Search

Your search keyword '"Jobst, Matthias"' showing total 17 results
Start Over Author "Jobst, Matthias"
17 results on '"Jobst, Matthias"'

2. Deploying Machine Learning Models to Ahead-of-Time Runtime on Edge Using MicroTVM

Author

Liu, Chen, Jobst, Matthias, Guo, Liyuan, Shi, Xinyue, Partzsch, Johannes, and Mayr, Christian

Subjects
Computer Science - Machine Learning
Abstract

In the past few years, more and more AI applications have been applied to edge devices. However, models trained by data scientists with machine learning frameworks, such as PyTorch or TensorFlow, can not be seamlessly executed on edge. In this paper, we develop an end-to-end code generator parsing a pre-trained model to C source libraries for the backend using MicroTVM, a machine learning compiler framework extension addressing inference on bare metal devices. An analysis shows that specific compute-intensive operators can be easily offloaded to the dedicated accelerator with a Universal Modular Accelerator (UMA) interface, while others are processed in the CPU cores. By using the automatically generated ahead-of-time C runtime, we conduct a hand gesture recognition experiment on an ARM Cortex M4F core., Comment: CODAI 2022 Workshop - Embedded System Week (ESWeek)

Published
2023

5. Neuromorphic intermediate representation: A unified instruction set for interoperable brain-inspired computing

Author

Pedersen, Jens, Abreu, Steven, Jobst, Matthias, Lenz, Gregor, Fra, Vittorio, Bauer, Felix Christian, Muir, Dylan Richard, Zhou, Peng, Vogginger, Bernhard, Heckel, Kade, Urgese, Gianvito, Shankar, Sadasivan, Stewart, Terrence C., Sheik, Sadique, Eshraghian, Jason K., Pedersen, Jens, Abreu, Steven, Jobst, Matthias, Lenz, Gregor, Fra, Vittorio, Bauer, Felix Christian, Muir, Dylan Richard, Zhou, Peng, Vogginger, Bernhard, Heckel, Kade, Urgese, Gianvito, Shankar, Sadasivan, Stewart, Terrence C., Sheik, Sadique, and Eshraghian, Jason K.

Abstract

Spiking neural networks and neuromorphic hardware platforms that simulate neuronal dynamics are getting wide attention and are being applied to many relevant problems using Machine Learning. Despite a well-established mathematical foundation for neural dynamics, there exists numerous software and hardware solutions and stacks whose variability makes it difficult to reproduce findings. Here, we establish a common reference frame for computations in digital neuromorphic systems, titled Neuromorphic Intermediate Representation (NIR). NIR defines a set of computational and composable model primitives as hybrid systems combining continuous-time dynamics and discrete events. By abstracting away assumptions around discretization and hardware constraints, NIR faithfully captures the computational model, while bridging differences between the evaluated implementation and the underlying mathematical formalism. NIR supports an unprecedented number of neuromorphic systems, which we demonstrate by reproducing three spiking neural network models of different complexity across 7 neuromorphic simulators and 4 digital hardware platforms. NIR decouples the development of neuromorphic hardware and software, enabling interoperability between platforms and improving accessibility to multiple neuromorphic technologies. We believe that NIR is a key next step in brain-inspired hardware-software co-evolution, enabling research towards the implementation of energy efficient computational principles of nervous systems. NIR is available at neuroir.org, QC 20240926

Published
2024
Full Text
View/download PDF

8. Neuromorphic Intermediate Representation: A Unified Instruction Set for Interoperable Brain-Inspired Computing

Author

Pedersen, Jens E., Abreu, Steven, Jobst, Matthias, Lenz, Gregor, Fra, Vittorio, Bauer, Felix C., Muir, Dylan R., Zhou, Peng, Vogginger, Bernhard, Heckel, Kade, Urgese, Gianvito, Shankar, Sadasivan, Stewart, Terrence C., Eshraghian, Jason K., Sheik, Sadique, Pedersen, Jens E., Abreu, Steven, Jobst, Matthias, Lenz, Gregor, Fra, Vittorio, Bauer, Felix C., Muir, Dylan R., Zhou, Peng, Vogginger, Bernhard, Heckel, Kade, Urgese, Gianvito, Shankar, Sadasivan, Stewart, Terrence C., Eshraghian, Jason K., and Sheik, Sadique

Abstract

Spiking neural networks and neuromorphic hardware platforms that emulate neural dynamics are slowly gaining momentum and entering main-stream usage. Despite a well-established mathematical foundation for neural dynamics, the implementation details vary greatly across different platforms. Correspondingly, there are a plethora of software and hardware implementations with their own unique technology stacks. Consequently, neuromorphic systems typically diverge from the expected computational model, which challenges the reproducibility and reliability across platforms. Additionally, most neuromorphic hardware is limited by its access via a single software frameworks with a limited set of training procedures. Here, we establish a common reference-frame for computations in neuromorphic systems, dubbed the Neuromorphic Intermediate Representation (NIR). NIR defines a set of computational primitives as idealized continuous-time hybrid systems that can be composed into graphs and mapped to and from various neuromorphic technology stacks. By abstracting away assumptions around discretization and hardware constraints, NIR faithfully captures the fundamental computation, while simultaneously exposing the exact differences between the evaluated implementation and the idealized mathematical formalism. We reproduce three NIR graphs across 7 neuromorphic simulators and 4 hardware platforms, demonstrating support for an unprecedented number of neuromorphic systems. With NIR, we decouple the evolution of neuromorphic hardware and software, ultimately increasing the interoperability between platforms and improving accessibility to neuromorphic technologies. We believe that NIR is an important step towards the continued study of brain-inspired hardware and bottom-up approaches aimed at an improved understanding of the computational underpinnings of nervous systems., Comment: NIR is available at https://github.com/neuromorphs/NIR

Published
2023

12. Review of SGTR and LOCA available experimental data

Author

HOZER Zoltan, KECEK Adam, DIESCHBOURG Katia, TAURINES Tatiana, ADORNI Martina, Müllner Nikolaus, MASSONE Mattia, JOBST Matthias, ARKOMA Asko, Szabo Peter, LISHCHUK Ruslan, SHOLOMITSKY Stanislav, SCHÖPPNER Michael, LECLERE Cédric, HERRANZ Luis, IGLESIAS Rafael, and BUSSER Vincent

Subjects
Loss Of Coolant Accidents, Steam Generator Tube Ruptures, Experimental database
Abstract

This report is a result of a detailed and comprehensive review of available experimental database focusing on fuel rod failures, on fission product releases from fuel rods and on activity transport up to the environment during LOCA and SGTR events. It also includes available information issued from the monitoring of normal operation and transients in NPPs and from some accidental scenarios. The review covers all important LOCA and SGTR phenomena in PWRs and VVERs by experimental data and is expected to support the code and model validation activities within the R2CA H2020 project. More than 40 test series is included in the report. For each test series a description is provided where the test objectives, the tested materials, the experimental facility, the measured parameters and Post Irradiation Examination data are provided together with the general conclusions drawn after the test completion. In addition, three large matrices have been compiled to support detailed discussions on 1/ Phenomena 2/ Test characterization 3/ data availability. These matrices are displayed in the attachment of this report. The common use of data from small scale separate effect test and integral tests provides possibilities for the improved simulation of several phenomena (e.g. cladding burst, oxidation, hydrogen uptake, activity release and transport) under very different conditions. Most of these conditions can be considered typical for LOCA and SGTR events,while some of them cover wider parameter ranges than those which currently take place in these accidents.

Published
2020
Full Text
View/download PDF

13. Review of simulation codes and methodologies for LOCA and SGTR

Author

ARKOMA Asko, SZOGRADI Marton, LESKINEN Joona, BUSSER Vincent, JACQ François, BELON Sébastien, ADORNI Martina, MASCARI Fulvio, EDERLI Stephano, GUMENYUK Dmytro, ZHABIN Oleg, OVDIIENKO Iurii, LISHCHUK Ruslan, SHOLOMITSKY Stanislav, Müllner Nikolaus, KULACSY Katalin, Hozer Zoltan, Kecek Adam, JOBST Matthias, HERRANZ Luis, FERIA Francisco, IGLESIAS Rafael, KALIATKA Tadas, VAN-UFFELEN Paul, and BRADT Paul

Subjects
Loss Of Coolant Accidents, Steam Generatour Tube Ruptures, Design Basis Accidents, Fission product behaviour codes, Release evaluation methodologies, System codes, Code coupling, Modelling computer codes, Fuel performance codes
Abstract

This report covers the current capabilities of the available codes that will be used within the R2CA H2020 project in the release evaluation methodologiesof LOCA and SGTR transients in different kinds of reactors (PWRs, VVERs, EPR, BWR) displayingboth code descriptions and model comparison tables. It includes both system codes and codes focusing on a part of the phenomenology only, in particular focusing on the determination of the fuel rod failures as well as on the consequences of these failures. A state of the art of the combined use of these codes in implementation of the release evaluation methodologies is also described. The work that has been done covers fuel performance, sub-channel and system thermal-hydraulics, radionuclide transport and integral system behaviour. Codes modelling only the steady-state operation (especially fuel codes) have also been included as they produce initial conditions for the transient codes and are integral part of the calculation chains. Finally, the developments items for the codes, that will be addressed within the R2CA project, are also summarized dealing with cladding modelling, fuel and fission product modelling, code coupling and whole core modelling and nodalization.&nbsp

Published
2020
Full Text
View/download PDF

15. Justification of best-estimate transient calculations in comparison to the steady-state bounding-case approach

Author

Sangiorgi, M., Carenini, L., Brumm, S., Le Tellier, R., Viot, L., Wu, Z., Xia, S., Bakouta, N., Ederli, S., Mascari, F., Harti, M., Lecomte, M., Sagan, M., Pandazis, P., Jobst, Matthias, Gencheva, R., Groudev, P., Barnak, M., Matejovic, P., Villanueva, W., Chen, Y., Ma, W., Bechta, S., Kaliatka, A., Valinius, M., Kostka, P., Techy, Z., Vorobyov, Y., Thomas, R., Vokac, P., Kotouc, M., Korpinen, A., and Fichot, F.

Subjects
BWR, SBO, VVER-1000, Accident Management Measures, PWR, In-vessel melt retention, Severe Accidents, VVER-440, LOCA
Abstract

In the scope of the European IVMR (In-Vessel Melt Retention) project, calculations of In-Vessel retention (IVR) strategy with state of the art Severe Accident (SA) computer codes were performed, including the integral codes ASTEC, ATHLET-CD, MAAP, MELCOR and RELAP5/SCDAPSIM. Further codes dedicated to the study of lower plenum behaviour were also included. Simulations were performed for several types of reactors (PWR, VVER-440, VVER-1000, BWR) and several severe accident scenarios (Station Blackout (SBO) accidents and Loss-Of-Coolant accidents of several leak sizes combined with SBO). The code improvements for IVR simulation, implemented during the project, are summarized and the results obtained with the improved codes are presented in the paper.

Published
2020

17. DIGITAL data.

Author

Howard, Patrick, Arrowood, Lloyd, Jobst, Matthias, and Hansen, John

Subjects
NONDESTRUCTIVE testing, DIGITAL communications standards, DATA transmission systems standards, DIGITAL image processing, ELECTRONIC data processing
Abstract

The article focuses on the adoption of Digital Imaging and Communications in Nondestructive Evaluation (DICONDE) standard for the digital data exchange, storage and communication in nondestructive testing (NDT). It discusses the changes in DICONDE standard which has been based on the information object definition (IOD) of Digital Imaging and Communication in Medicine (DICOM) standard. It presents case studies including one on the benefit of DICONDE for NDT equipment and manufacturers.

Published
2010

Catalog

Books, media, physical & digital resources
See catalog results