: From performance to challenging the alchemist within

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It is a great pleasure to present this invited talk by Mauricio Reyes from the University of Bern on his great research in the field of Machine Learning and Medical Imaging:

Title: Medical image analysis in the era of deep learning: From performance to challenging the alchemist within
Prof. Dr. Mauricio Reyes, University Bern

Abstract: In this talk, Dr. Reyes will present our experience in the area of deep learning-based medical image analysis, going through the classical pillars of obtaining high accuracy, through then focusing on issues preventing clinical integration, including robustness, system monitoring via human-machine interfaces, interpretability and fast active learning. The talk will focus on neuroimaging but a few examples in other areas will be provided.

Biosketch: Mauricio Reyes, conducted graduate studies in Electrical Engineering at the University of Santiago, Chile where he was awarded the best electrical engineer thesis by the Chilean Institute of Engineers School. He conducted postgraduate studies to obtain a Ph.D. in Computer Sciences with a specialization in Medical Image Analysis from INRIA, France (2006). He is an associate professor at the medical faculty of the University of Bern and is currently leading the Medical Image Analysis group at the ARTORG Center for Biomedical Engineering Research of the University of Bern. His research focuses on basic and applied machine learning technologies as well as biomedical engineering solutions to improve healthcare through medical image computation technologies. A particular strength of his research has been the emphasis on developing solutions that are designed to be integrated into the clinical workflow. Dr. Reyes has participated in several Swiss National Science Foundation projects, Commission of Technology and Innovation projects, EU-FP7 projects on computational oncology and computational anatomy, and several further projects supported by Swiss foundations. From 2006 he has secured over 7.6M EUR research funds. He has an H-index: 34, has authored over 230 articles, with over 6000 citations. His entrepreneurial work has also led to the creation of one consolidated company and the second one in its first steps.

References

Suter Y., Knecht U., Alao M., Valenzuela W., Hewer E., Schucht P., Wiest R., and Reyes M. Radiomics for glioblastoma survival analysis in pre-operative mri: Exploring feature robustness, class boundaries, and machine learning techniques. Cancer Imaging, 20(2):1-13, June 2020.

Jungo A., Balsiger F., and Reyes M. Analyzing the quality and challenges of uncertainty estimations for brain tumor segmentation. Frontiers in Neuroscience, 14:282, 2020. Jungo A. and Reyes M. Assessing reliability and challenges of uncertainty estimations for medical image segmentation. In Medical Image Computing and Computer-Assisted Intervention { MICCAI 2019 , volume In Press of Lecture Notes in Computer Science, 2019.

Silva W., Cardoso J., and Reyes M. Interpretability-guided content-based medical image retrieval. In Medical Image Computing and Computer-Assisted Intervention { MICCAI 2020, volume In Press, 2020.

Reyes M., Meier R., Pereira S., Silva C., Dahlweid FM., von Teng-Kobligk H., Summers R., and Wiest R. On the interpretability of artificial intelligence in radiology: Challenges and opportunities. Radiology: Articial Intelligence , 2(3):e190043, 2020.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: Damiano Baldoni – Thinking of You

Beyond the Patterns- Teil 7: Jong Chul Ye - GAN for Medical image Reconstruction

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to welcome Prof. Dr. Jong Chul Ye from KAIST for a presentation to our lab!

Abstract: Although deep neural networks have been widely studied for medical imaging applications, most of them are supervised learning framework that requires matched label data. Unfortunately, in many medical imaging applications, high-quality label data is often difficult to obtain, so the need for unsupervised learning is increasing. In this talk, I will mainly focus on unsupervised learning in medical image reconstruction problems such as low-dose X-ray CT, accelerated MRI, ultrasound, optics, etc., when the matched target data are not available. In particular, we introduce a recent advance of generative models, in particular optimal transport-driven CycleGAN framework, which has a strong mathematical background and can be readily incorporated with the imaging physics. The use of optimal transport driven cycleGAN for low-dose X-ray CT, CT metal artifact removal, accelerated MRI, MR motion artifact removal, ultrasound imaging artifact removal, etc., which have been pioneered by our lab, will be also introduced.

Short Bio:Jong Chul Ye is a Professor of the Dept. of Bio/Brain Engineering and Adjunct Professor at Dept. of Mathematical Sciences of Korea Advanced Institute of Science and Technology (KAIST), Korea. He received the B.Sc. and M.Sc. degrees from Seoul National University, Korea, and the Ph.D. from Purdue University, West Lafayette. Before joining KAIST, he was a Senior Researcher at Philips Research, GE Global Research in New York, and a postdoctoral fellow at the University of Illinois at Urbana Champaign. He has served as an associate editor of IEEE Trans. on Image Processing, IEEE Trans. on Computational Imaging, and an editorial board member for Magnetic Resonance in Medicine. He is currently an associate editor for IEEE Trans. on Medical Imaging, and a Senior Editor of IEEE Signal Processing Magazine. He is an IEEE Fellow, Chair of IEEE SPS Computational Imaging TC, and IEEE EMBS Distinguished Lecturer. He was a General Co-chair for 2020 IEEE Symp. On Biomedical Imaging (ISBI) (with Mathews Jacob). His current research focus is deep learning theory and algorithms for various imaging reconstruction problems in x-ray CT, MRI, optics, ultrasound, remote sensing, etc.

References
Continuous Conversion of CT Kernel using Switchable CycleGAN with AdaIN
S Yang, EY Kim, JC Ye
arXiv preprint arXiv:2011.13150 2020

Unsupervised MR Motion Artifact Deep Learning using Outlier-Rejecting Bootstrap Aggregation
G Oh, JE Lee, JC Ye
arXiv preprint arXiv:2011.06337

Unpaired Training of Deep Learning tMRA for Flexible Spatio-Temporal Resolution
E Cha, H Chung, EY Kim, JC Ye
IEEE Transactions on Medical Imaging

Unpaired deep learning for accelerated MRI using optimal transport driven cycleGAN
G Oh, B Sim, HJ Chung, L Sunwoo, JC Ye
IEEE Transactions on Computational Imaging 6, 1285-1296

AdaIN-Switchable CycleGAN for Efficient Unsupervised Low-Dose CT Denoising
J Gu, JC Ye
arXiv preprint arXiv:2008.05753

Two-Stage Deep Learning for Accelerated 3D Time-of-Flight MRA without Matched Training Data
H Chung, E Cha, L Sunwoo, JC Ye
arXiv preprint arXiv:2008.01362 2020

Cyclegan with a blur kernel for deconvolution microscopy: Optimal transport geometry
S Lim, H Park, SE Lee, S Chang, B Sim, JC Ye
IEEE Transactions on Computational Imaging 6, 1127-1138

Unsupervised CT Metal Artifact Learning using Attention-guided beta-CycleGAN
J Lee, J Gu, JC Ye
arXiv preprint arXiv:2007.03480

Assessing the importance of magnetic resonance contrasts using collaborative generative adversarial networks
D Lee, WJ Moon, JC Ye
Nature Machine Intelligence 2 (1), 34-42

Optimal transport, cyclegan, and penalized ls for unsupervised learning in inverse problems
B Sim, G Oh, S Lim, JC Ye
arXiv preprint arXiv:1909.12116

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)

 

Beyond the Patterns- Teil 8: Mathias Unberath - Bridging Domains in Medical Imaging — Differentiable Mappings Between 2- and 3-Dimensional Data Domains

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to welcome Prof. Dr. Mathias Unberath back to FAU.

Abstract: Differentiably connecting 2- and 3-dimensional domains is of substantial interest in medical imaging as it enables transformational image processing techniques that substantially add value without disrupting clinical workflow. In this talk, I will introduce our recent advances in dense 3D reconstruction and differentiable rendering using examples in endoscopic and X-ray-guided surgery.

Short Bio: Mathias Unberath is an Assistant Professor in the Department of Computer Science at Johns Hopkins University with affiliations to the Laboratory for Computational Sensing and Robotics and the Malone Center for Engineering in Healthcare. He has created and is leading the Advanced Robotics and Computationally AugmenteD Environments (ARCADE) Lab that conducts research at the intersection of computer vision, machine learning, augmented reality, robotics, and medical imaging to develop collaborative systems that assist clinical professionals across the healthcare spectrum.

Previously, Mathias was an Assistant Research Professor in Computer Science and postdoctoral fellow in the Laboratory for Computational Sensing and Robotics at Hopkins and completed his Ph.D. in Computer Science at the Friedrich-Alexander-Universität Erlangen-Nürnberg from which he graduated summa cum laude in 2017. While completing a Bachelor’s in Physics and Master’s in Optical Technologies at FAU Erlangen, Mathias studied at the University of Eastern Finland as an ERASMUS scholar in 2011 and joined Stanford University as a DAAD fellow throughout 2014.

Mathias has published more than 80 journal and conference articles and has received numerous awards, grants, and fellowships, including the NIH NIBIB R21 Trailblazer Award.

References

  • Gao, C., Liu, X., Gu, W., Killeen, B., Armand, M., Taylor, R., & Unberath, M. (2020). Generalizing Spatial Transformers to Projective Geometry with Applications to 2D/3D Registration. MICCAI 2020.
  • Gu, W., Gao, C., Grupp, R., Fotouhi, J., & Unberath, M. (2020, October). Extended Capture Range of Rigid 2D/3D Registration by Estimating Riemannian Pose Gradients. In International Workshop on Machine Learning in Medical Imaging (pp. 281-291). Springer, Cham.
  • Grupp, R. B., Unberath, M., Gao, C., Hegeman, R. A., Murphy, R. J., Alexander, C. P., … & Taylor, R. H. (2020). Automatic annotation of hip anatomy in fluoroscopy for robust and efficient 2D/3D registration. International Journal of Computer Assisted Radiology and Surgery, 1-11.
  • Liu, X., Sinha, A., Ishii, M., Hager, G. D., Reiter, A., Taylor, R. H., & Unberath, M. (2019). Dense depth estimation in monocular endoscopy with self-supervised learning methods. IEEE transactions on medical imaging39(5), 1438-1447.
  • Liu, X., Zheng, Y., Killeen, B., Ishii, M., Hager, G. D., Taylor, R. H., & Unberath, M. (2020). Extremely Dense Point Correspondences using a Learned Feature Descriptor. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 4847-4856).
  • Liu, X., Stiber, M., Huang, J., Ishii, M., Hager, G. D., Taylor, R. H., & Unberath, M. (2020). Reconstructing Sinus Anatomy from Endoscopic Video–Towards a Radiation-free Approach for Quantitative Longitudinal Assessment. MICCAI 2020.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 9: Manami Sasaki - Cosmic Structures

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to welcome Prof. Dr. Manami Sasaki for an invited talk on Cosmic Structures

Abstract: The universe contains structures on all scales, from large systems of clusters of galaxies to smallest systems like stars and planets. The study of these organized structures allows us to understand how the matter in the universe is distributed and how it has developed, how galaxies have formed and changed, how stars form and evolve in galaxies, and how the different types of structures influence each other. I will present examples of these cosmic structures and what observations and numerical simulations reveal about their properties and evolution.

Short Bio: Manami Sasaki studied Physics at the Ruprecht-Karls Universität Heidelberg. After working as a doctoral student at the Max-Planck-Institute for extraterrestrial Physics in Garching, where she studied the X-ray source population in the largest satellite galaxies of the Milky Way, the Magellanic Clouds, she obtained the degree of Dr. rer. nat. in Astronomy at the Ludwig-Maximilians-Universität München.She continued her research in Astronomy and Astrophysics as a postdoc at the Harvard-Smithsonian Center for Astrophysics in Cambridge, USA, and as an Emmy Noether junior research group leader at the Eberhard Karls Universität Tübingen, before she became a professor for Multiwavelength Astronomy at the FAU.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)

Beyond the Patterns- Teil 10: Markus Haltmeier - Learned Analysis and Synthesis Regularisation of Inverse Problems

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to welcome Prof. Dr. Markus Haltmeier for an invited talk on Image Reconstruction.

Abstract: Inverse problems consist of finding accurate approximations for the unknown. Unknown x from noisy data y = A (x) + b, where A is the so-called forward operator, b represents the noise (data perturbation) and y are given noisy data. The characteristic property of inverse problems is their ill-posedness, which means that A(x) = y has no unique solution or the solution depends unstably on the given data. To obtain stable and accurate solutions, regularisation methods incorporate additional available information about the unknown and the noise.  In this talk, we review classical frame-based analysis and synthesis regularisation. We then present recent extensions that use neural networks as nonlinear synthesis and analysis operators. A mathematical analysis is given, possible training strategies are discussed and connections to related work are presented. This talk is based on:
[1] H. Li, J. Schwab, S. Antholzer, M. Haltmeier, M. NETT: Solving inverse problems with deep neural networks. Inverse Problems 36, 2020.
[2] D. Obmann, L. Nguyen, J. Schwab, M. Haltmeier. Sparse l^q-regularization of inverse problems with deep learning, arXiv:1908.03006 [math.NA], 2020.
[3] D. Obmann, J. Schwab, M. Haltmeier. Deep synthesis regularization of inverse problems, Inverse Problems 37, 2021.

Short Bio: Markus Haltmeier received his Ph.D. in mathematics from the University of Innsbruck, Tyrol, Austria, in 2007 for his work on computed tomography. He then worked as a researcher at the University of Innsbruck, the University of Vienna, Austria, and the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany, on various aspects of inverse problems. Since 2012, he is a full professor at the Department of Mathematics, University of Innsbruck. His current research interests include inverse problems, regularisation theory, signal and image processing, computed tomography, photoacoustic imaging, and machine learning.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

 

Beyond the Patterns- Teil 11: Matthias Niessner - 3D Semantic Scene Understanding

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It is a great pleasure to announce a new guest presentation by Matthias Niessner from TU Munich!

Abstract: In recent years, commodity 3D sensors, such as the Microsoft Kinect, have become easily and widely available. These advances in sensing technology have inspired significant interest in using the captured data for mapping and understanding 3D environments. In this talk, I will present our current research in this fascinating field, show potential future research directions, and talk about long-term goals. More specifically, I will show how we can now easily obtain a 3D reconstruction of an environment, and how we can exploit these results in order to infer semantics of a scene.

Bio: Dr. Matthias Nießner is a Professor at the Technical University of Munich where he leads the Visual Computing Lab. Before, he was a Visiting Assistant Professor at Stanford University. Prof. Nießner’s research lies at the intersection of computer vision, graphics, and machine learning, where he is particularly interested in cutting-edge techniques for 3D reconstruction, semantic 3D scene understanding, video editing, and AI-driven video synthesis. In total, he has published over 70 academic publications, including 22 papers at the prestigious ACM Transactions on Graphics (SIGGRAPH / SIGGRAPH Asia) journal and 24 works at the leading vision conferences (CVPR, ECCV, ICCV); several of these works won best paper awards, including at SIGCHI’14, HPG’15, SPG’18, and the SIGGRAPH’16 Emerging Technologies Award for the best Live Demo.

Prof. Nießner’s work enjoys wide media coverage, with many articles featured in main-stream media including the New York Times, Wall Street Journal, Spiegel, MIT Technological Review, and many more, and his was work led to several TV appearances such as on Jimmy Kimmel Live, where Prof. Nießner demonstrated the popular Face2Face technique; Prof. Nießner’s academic Youtube channel currently has over 5 million views.

For his work, Prof. Nießner received several awards: he is a TUM-IAS Rudolph Moessbauer Fellow (2017 – ongoing), he won the Google Faculty Award for Machine Perception (2017), the Nvidia Professor Partnership Award (2018), as well as the prestigious ERC Starting Grant 2018 which comes with 1.500.000 Euro in research funding; in 2019, he received the Eurographics Young Researcher Award honoring the best upcoming graphics researcher in Europe.

In addition to his academic impact, Prof. Nießner is a co-founder and director of Synthesia Inc., a brand-new startup backed by Marc Cuban, whose aim is to empower storytellers with cutting-edge AI-driven video synthesis.

Links:
http://www.scan-net.org/
http://niessnerlab.org/

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 12: Xin Lai - Network- and systems-based re-engineering of dendritic cells with microRNAs for cancer immunotherapy

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to announce Dr. Xin Lai from Erlangen’s University Clinic!

Abstract: Dendritic cells (DCs) are professional antigen-presenting cells that induce and regulate adaptive immunity by presenting antigens to T cells. Due to their coordinative role in adaptive immune responses, DCs have been used as cell-based therapeutic vaccination against cancer. The capacity of DCs to induce a therapeutic immune response can be enhanced by re-wiring of cellular signalling pathways with microRNAs (miRNAs).

Since the activation and maturation of DCs is controlled by an interconnected signalling network, we deploy an approach that combines RNA sequencing data and systems biology methods to delineate miRNA-based strategies that enhance DC-elicited immune responses.

Through RNA sequencing of IKKβ-matured DCs that are currently being tested in a clinical trial on therapeutic anti-cancer vaccination, we identified 44 differentially expressed miRNAs. According to a network analysis, most of these miRNAs regulate targets that are linked to immune pathways, such as cytokine and interleukin signalling. We employed a network topology-oriented scoring model to rank the miRNAs, analysed their impact on immunogenic potency of DCs, and identified dozens of promising miRNA candidates, with miR-15a and miR-16 as the top ones. The results of our analysis are presented in a database that constitutes a tool to identify DC-relevant miRNA-gene interactions with therapeutic potential (www.synmirapy.net/dc-optimization).

Our approach enables the systematic analysis and identification of functional miRNA-gene interactions that can be experimentally tested for improving DC immunogenic potency.

Short Bio: Xin Lai is a research fellow at University of Erlangen-Nürnberg. He obtained his doctorate in systems biology at University of Rostock in 2013. His research makes use of methods from biomathematics and bioinformatics to analyze biological data. He developed and proposed a systems biology approach to identify therapeutic microRNAs in cancer. His interests broadened into human genomics, network biology, and computational modelling through lecturing and from tutoring engineer and medical students from primary research papers. More than 10-year collaboration with experimental researchers led him to realize the importance of a systematic and integrative overview of biomedical research. As a researcher, he is now using this perspective to conduct research in medical systems biology. In addition, he is a father with a great passion for basketball, photography, and cooking.

Further Reading
https://www.amazon.com/Introduction-Systems-Biology-Mathematical-Computational-dp-1439837171/dp/1439837171/ref=dp_ob_title_bk
https://www.amazon.com/Network-Medicine-Complex-Systems-Therapeutics/dp/0674436539

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 13: Julio Vera Gonzalez - Computational modelling of gene regulatory networks in cancer

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It’s a great pleasure to announce Dr. Julio Vera Gonzalez from Erlangen’s University Clinic!

Abstract: Multifactorial diseases like cancer are not controlled by single genes, but instead by dense networks of interacting genes, proteins and RNAs. Comprehensive reconstruction and analysis of these networks requires quantitative data and network analysis algorithms.

In addition, these networks are enriched in feedback and feedforward loops that make their dynamics complex and highly non-linear. One can use different types of computational models to simulate the dynamics of the networks. The information obtained can be used to delineate gene signatures that predict for cancer aggressiveness or resistance to therapy.

In this talk we discuss these concepts and illustrate the ideas with published case studies.

Short Bio: Prof. Dr. Julio Vera is a physicist working in medical systems biology since 2005. Since 2013 he is professor of Systems Tumor Immunology at the UK Erlangen and FAU Erlangen-Nürnberg. His expertise is in mathematical modelling, bioinformatics and network biology. He applies multi-criteria decision algorithms to biomedicine.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 14: Isabel Valera - Ethical Machine Learning: Mind the Assumptions!

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It is a great pleasure to announce Isabel Valera from Max Planck Institute (MPI) Saarbrücken as invited speaker at our lab!

Title: “Ethical ML: mind the assumptions”
Speaker: Prof. Dr. Isabel Valera (Saarland University & Max Planck Institute for Software Systems)
Website: https://ivaleram.github.io/

Abstract: As automated data analysis supplements and even replaces human supervision in consequential decision-making (e.g., pretrial bail and loan approval), there are growing concerns from civil organizations, governments, and researchers about potential unfairness and lack of transparency of these algorithmic systems. To address these concerns, the emerging field of ethical machine learning has focused on proposing definitions and mechanisms to ensure the fairness and explicability of the outcomes of these systems. However, as we will discuss in this work, existing solutions are still far from being perfect and encounter significant technical challenges. Specifically, I will show that, in order for ethical ML, it is essential to have a holistic view of the system, starting from the data collection process before training, all the way to the deployment of the system in the real-world. Wrong technical assumptions may indeed come at a high social cost.As an example, I will first  focus on my recent work on  both fair algorithmic decision-making, and algorithmic recourse. In particular, I will show that algorithms may indeed amply the existing unfairness level in the data, if their assumptions do not hold in practice. Then, I will focus on algorithmic recourse, which aims to guide individuals affected by an algorithmic decision system on how to achieve the desired outcome. In this context, I will discuss the inherent limitations of counterfactual explanations, and argue for a shift of paradigm from recourse via nearest counterfactual explanations to recourse through interventions, which directly accounts for the underlying causal structure in the data. Finally, we will then discuss how to achieve recourse in practice when only limited causal information is available.

Short Bio: Prof. Valera is a full Professor on Machine Learning at the Department of Computer Science of Saarland University in Saarbrücken (Germany), and Adjunct Faculty at MPI for Software Systems in Saarbrücken (Germany).

She is a fellow of the European Laboratory for Learning and Intelligent Systems ( ELLIS), where she is part of the Robust Machine Learning Program and of the Saarbrücken Artificial Intelligence & Machine learning (Sam) Unit.

Prior to this, she was an independent group leader at the MPI for Intelligent Systems in Tübingen (Germany). She has held a German Humboldt Post-Doctoral Fellowship, and a “Minerva fast track” fellowship from the Max Planck Society. Prof. Valera obtained her PhD in 2014 and MSc degree in 2012 from the University Carlos III in Madrid (Spain), and worked as postdoctoral researcher at the MPI for Software Systems (Germany) and at the University of Cambridge (UK).

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 15: Gary Marcus - The Next Decade in AI

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

We had the great pleasure to have Gary Marcus as an invited speaker in our series of talks to present his ideas on the next decade in AI.

Abstract: Recent research in artificial intelligence and machine learning has largely emphasized general-purpose learning and ever-larger training sets and more and more compute. In contrast, I propose a hybrid, knowledge-driven, reasoning-based approach, centered around cognitive models, that could provide the substrate for a richer, more robust AI than is currently possible.

Short Bio: Gary Marcus is CEO and Founder of Robust AI, well-known machine learning scientist and entrepreneur, author, and Professor Emeritus at New York State University.

Dr. Marcus attended Hampshire College, where he designed his own major, cognitive science, working on human reasoning. He continued on to graduate school at Massachusetts Institute of Technology, where his advisor was the experimental psychologist Steven Pinker. He received his Ph.D. in 1993.

His books include The Algebraic Mind: Integrating Connectionism and Cognitive Science, The Birth of the Mind: How a Tiny Number of Genes Creates the Complexities of Human Thought, Kluge: The Haphazard Construction of the Human Mind, a New York Times Editors‘ Choice, and Guitar Zero, which appeared on the New York Times Bestseller list. He edited The Norton Psychology Reader, and was co-editor with Jeremy Freeman of The Future of the Brain: Essays by the World’s Leading Neuroscientist, which included Nobel Laureates May-Britt Moser and Edvard Moser. Together with Ernie Davis, he authored Rebooting AI and is well known to deconstruct myths of the AI community.

In 2014, he founded Geometric Intelligence, a machine learning company. It was acquired by Uber in 2016. In 2019, he founded Robust AI and acts currently as Robust AI’s CEO.

Links:
http://rebooting.ai
https://arxiv.org/abs/2002.06177

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference:
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 16: Essam Rashed - Human head models with deep learning enabled dielectric

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Dr. Essam Rashed has been working in many fields of medical image processing. Therefore, it’s a great pleasure to host him as a virtual guest in our lab!

Abstract: Transcranial magnetic stimulation (TMS) is a commonly used clinical procedure for neurophysiological characterization. Personalized TMS requires a pipeline for individual head model generation to provide target-specific brain stimulation. This process includes intensive segmentation of several head tissues based on MRI data, which has significant potential for segmentation error, especially for low-contrast tissues. Uniform electrical dielectric properties are assigned to each tissue in the model, which is an unrealistic assumption based on conventional volume conductor modeling. In this talk, I will briefly highlight this problem and discuss new approaches for fast and automatic estimation of the dielectric properties in the human head models without anatomical segmentation.

Short Bio: Essam Rashed received his B.Sc. in Scientific Computing in 1998 and M.Sc. in Computer Science in 2002, both from Suez Canal University, Ismailia, Egypt. He received Ph.D. (Eng.) in Computer Science from the University of Tsukuba, Tsukuba, Japan in 2010. From 2010 to 2012, he was a Research Fellow of the Japan Society for the Promotion of Science (JSPS) at the University of Tsukuba, Japan. He served as Assistant Professor of Computer Science at the Department of Mathematics, Faculty of Science, Suez Canal University from 2012 to 2016. Since then, he was promoted to Associate Professor at Suez Canal University, Egypt and worked at Faculty of Informatics and Computer Science, The British University in Egypt on Secondment. Currently, he is a Research Associate Professor at Nagoya Institute of Technology. His research interests include medical image processing, data analysis and visualization, deep learning and pattern recognition. Dr. Rashed is  IEEE Senior Member and Associate Editor of IEEE Access. He is a recipient of the Egyptian National Doctoral Scholarship (2006), JSPS postdoctoral fellowship (2010), JAMIT best presentation award (2008 & 2012), and Chairman Award, Department of Computer Science, University of Tsukuba (2010). He participated as a PI and CoI for several external funded projects.

References:
https://www.dropbox.com/s/r1d9skmm3yzrdtg/Rashed_NI2019.pdf?dl=0
https://www.dropbox.com/s/a67nq358dsc6u23/Rashed_NN2020.pdf?dl=0
https://www.dropbox.com/s/tyeyxtj69xokq1t/Rashed_PMB2020.pdf?dl=0
https://www.dropbox.com/s/eg1fk6pnaqibebb/Rashed_TMI2020.pdf?dl=0

Essam on github: 
https://github.com/erashed/ 

ForkNet available from Mathematica: 
https://resources.wolframcloud.com/NeuralNetRepository/resources/ForkNet-Brain-Segmentation-Net-Trained-on-NAMIC-Data

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 17: Emil Sidky - Inverse problems in imaging and evidence for solution by convolutional neural networks

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

 

It would have been great to welcome Emil to the Bergkirchweih this year. Unfortunately, the festival was cancelled. Yet, we still have the pleasure to have Emil virtually here in Erlangen!

Abstract: This talk examines the claim made in the literature that ill-posed inverse problems associated with image reconstruction in computed tomography (CT) can be solved with a convolutional neural network (CNN). To lay the groundwork, a brief overview of inverse problems will be given including a discussion on what makes an inverse problem ill-posed and what constitutes its solution. Examples of how inverse problem investigations play a role in CT image reconstruction will be presented in order to appreciate the value of the generalizable knowledge gained in such studies. Having set the stage, the talk will the discuss the evidence that deep-learning with convolutional neural networks solve the CT inverse problem. Finally, I will cover our own investigation into the use of CNNs to solve the sparse-view CT inverse problem in the context of a breast CT simulation.

Short Bio: Dr. Sidky is Research Professor in the Department of Radiology at The University of Chicago. He received his B.S. degree (1988) in Physics, Astronomy-Physics, and Mathematics from the University of Wisconsin-Madison. He went on to obtain his M.S (1991) and Ph.D (1993) in Physics from The University of Chicago. Dr. Sidky worked as a post-doctoral research assistant in Atomic Physics at the University of Copenhagen (1993-1996), University of Bielefeld (1996), and Kansas State University (1996-2001). In 2001, Dr. Sidky switched to medical imaging and joined the lab of Dr. Xiaochuan Pan; most recently, he was promoted to Research Professor in 2018. Dr. Sidky has published approximately 100 papers, and about 70 of them are in medical imaging. His theoretical work has mainly focused on X-ray tomography with sparse or limited-angular range sampling. He has also applied advanced techniques for non-smooth or non-convex large-scale optimization applied to imaging. His application work has centered on tomographic breast imaging, CT and tomosynthesis, and developing image reconstruction algorithms and calibration techniques for spectral CT scanners based on photon-counting detectors.

Challenge Website

Paper

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 18: Dr. Mike Kestemont - Ecology and Cultural Heritage: Modelling the Historic Survival of Books and Authors with Unseen Species Models

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Dr. Mike Kestemont is a long-term collaborator of FAU with respect to Digital Humanities. We finally managed to get him for an invited presentation on his latest research:

Abstract: In this talk, I will report on recent advances in applying quantitative methods from ecology to data from the cultural heritage domain, in particular historic literature. With biodiversity under global threat, ecologists rely on unseen species models to monitor species richness and account for the unobserved species in a sample. I hope to demonstrate that similar bias mitigation strategies are useful in the historic study of culture, which is prone to survivorship bias in the face of the incomplete survival of sources. In collaborative work, we have applied established estimators from ecology to model the loss of chivalric narrative fiction from medieval Europe, including the well-known courtly romances about King Arthur. In more recent work, we explore to which other kinds of heritage data these methods can be applied, such as the number of premodern authors that were not saved from oblivion. This work has been carried out with multiple co-authors, in particular dr. Folgert Karsdorp (Meertens Institute Amsterdam), who will be duly credited in the talk.

Short Bio: Mike Kestemont, PhD, is associate research professor in the department of Literature at the University of Antwerp (Belgium). He specializes in computational text analysis for the Computational Humanities. His work has a strong focus on historic literature and his previous research has covered a wide range of topics in literary history, including classical, medieval, early modern and modernist texts. Together with Folgert Karsdorp and Allen Riddell he has just published a textbook on data science for the Humanities with Princeton University Press. Mike recently took up an interest in ecology and how its quantitative methods can be meaningfully applied in the study of culture. Mike lives in Brussels (www.mike-kestemont.org), tweets in English (@Mike_Kestemont) and codes in Python (https://github.com/mikekestemont).

M. Kestemont & F. Karsdorp, ‚Estimating the Loss of Medieval Literature with an Unseen Species model from Ecodiversity‘. Computational Humanities Research Workshop. Amsterdam [online], 18-20 november 2020. https://zenodo.org/record/4030681#.YElLeC1XYUE

Humanities Data Analysis: Case Studies With Python, Folgert Karsdorp, Mike Kestemont, and Allen Riddell. A practical guide to data-intensive humanities research using the Python programming language. https://press.princeton.edu/books/hardcover/9780691172361/humanities-data-analysis

Novels by Mike Kestemont:

De zwarte koning, https://www.amazon.de/zwarte-koning-Michael-Kestemont/dp/9401458685
De witte weduwe, https://www.amazon.de/-/en/Michael-Kestemont/dp/9401467870

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 19: Timothy Odonga – Fairness of Classifiers Across Skin Tones in Dermatology

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

It is a great pleasure to announce Timothy Odonga as speaker at our lab! Timothy will present his latest research on fairness of classifiers that was already featured on MICCAI and NeurIPS Fair ML for Health.

Abstract: Recent advances in computer vision have led to breakthroughs in the development of automated skin image analysis. However, no attempt has been made to evaluate the consistency in performance across populations with varying skin tones. In this paper, we present an approach to estimate skin tone in skin disease benchmark datasets and investigate whether model performance is dependent on this measure. Specifically, we use Individual Typology Angle (ITA) to approximate skin tone in dermatology datasets. We look at the distribution of ITA values to better understand skin color representation in two benchmark datasets: 1) the ISIC 2018 Challenge dataset, a collection of dermoscopic images of skin lesions for the detection of skin cancer, and 2) the SD-198 dataset, a collection of clinical images capturing a wide variety of skin diseases. To estimate ITA, we first develop segmentation models to isolate non diseased areas of skin. We find that the majority of the data in the two datasets have ITA values between 34.5 and 48, which are associated with lighter skin, and is consistent with under-representation of darker skinned populations in these datasets. We also find no measurable correlation between accuracy of machine learning models and ITA values, though more comprehensive data is needed for further validation.

Short Bio: Timothy holds a master’s degree in Electrical and Computer Engineering from Carnegie Mellon University, and two bachelor’s degrees in Physics and Electrical Engineering from Gordon College and the University of Southern California, respectively. He has experience working on research projects in machine learning at CMU and IBM Research. During his time at IBM Research, he was an IBM Great Minds scholar and an AI for Social Good fellow as he worked on a project on AI Fairness in dermatology. The research papers from this work were accepted and published in the MICCAI 2020 conference, and the NeurIPS Fair ML for Health Workshop in 2019. His research interests include machine learning for healthcare focusing on topics like fairness, explainability, and causality

Paper at MICCAI 2020: Fairness of Classifiers Across Skin Tones in Dermatology – https://link.springer.com/chapter/10.1007/978-3-030-59725-2_31

NeurIPS Fair ML for Health Workshop with Timothy’s Paper – https://www.fairmlforhealth.com/accepted-papers

 

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 20: Julia Schnabel - Smart Medical Imaging – from Sensors to Information

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Prof. Dr. Schnabel has been shaping the world of medical imaging like few other persons. It’s great pleasure to have her here in Erlangen for an invited talk!

Abstract:  Medical imaging spans the entire process from acquisition, reconstruction, and quality control to image segmentation, classification, and interpretation. Recent years have increasingly seen the use of machine learning and deep learning architectures along the entire imaging pipeline, providing innovative end-to-end learning solutions that can operate directly on the imaging sensor during image acquisition, for online interpretation by the clinician.  In this talk I will focus on some recently developed “smart” medical imaging approaches applied to imaging problems in three major healthcare challenges: cancer, cardiovascular disease, and premature birth. I will specifically focus on physically and biologically realistic data augmentation, as well as real-time applications of our methods during scan-time, showing promise in image interpretation tasks that are typically only performed further down-stream, but that can equally contribute to achieving better image quality and more robust extraction of clinically relevant information.

Short Bio: Julia Schnabel graduated with an MSc in Computer Science at Technical University of Berlin (1993) and a PhD in Computer Science at University College London (1998), and subsequently held post-doctoral positions at University College London, King’s College London and University Medical Center Utrecht, before becoming first Associate Professor (2007) and then Full Professor (2014) of Engineering Science at the University of Oxford. In 2015 she joined King’s College London as Chair in Computational Imaging. Julia’s research focusses on machine/deep learning, complex motion modelling, as well as multi-modality and quantitative imaging for a range of medical imaging applications. She is serving on the Editorial Board of Medical Image Analysis, is Associate Editor for IEEE Transactions on Medical Imaging and  IEEE Transactions on Biomedical Engineering, and has recently founded the new free open-access Journal of Machine Learning for Biomedical Imaging (melba-journal.org). She has been Program Chair of  the MICCAI 2018 conference, is General Chair of IPMI 2021, and will be General Chair of MICCAI 2024, to be held for the first time in Africa. She is elected member of the IEEE EMBS Administrative Committee and the MICCAI Society Board of Directors, and an elected Fellow of the MICCAI Society (2018), ELLIS (2019), and IEEE (2021).

References
Oksuz I, Ruijsink JB, Puyol Anton E, Clough JR, Lima da Cruz, GJ, Bustin, A, Prieto Vasquez C, Botnar RM, Rueckert D, Schnabel JA, King AP. Automatic CNN-based detection of cardiac MR motion artefacts using k-space data augmentation and curriculum learning. Medical Image Analysis (2019). 10.1016/j.media.2019.04.009
 
Oksuz I, Clough J, Ruijsink B, Puyol-Antón E, Bustin A, Cruz G, Prieto C, King AP, Schnabel JA. Deep Learning Based Detection and Correction of Cardiac MR Motion Artefacts During Reconstruction for High-Quality Segmentation. IEEE Transactions on Medical Imaging (2020).  10.1109/TMI.2020.3008930
 
Ruijsink B, Puyol-Antón E, Oksuz I, Sinclair M, Bai W, Schnabel JA, Razavi R, King AP. Fully Automated, Quality-Controlled Cardiac Analysis From CMR: Validation and Large-Scale Application to Characterize Cardiac Function. JACC: Cardiovascular Imaging (2019). 10.1016/j.jcmg.2019.05.030
 
Martinez O, Ellis S, Baltatzis V, Devaraj A, Desai S, Le Golgoc, Nair A, Glocker B, Schnabel JA. Data Augmentation for Early Stage Lung Nodules using Deep Image Prior and CycleGan. In: MED-NEURIPS (2019).
 
Martinez O, Ellis S, Baltatzis V, Nair A, Le Folgoc L, Desai S, Glocker B, Schnabel JA. Patient-Specific 3D Cellular Automata Nodule Growth Synthesis in Lung Cancer without the Need of External Data. Accepted for IEEE Symposium on Biomedical Imaging – ISBI 2021.
 
Zimmer VA, Gómez A, Skelton E, Toussaint N, Zhang T, Khanal B, Wright R, Noh Y, Ho A, Matthew J, Hajnal JV, Schnabel JA. Towards Whole Placenta Segmentation at Late Gestation Using Multi-view Ultrasound Images. In: Medical Image Computing and Computer-Assisted Intervention – MICCAI 2019, Lecture Notes in Computer Science, vol 11768, Springer (2019) https://doi.org/10.1007/978-3-030-32254-0_70
 
Zimmer VA, Gómez A, Skelton E, Ghavami N, Wright R, Li L, Matthew J, Hajnal JV, Schnabel JA. A Multi-task Approach Using Positional Information for Ultrasound Placenta Segmentation. In: Medical Ultrasound, and Preterm, Perinatal and Paediatric Image Analysis. ASMUS 2020, PIPPI 2020. Lecture Notes in Computer Science, vol 12437, Springer (2020) https://doi.org/10.1007/978-3-030-60334-2_26
 
Wright R, Toussaint N, Gómez A, Zimmer VA, Khanal B, Matthew J, Skelton E, Kainz B, Rueckert D, Hajnal JV, Schnabel JA. Complete Fetal Head Compounding from Multi-view 3D Ultrasound. In: Medical Image Computing and Computer-Assisted Intervention – MICCAI 2019, Lecture Notes in Computer Science, vol 11766, Springer (2019) https://doi.org/10.1007/978-3-030-32248-9_43
 
Toussaint N, Khanal B, Sinclair M, Gomez A, Skelton E, Matthew J, Schnabel JA. Weakly Supervised Localisation for Fetal Ultrasound Images. In: Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. DLMIA 2018, ML-CDS 2018. Lecture Notes in Computer Science, vol 11045, Springer (2018). 10.1007/978-3-030-00889-5_22

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

 

Beyond the Patterns- Teil 21: Florian Willomitzer - Fundamental Limits in Computational 3D Imaging: From Novel 3D Cameras to Looking around Corners

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Florian Willomitzer is former PhD graduate of FAU and is now back to report on his latest exciting research that he conducted during his time as research assistant professor at Northwestern University.

Abstract: In recent years, the introduction of modern computer vision algorithms has led to new and exciting developments in imaging sciences, such as lidar 3d mapping for autonomous driving or medical imaging and displaying tools that assist doctors in diagnosis and therapy. In light of the seemingly limitless opportunities of these developments, the knowledge about fundamental limits has become even more important: By knowing that our imaging device already operates at the physical limit (e.g., of resolution), we can avoid unnecessary investments in better hardware, such as faster detectors, better optics, or cameras with higher pixel resolution. Moreover, limits often appear as uncertainty products, making it possible to optimize our measurement towards a specific quantity (e.g., speed) by trading in information less critical for the respective application. Although the imaging device is essential in this optimization, the central role is assumed by the illumination, which serves as an encoder of the desired information.

In this talk, I will discuss the virtue of limits and merit of illumination modalities in computational 3D imaging systems using examples of my research. Among other projects, I will introduce a novel method to image hidden objects through scattering media or around corners and the ‘single-shot 3D video camera’ – a highly precise 3D sensor for the dense measurement of fast macroscopic live scenes.

Short Bio: Florian Willomitzer is a Research Assistant Professor at Northwestern University, USA. He graduated from the University of Erlangen-Nuremberg, Germany, where he received his Ph.D. degree with honors (‘summa cum laude’) in 2017. During his doctoral studies Florian investigated physical and information theoretical limits of optical 3D-sensing and implemented sensors that operate close to these limits.
At Northwestern University, Florian and his students develop novel techniques to overcome traditional resolution limitations and dynamic range restrictions in 3D and 2D imaging. Moreover, Florian’s research is focused on new methods to image hidden objects through scattering media or around corners, high-resolution holographic displays, and the implementation of high precision metrology methods in low-cost mobile handheld devices.

Florian is currently Chair of the OSA COSI conference and has served as reviewer for OSA, IEEE, SPIE, and the Nature Portfolio. His Ph.D. thesis was awarded with the Springer Theses Award for Outstanding Ph.D. Research.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 22: Udaranga Wickramasinghe - Voxel2Mesh: 3D Mesh Model Generation from Volumetric Data

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

It’s a great pleasure to welcome Udaranga Wickramasinghe from EPFL, Lausanne, Switzerland at our lab for an invited talk!

Abstract: CNN-based volumetric methods that label individual voxels dominate the field of biomedical image segmentation. However, 3D surface representations of the segmented structures are often required for tasks like shape analysis. They can be obtained by post-processing the labeled volumes which typically introduces artifacts and prevents end-to-end training. In this talk, I introduce Voxel2Mesh, a novel architecture that goes from 3D image volumes to 3D surfaces directly without any post-processing and with better accuracy than current methods when using smaller training datasets. I will discuss in detail about the motivation, design choices, strengths and limitations of the architecture. I will also discuss how this can help to accelerate the adoption of deep learning techniques for shape analysis in medical imaging.

Short Bio: Udaranga Wickramasinghe is a PhD student at CVLAB – EPFL advised by Prof. Pascal Fua. His research focuses on 3D surface extraction from volumetric images and ways to introduce prior knowledge into deep neural networks. Prior to joining CVLAB, he completed his master’s degree in Computer Science at EPFL, Switzerland in 2017 and his bachelor’s degree in Electronics and Telecommunication Engineering at University of Moratuwa, Sri Lanka in 2014.

Links & References
Voxel2Mesh on arxiv: https://arxiv.org/abs/1912.03681
Voxel2Mesh code: https://github.com/cvlab-epfl/voxel2mesh
Heart segmentation: https://arxiv.org/abs/2102.07899
Deep Active Surface models: https://arxiv.org/abs/2011.08826

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 23: Anton Batliner - Moving to a World Beyond

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

We welcome Dr. Anton Batliner back to the Pattern Recognition Lab. In his presentation, he will inform us on the problems of worshipping the excessive use of significance testing.

Abstract: Null Hypothesis Testing (NHT) with p-values as decisive criteria has been criticized from its very beginning, back in the last century. The American Statistical Association published two position papers in 2016 and 2019, questioning its role in science and envisioning a “World Beyond p < 0.05“. Yet, NHT as ritual prevails until today. We will describe the shortcomings of NHT and sketch alternative ways of evaluating results such as explorative statistics, bootstrapping, and confidence intervals.

Short Bio: ANTON BATLINER received his doctor degree in Phonetics in 1978 at LMU Munich. He has been with the Institute for Nordic Languages and the Institute for German Philology, both at LMU Munich, the IMS at Stuttgart University, the Pattern Recognition Lab at FAU Erlangen, the Institute for Human-Machine Communication at TUM, the Chair of Complex and Intelligent Systems, University of Passau, and the Chair of Embedded Intelligence for Health Care and Wellbeing at University of Augsburg. He is co-editor/author of two books and author/co-author of more than 300 technical articles, with an h-index of 48 and >11000 citations. His main research interests are all (cross-linguistic) aspects of prosody and (computational) paralinguistics. He repeatedly served as co-organiser for Workshops/Sessions/Challenges on emotion and other paralinguistic events at LREC, ICPhS, Speech Prosody, and Interspeech. He was guest editor for AHCI, CSL, and Speech Communication, Associated Editor for the IEEE Transactions on Affective Computing, as well as reviewer for numerous leading journals, conferences, and workshops.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 24: Alicia Fornés - Computer Vision for Handwritten Document Analysis

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Alicia Fornes is performing ground-breaking research in hand-written document analysis at the Computer Vision Center, Barcelona, Spain. So, it’s a great pleasure to welcome her to Beyond the Patterns!

Abstract: Lately, document image analysis and recognition systems have become fundamental for recognizing, searching and extracting information from historical manuscripts, easing the access and indexing of our cultural heritage. However, and even with the recent advances in deep learning, historical handwritten documents are difficult due to the variability in the handwriting styles and the few available labelled data. For this reason, synthetic data generation, domain adaptation or few-shot learning techniques have been proposed for alleviating those problems. This talk will overview some of these techniques, showing examples of their application to textual documents, music scores or enciphered manuscripts, and discussing some of the open challenges.

Short Bio: Dr. Alicia Fornés is a Senior Research Fellow at the Computer Vision Center (CVC) and the Universitat Autònoma de Barcelona (UAB). She has more than 100 publications in international conferences and journals. She has participated in many research and technology transfer projects related to the recognition of handwritten documents. She received the IAPR/ICDAR Young Investigator Award in 2017 for outstanding contributions in the recognition of handwriting, text and graphics, with high impact to the field of Digital Humanities. Her research interests include historical document image analysis, handwriting recognition and optical music recognition.

References & Links
Pianola Roll Digitizer: 

Document Information Extraction http://dag.cvc.uab.es/infoesposalles/media-gallery/

Historical Social Network http://dag.cvc.uab.es/xarxes/

DECRYPT project: https://de-crypt.org/

J.I.Toledo, S.Dey, A.Fornés, J.Lladós. Handwriting Recognition by Attribute embedding and Recurrent Neural Networks. ICDAR, 2017

L.Kang, P.Riba, M.Villegas, A.Fornes, M.Rusiñol. Candidate fusion: Integrating language modelling into a sequence-to-sequence handwritten word recognition architecture. Pattern Recognition, 2021

Lei Kang, Pau Riba, MarçalRusiñol, Alicia Fornés, Mauricio Villegas. Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition. https://arxiv.org/abs/2005.13044, 2020.

Pau Riba, Andreas Fischer, JosepLladós, Alicia Fornés. Learning Graph Edit Distance by Graph Neural Networks. https://arxiv.org/abs/2008.07641, 2020.

J.I.Toledo, M. Carbonell, A.Fornés, J.Lladós. Information Extraction from Historical Handwritten Document Images with a Context-aware Neural Model, Pattern Recognition, 2019.

M.Carbonell, A.Fornés, M Villegas, J.Lladós. A Neural Model for Text Localization, Transcription and Named Entity Recognition in Full Pages. Pattern Recognition Letters, 2020.

J.Chen, P.Riba, A.Fornés, J.Mas, J.Lladós, J..M.Pujadas-Mora, Word-Hunter: A GamesourcingExperience to Validate the Transcription of Historical Manuscripts. ICFHR (2018)

A.Baró, P.Riba, J.Calvo-Zaragoza, A.Fornés. From Optical Music Recognition to Handwritten Music Recognition: a Baseline. Pattern Recognition Letters, 2019.

M.Visani, V.C.Kieu, A.Fornés, N.Journet. The ICDAR 2013 Music Scores Competition: Staff Removal. ICDAR, 2013.

L.Kang, M.Rusiñol, A.Fornés, P.Riba, M.Villegas. Unsupervised Adaptation for Synthetic-to-Real Handwritten Word Recognition. Winter Conf. on Applications of Computer Vision (WACV), 2020.

L.Kang, P.Riba, Y.Wang, M.Rusiñol, A.Fornés, M.Villegas. GANwriting: Content-Conditioned Generation of Styled Handwritten Word Images. ECCV, 2020.

L.Kang, P.Riba, M.Rusiñol, A.Fornés, M.Villegas. Distilling Content from Style for Handwritten Word Recognition. International Conference on Frontiers in Handwriting Recognition (ICFHR), 2020.

J.Chen, M.A.Souibgui, A. Fornes, B.Megyesi. A Web-BasedInteractiveTranscriptionTool forEncryptedManuscripts. HistoCrypt, 2020

M.A.Souibgui, A.Fornés, Y.Kessentini, C.Tudor. A Few-shot Learning Approach for Historical Encoded Manuscript Recognition, International Conference on Pattern Recognition (ICPR), 2020.

M.A.Souibgui, A.Bensalah, J.Chen, A.Fornes, M.Waldispühl. Abstract submitted to the International Symposium on Runes and Runic Inscriptions (ISRRI), 20|21

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 25: Jochen Hoffmann - Information exchange as an infringement of competition law

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

It’s a great pleasure to announce our next guest in our invited lecture series “Beyond the Patterns”.

Abstract: European competition law prohibits the coordination of economic decisions among market participants, esp. price fixing. Information exchange or even unilateral advance communication of prices may amount to such an illegal collusion. The lecture gives a brief introduction to the rules against coordinated behavior and explains the difference between a coordinated behavior based on information exchange and a legal parallel behavior based on available information.

Short Bio: Jochen Hoffmann (born in 1971) studied law and business at the University of Bayreuth where he passed the first state examination in law in 1995 and received his doctorate in 1998. After passing the second state examination in law in 1998 he worked as a research assistant at the University of Bayreuth from 1999 to 2006 and received his habilitation there in 2005. From 2006 to 2009 he was Professor for civil law, business law and international economic law at the University of Hamburg. Since October, 1st 2009 Professor Hoffmann holds the chair for Business Law at the Friedrich-Alexander-Universität Erlangen-Nürnberg and has been Dean of the Faculty of Business, Economics and Law and Speaker of the School of Law since 2020.

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 26: Daniel Cremers - Deep Direct Visual SLAM

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

We welcome Prof. Dr. Daniel Cremers as an invited speaker to our lab next week and are looking forward to his presentation.

Abstract: While neural networks have swept the field of computer vision and replaced classical methods in most areas of image analysis and beyond, extending their power to the domain of camera-based 3D reconstruction and visual SLAM remains an important open challenge. In my talk, I will discuss the problem of image-based reconstruction and visual Simultaneous Localization and Mapping (SLAM). In particular, I will advocate direct methods that recover 3D structure and camera motion directly from the intensity images. Moreover, I will discuss how the performance of visual SLAM methods can be drastically enhanced using the predictive power of deep networks.

Short Bio: Daniel Cremers received a PhD in Computer Science (2002) from the University of Mannheim, Germany. Subsequently he spent two years as a postdoctoral researcher at the University of California at Los Angeles (UCLA) and one year as a permanent researcher at Siemens Corporate Research in Princeton, NJ. From 2005 until 2009 he was associate professor at the University of Bonn. Since 2009 he holds the Chair of Computer Vision and Artificial Intelligence at the Technical University of Munich. His publications received numerous awards, including the ‘Best Paper of the Year 2003’ (Int. Pattern Recognition Society), the ‘Olympus Award 2004’ (German Soc. for Pattern Recognition) and the ‘2005 UCLA Chancellor’s Award for Postdoctoral Research’. For pioneering research he received five grants from the European Research Council, including a Starting Grant, a Consolidator Grant and an Advanced Grant. In 2018 he organized the largest ever European Conference on Computer Vision in Munich. He is member of the Bavarian Academy of Sciences and Humanities. In December 2010 he was listed among “Germany’s top 40 researchers below 40” (Capital). On March 1st 2016, Prof. Cremers received the Gottfried Wilhelm Leibniz Award, the biggest award in German academia. He is co-founder of several companies, most recently the high-tech startup Artisense.

Additional References

Weiss, Sebastian, Robert Maier, Rüdiger Westermann, Daniel Cremers, and Nils Thuerey. „Sparse Surface Constraints for Combining Physics-based Elasticity Simulation and Correspondence-Free Object Reconstruction.“ arXiv preprint arXiv:1910.01812 (2019).

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 27: Luis Pineda - Active MR k-space Sampling with Reinforcement Learning

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

Our invited speaker in this video is Luis Pineda from Facebook AI Research!

Abstract: Deep learning approaches have recently shown great promise in accelerating magnetic resonance image (MRI) acquisition. The majority of existing work have focused on designing better reconstruction models given a pre-determined acquisition trajectory, ignoring the question of trajectory optimization. In this paper, we focus on learning acquisition trajectories given a fixed image reconstruction model. We formulate the problem as a sequential decision process and propose the use of reinforcement learning to solve it. Experiments on a large scale public MRI dataset of knees show that our proposed models significantly outperform the state-of-the-art in active MRI acquisition, over a large range of acceleration factors.

Bio: Luis Pineda is a researcher at Facebook AI Research in Montreal. He obtained his PhD from University of Massachusetts Amherst in 2018, advised by Prof. Shlomo Zilberstein; during his PhD, he focused on developing heuristic search algorithms for probabilistic planning and their applications to robotics problems. At FAIR, his focus has been on studying deep reinforcement learning and its applications. His recent work includes exploring the use of deep RL for active MRI acquisition, and developing novel RL-based methods for multi-agent collaboration in Hanabi. 

Reference:
Pineda, Luis, Sumana Basu, Adriana Romero, Roberto Calandra, and Michal Drozdzal. „Active MR k-space sampling with reinforcement learning.“ In International Conference on Medical Image Computing and Computer-Assisted Intervention, pp. 23-33. Springer, Cham, 2020.
https://link.springer.com/chapter/10.1007/978-3-030-59713-9_3

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)

Beyond the Patterns- Teil 28: Petar Veličković - Geometric Deep Learning

Kurs-Verknüpfung     

Beyond the Patterns

Lehrende(r)

Prof. Dr. Andreas Maier

Sprache

Englisch

Einrichtung

Friedrich-Alexander-Universität Erlangen-Nürnberg

It’s a great pleasure to welcome Petar Velikovi from Deep Mind to our Lab!

Abstract: The last decade has witnessed an experimental revolution in data science and machine learning, epitomised by deep learning methods. Indeed, many high-dimensional learning tasks previously thought to be beyond reach –such as computer vision, playing Go, or protein folding – are in fact feasible with appropriate computational scale. Remarkably, the essence of deep learning is built from two simple algorithmic principles: first, the notion of representation or feature learning, whereby adapted, often hierarchical, features capture the appropriate notion of regularity for each task, and second, learning by local gradient-descent type methods, typically implemented as backpropagation.
While learning generic functions in high dimensions is a cursed estimation problem, most tasks of interest are not generic, and come with essential pre-defined regularities arising from the underlying low-dimensionality and structure of the physical world. This talk is concerned with exposing these regularities through unified geometric principles that can be applied throughout a wide spectrum of applications.

Such a ‘geometric unification’ endeavour in the spirit of Felix Klein’s Erlangen Program serves a dual purpose: on one hand, it provides a common mathematical framework to study the most successful neural network architectures, such as CNNs, RNNs, GNNs, and Transformers. On the other hand, it gives a constructive procedure to incorporate prior physical knowledge into neural architectures and provide principled way to build future architectures yet to be invented.

Bio:Petar Velikovi is a Senior Research Scientist at DeepMind. He holds a PhD in Computer Science from the University of Cambridge (Trinity College), obtained under the supervision of Pietro Liò. His research interests involve devising neural network architectures that operate on nontrivially structured data (such as graphs), and their applications in algorithmic reasoning and computational biology. He has published relevant research in these areas at both machine learning venues (NeurIPS, ICLR, ICML-W) and biomedical venues and journals (Bioinformatics, PLOS One, JCB, PervasiveHealth). In particular, he is the first author of Graph Attention Networks—a popular convolutional layer for graphs—and Deep Graph Infomax—a scalable local/global unsupervised learning pipeline for graphs (featured in ZDNet). Further, his research has been used in substantially improving the travel-time predictions in Google Maps (covered by outlets including the CNBC, Endgadget, VentureBeat, CNET, the Verge and ZDNet).

Geometric Deep Learning Website:
https://geometricdeeplearning.com

Michael Bronstein’s Blog Post on Geometric Deep Learning:
https://towardsdatascience.com/geometric-foundations-of-deep-learning-94cdd45b451d

Petar’s Talk at Cambridge:

This video is released under CC BY 4.0. Please feel free to share and reuse.

For reminders to watch the new video follow on Twitter or LinkedIn. Also, join our network for information about talks, videos, and job offers in our Facebook and LinkedIn Groups.

Music Reference: 
Damiano Baldoni – Thinking of You (Intro)
Damiano Baldoni – Poenia (Outro)