- Evaluating acceleration models and respective stochasticity towards a real-time forecasting model for landslides. Masterarbeit, 2023 mehr…
- Physics-Informed Geometric Deep Learning for Molecular Property Prediction. Masterarbeit, 2023 mehr…
- Machine Learning in the Context of a Real-Time Education in Gaming. Masterarbeit, 2023 mehr…
- Decentralized Mutual Exclusion for Mobile Robots at Shared Resources in Complex. Masterarbeit, 2023 mehr…
- Implementing a learning-rate scheduler in a Newton-CG Optimizer for Deep Learning. Bachelorarbeit, 2022 mehr…
- Transfer Learning and Dynamic Loading in TUM-Lens. Masterarbeit, 2022 mehr…
- Efficient and Scalable Kernel Matrix Approximation using Hierarchical Decomposition. Masterarbeit, 2022 mehr…
- Second-Order Optimization Methods for Bayesian Neural Networks. Masterarbeit, 2021 mehr…
- Extending a Newton-CG Second-order Optimizer to Natural Language Processing. Bachelorarbeit, 2021 mehr…
- Training Deep Convolutional Neural Networks on the GPU Using a Second-Order Optimizer. Bachelorarbeit, 2020 mehr…
- Implementing a TensorFlow-Slim based Android app for image classification. Bachelorarbeit, 2020 mehr…
- Application of second-order optimisation for large-scale deep learning. Masterarbeit, 2020 mehr…
- Integrated approach of Random Projections and Sparse Grids for Density estimation. Masterarbeit, 2020 mehr…
- Python Software Suite of Geometric-Oblivious FMM. IDP-Arbeit, 2020 mehr…
- Comparison of distance metrics for MDS based NLDR using CNNs. Bachelorarbeit, 2020 mehr…
Severin Reiz
Technical University of Munich
TUM School of CIT
Department of Computer Science
Boltzmannstrasse 3
85748 Garching
Germany
Office: MI 02.05.058
Mail: reiz (at) in.tum.de
Tel: +49-89-289-18603
Office Hours: by arrangement
Background
- Doctoral candidate, TUM Graduate School - IGSSE
- Research Associate (Wissenschaftlicher Mitarbeiter) at TUM SCCS since July 2017
- M.Sc. with Honours in Computational Science and Engineering, Technical University of Munich, 2017
- M. Sc. Thesis at University of Texas at Austin, USA (George Biros)
- Seminar Thesis at Norwegian University of Science and Technology, Trondheim and SINTEF
- B.Sc. Engineering Science, Technical University of Munich, 2014
Open and running student projects
Topics change (frequently) with my ongoing work and depend mostly on the interests and experience of the student. It is best to contact me directly if you are interested in a thesis or student project.
See also the list of Student Projects at our chair.
Do you want to know what other students are working on in our chair? You are warmly encouraged to attend their presentations at the SCCS Colloquium! Come to get ideas, meet your potential supervisor, or to learn from the style of others for your own presentation.
Open student projects
Running student projects
We are working on an android app for live image classification (locally on your device). Feel free to download the Android package (apk)
Past student projects and Theses advised
Teaching
Winter semester 2020/21
- Seminar Computational Aspects of Machine Learning: Link to application page Mooodle
Summer semester 2020
Publications
- Efficient and Scalable Kernel Matrix Approximations Using Hierarchical Decomposition. In: Intelligent Computers, Algorithms, and Applications. Springer Nature Singapore, 2024 mehr…
- Efficient and Scalable Kernel Matrix Approximations Using Hierarchical Decomposition. In: Intelligent Computers, Algorithms, and Applications. Springer Nature Singapore, 2024 mehr…
- Neural Nets with a Newton Conjugate Gradient Method on Multiple GPUs. In Proceedings of the 14th International Conference on Parallel Processing and Applied Mathematics, 2022, 13 mehr…
- Fast Evaluation and Approximation of the Gauss-Newton Hessian Matrix for the Multilayer Perceptron. SIAM Journal on Matrix Analysis and Applications, 2021 mehr…
- Software for Exascale Computing: Some Remarks on the Priority Program SPPEXA. In: Springer International Publishing, 2020 mehr…
- Software for Exascale Computing - SPPEXA 2016-2019. Band LNCSE 136. Lecture Notes in Computational Science and Engineering (LNCSE) 136. Springer, 2020 mehr…
- H-matrix approximation of the Gauss-Newton Hessian matrix for the multilayer perceptron. 33rd Conference on Neural Information Processing Systems (NeurIPS 2019), 2019 mehr…
- Distributed O(N) Linear Solver for Dense Symmetric Hierarchical Semi-Separable Matrices. Embedded Multicore/Many-core Systems on a chip, IEEE, 2019 mehr…
- Distributed-Memory Hierarchical Compression of Dense SPD Matrices. SC '18: Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, 2018 mehr…
- Black Box Hierarchical Approximations for SPD Matrices. Master's thesis, 2017 mehr…
- CAD-integrated Topology Optimization (BGCE Honours Project). , 2016 mehr…
- CFD Study of Fuel-air Mixing in a Novel Low-NOx Burner. Projektarbeit, 2015 mehr…
- Water Injections in Gas Turbines - Kinetic Modeling with Cantera. Bachelor's thesis, 2014 mehr…
Talks
- SPPEXA and beyond: Perspectives of Exascale Computing in the Context of AI. Human-centric Artificial Intelligence, 20202nd French-German-Japanese Symposium mehr…
- Some words from SPPEXA. (Vortrag / Parallel Programming Models - Productivity and Applications for Exascale and Beyond - 4th edition) 2019 mehr…
- A Fast Multipole Method for Training Neural Networks. PhD Forum: IEEE International Supercomputing Conference, 2019 mehr…
- Numerical approaches to large-scale machine learning. (Vortrag / Leogang High Performance Computing Workshop) 2019 mehr…
- Second-order optimization for AI using HSS Matrices on distributed-memory setup. Convergence of HPC and Data Science for Future Extreme Scale Intelligent Applications, 2019 mehr…
Posters
- Scalable Hierarchical Approximation of Dense Kernel Matrices. SIAM-PP 2024Baltimore mehr…
- Training Large-Scale Neural Networks with a Newton Conjugate Gradient Method (Newton-CG). SIAM Conference on Computational Science and Engineering (CSE23), 2023Amsterdam mehr…
- An Exascale Library for Numerically Inspired Machine Learning (ExaNIML). ISC High Performance, 2020Frankfurt (DIGITAL due corona) mehr…
- ExaNIML: An Exascale Library for Numerically Inspired Machine Learning. 13th IGSSE Forum, Raitenhaslach, 201913th IGSSE Forum, Raitenhaslach mehr…
- Nonsymmetric Algebraic Fast Multipole Method. Computational Science at Scale (CoSaS) 2018, 2018 mehr…