Machine-Learning in Engineering (MLE@TUHH) ist eine Initiative zur Bündelung der Kompetenzen im Bereich Machine Learning an der TUHH mit dem Ziel des Wissenstransfers in Richtung Wirtschaft und Industrie. Am 4. November 2020 um 10:00 Uhr möchte sich MLE allen Mitglieder der Technischen Universität Hamburg vorstellen. Die Veranstaltung wendet sich an alle Mitglieder der TUHH, vom Institutsleiter bis zum Studierenden. Wir wollen so auf die verschiedenen Angebote der MLE Initiative aufmerksam machen und den Informationsaustausch innerhalb der Universität fördern. Insgesamt soll die Sichtbarkeit der Machine-Learning-Aktivitäten innerhalb der TUHH gesteigert, Kooperationen gefördert und interessierten Studierenden ein Einblick ermöglicht werden. Auch die Verankerung von MLE in der der Lehre soll aufgezeigt werden.
Grußwort durch den geschäftsführenden Präsidenten der TUHH Prof. Dr. Andreas Timm-Giel
Vorstellung des Studiengangs „Data Science“ durch Prof. Dr. Tobias Knopp
3 Pitches von Doktoranden (aus diversen Bereichen), wie sie ML in ihrer Forschung anwenden:
Vorstellung Kooperationspartner Artificial Intelligence Center Hamburg e.V. (ARIC) durch CEO Alois Krtil
The events are free for members of TUHH and HZG.
In two interactive sessions, we show examples of engineers and scientists using MATLAB for building AI-driven systems. We explore the complete workflow of developing machine learning and deep learning applications with MATLAB using a real-world ECG data set. We also show how to easily develop and apply data analytics solutions that take advantage of enhanced signal processing and AI techniques, including automated feature extraction, model selection and tuning.
We recommend taking the free, two-hour interactive introductions to practical machine learning methods and deep learning methods, respectively, before each session. In between the two sessions, participants will have access to the used data set, to apply your acquired knowledge and share their achieved results. The second session does not have the first session as a prerequisite and can be attended independently.
|2||19.04.21||Hidde Lekanne Deprez||Enhancing simulation images with GANs|
|4||03.05.21||Merten Stender||Physics-Informed Learning (Zoom)|
|5||17.05.21||Daniel Höche||Towards predictive maintenance (Zoom)|
|6||31.05.21||Amir Kotobi||Dynamic structure investigation of biomolecules with pattern recognition algorithms and X-ray experiments (Zoom)|
|7||07.06.21||Mijail Guillemard||Basics of Persistent Homology in Machine Learning (Zoom)|
|8||14.06.21||Nihat Ay||Information Geometry for Deep Learning|
|(Prof. Date)||Mirko Skiborowski||tba|
|9||21.06.21||Benedikt Kriegesmann||Efficient uncertainty quantification using surrogate models: application to fiber composite structures|
|(Prof. Date)||Matthias Mnich||Reinforcement Learning for Integer Programming|
|10||28.06.21||Frederic E. Bock||Data-driven Machine Learning Corrections of Physics-Based Analytical Model Predictions towards High-Fidelity Simulation Solutions – a Hybrid Modelling Approach|
|11||05.07.21||Patrick Göttsch||MARL Simulation – Multi-Agent Reinforcement Learning Simulation Software|
Hidde Lekanne Deprez: Enhancing simulation images with GANs
In the Standard Platform League, certain types of annotations such as semantic segmentations, depth maps and object localization are difficult to obtain from real world recordings. The use of synthetic data could circumvent this problem as obtaining these annotations within a simulation is trivial. However, there is a catch, the reality gap makes algorithms trained on the synthetic images perform much worse in actual applications. Researchers can painstakingly implement more features to the simulation to close this gap. However, there are alternatives such as the neural networks presented here. The CycleGan and MUNIT architectures are able to make a domain translation, maintaining semantic information but changing the style, without any labels or matchings. This could mean that a translation between simulation and real images is possible as long as we have images of both domains. For my bachelor thesis I experimented with using these two neural networks to make this translation and my insights are presented in this talk.
Merten Stender: Physics-Informed Learning
Machine Learning and Deep Learning have brought disruptive innovations to many fields since 2012. Today the application of those data-driven, and mostly black-box type models, can be regarded state-of-the-art in many scientific disciplines. However, the question of knowledge conservation arises: how to bring prior knowledge from generations of research and experience into the modeling process? This talk summarizes recent advances, lines of research and perspectives on “Physics-Informed Learning”, which is an umbrella term for blending first principles into evidence-based and data-driven models. Particular focus is put on engineering vibrations and spatio-temporal dynamics, e.g. water waves.
Daniel Höche: Towards predictive maintenance
Sustainable engineering requires reliable and plannable material behaviour in critical working environments like offshore. The extension of digital-twins towards virtual engineering assisted circular economy therefore needs computational models that enable the calculation of maintenance intervals or even the material condition at the end of its service life. The talk outlines how the combination of AI tools, data based models and physics based models facilitate predictive maintenance for metallic engineering materials exposed to severe conditions in-service. Aspects related to uncertainty, data availability or validation will be discussed.
Amir Kotobi: Dynamic structure investigation of biomolecules with pattern recognition algorithms and X-ray experiments
The biological functions of macromolecular systems, such as peptides and proteins, are largely defined by their spatial and electronic structures and thus it is of great importance to have high resolution view over these structures. Dynamic structure investigation of biomolecules with advanced molecular dynamic simulations and machine learning approaches on the basis of free energy calculations can give valuable opportunity in analysing the trajectories.
Mijail Guillemard: Basics of Persistent Homology in Machine Learning
Persistent Homology is a recent development in applied algebraic topology that has been used in several machine learning strategies. In this talk, we present a short introduction to this topic with several applications in signal processing and data analysis.
Prof. Date: Nihat Ay: Information Geometry for Deep Learning
In the first part of my presentation I will highlight the importance of the geometric perspective when dealing with learning systems. Information geometry offers a general framework for the identification of natural geometric structures for learning. The impact of this approach has been demonstrated in terms of the natural gradient method, one of the most prominent information-geometric methods within the field of machine learning. It was proposed by Amari in 1998 and uses the Fisher-Rao metric as a Riemannian metric for the definition of a gradient within optimisation tasks. Since then it proved to be extremely efficient in the context of neural networks, reinforcement learning, and robotics. However, training deep neural networks with this method remains a difficult task. I will present recent results that allow us to greatly simplify the natural gradient for deep learning. I will conclude my talk with an outline of further applications and extensions of information geometry that are particularly important for mathematical data science.
Prof. Date: Mirko Skiborowski
Prof. Date: Benedikt Kriegesmann: Efficient uncertainty quantification using surrogate models: application to fiber composite structures
Uncertainty quantification in engineering sciences takes into account the uncertainties that may exist and affect a certain physical system in an a priori unknown manner. If the input parameters of a system or model are subject to stochastic scatter, then also the output parameters (i.e. objective values) scatter randomly. The stochastic distribution of an objective value can be determined with uncertainty quantification methods such as Monte Carlo simulations. This requires a multitude of evaluations of the underlying model (up to 103, 106). Hence, this approach is infeasible for computationally demanding models. For such applications, surrogate models, like artificial neural networks, Kriging and polynomial chaos expansions, can be first trained with a small amount of model evaluations and then used as a proxy instead of the expensive model.
In the current talk, this procedure is demonstrated by the example of fiber composite structures. Here, the random objective functions are the strength and the stiffness of a component. Random input parameters are (amongst others) material properties, geometric deviations and manufacturing defects. Some random input parameters can only be modelled on a smaller scale than the whole component. Therefore, surrogate-boosted Monte Carlo simulations are performed on different scales and the results in each case are propagated to the higher scale. Strength however can hardly be approximated with standard surrogate models. Here, hierarchical surrogate models are used, which link models of different fidelity, to still allow for an efficient and accurate prediction of the stochastic distribution of strength properties.
Prof. Date: Matthias Mnich: Reinforcement Learning for Integer Programming
The integer programming problem is one of the most fundamental problems in combinatorial optimization, where one seeks an optimal solution among a finite, but usually extremely large set of discrete alternatives. Integer programs are ubiquitous in engineering and industrial applications, as they can model a large variety of highly complex tasks by means of discrete variables which are tight together through constraints. Powerful commercial solvers exists, which can solve large-scale instances with thousands of variables generally quite fast, but there are still several important integer programming models which cannot be solved at all. Theoretical and design and analysis of integer programming algorithms usually fails to explain both the successes, and the failures, of industrial solvers, as worst-case run times of those algorithms are often super-exponential in the number of variables. We discuss novel approaches based on methods of reinforcement learning to attack some of the most prominent integer programming models for which classical methods suffer from expensive time and memory usage, talking about their advantages and limitations.
Frederic E. Bock: Data-driven Machine Learning Corrections of Physics-Based Analytical Model Predictions towards High-Fidelity Simulation Solutions – a Hybrid Modelling Approach
Patrick Göttsch: MARL Simulation – Multi-Agent Reinforcement Learning Simulation Software
In this talk a MARL Simulation environment will be introduced to let heterogeneous groups of agents controlled by hand crafted control algorithms or by learned control policies flock for search and rescue missions or to move in a formation by avoiding obstacles. This environment allows simulations of agents constrained by non-ideal communication.
We are a team of students from the Hamburg University of Technology. Since the foundation of our team in the year 2013, we are participating anually in the Standard Platform League of the RoboCup competition. The RoboCup is an international initiative with the goal of advancing research in humanoid robotics and artificial intelligence. Our team consists mostly of students who work on this project in their free time. Furthermore, we are one of the few teams that developed their entire framework from scratch.
Since 2014, we are anually hosting the Robotic Hamburg Open Workshop (RoHOW).
The RoHOW is an event hosted by the Hamburg University of Technology that serves the purpose of education and networking. Over 100 scientists and students from all over the world who are researching in the field of humanoid robotics participate in this event.
The goal of this event is the exchange of information between RoboCup teams from different universities and research institutes. Participants have the opportunity to visit lectures and workshops, challenge other teams in competitions, and get in direct contact with each other.
Here you find more information on RoHOW.
MLE ist eine Initiative zur Bündelung der Kompetenzen im Bereich Machine Learning an der TUHH, die kontinuierlich interne wie externe Veranstaltungen zum Thema durchführt:
Öffentliche Ringvorlesung für die interessierte Öffentlichkeit im Laufe des Wintersemesters 2020/21 mit Persönlichkeiten aus Wissenschaft und Industrie. Organisator: Marc-André Pick
MLE-Days @ TUHH: Zweitägiges Event zur Darstellung der Kompetenzen, Forschungsfelder und Zukunftskonzepte an der TUHH zu Anfang nächsten Jahres. Angebot insbesondere für Interessierte aus Wirtschaft, Industrie und öffentlichem Dienst. Ansprechpartner: Volker Turau