Show filters Hide filters

Refine your search

Publication Year
1-7 out of 7 results
Change view
  • Sort by:
42:16 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) English 2017

An introduction to inverse problems with applications in machine learning

The presentation starts with some motivating examples of Inverse Problems before introducing the general setting. We shortly review the most common regularization approaches (Tikhonov, iteration methods) and sketch some recent developments in sparsity and machine learning. Sparsity refers to additional expert information on the desired reconstruction, namely, that is has a finite expension in some predefined basis or frame. In machine learning we focus on 'multi colored' inverse problems, where part of the application can be formulated by a strict analytical framework but some part of the problem needs to modeled by a data driven approach. Those combined problems can be created by data- driven linear low rank approximations or more general black box models. In particular we review deep learning approaches to inverse problems. Finally, machine learning techniques by themselves are often inverse problems. We highlight basis learning techniques and applications to hyperspectral image analysis.
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: English
25:36 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) German 2017

MRI data models at low SNR

In this talk we elaborate the effect of low SNR on estimated parameters in models for neuroimaging data. We will present a new method for the local estimation of the noise parameter in the signal distribution and demonstrate how this can be used for improved estimation of model parameters. (joint work with Joerg Polzehl).
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: German
27:38 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) German 2017

Lost in the Web?

To get information of mathematical software is sometimes like finding a needle in a haystack. But otherwise, more and more research results will be achieved by using mathematical software. Comprehensive and precise information about software used is also an important condition for the credibility of research results. But software has some unique features which pose new requirements to information infrastructure, especially the dynamic character of software, its encoding in formal languages, dependencies from hardware, other software, and further context information. The information on mathematical software is widely distributed on Web sites, repositories, portals, Web archives, etc., and is not standardized. The talk will address two relevant activities which can significantly improve the information about mathematical software: the initiatives for a citation standard for software, and the development of a comprehensive portal for mathematical software which integrates the existing information about software from the Web. A state of the art report on the swMATH service will illustrate the concepts and approaches.
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: German
14:53 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) English 2017

Estimation of cardiovascular system parameter from real data

This talks shows preliminary results for the estimation of some parameters regarding the cardiovascular system (e.g., vessel mechanical properties, vasculature resistance) using real patient data. The mathematical model is based on efficient one-dimensional network for the blood flow simulation and on the unscented Kalman filtering for the parameter estimation.
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: English
13:52 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) German 2017

An online coupled Lagrangian particle dispersion model for COSMO

Lagrangian particle dispersion models (LPDM) are a well-known method for modeling exhaust gas distributions and similar problems. Open accessible LPDM's, e.g., the FLEXPART model are designed for meso-scale simulations and work offline coupled with the coarse frequented output data of any numerical weather prediction model. But the central issue there is that high resolution simulations of small scale phenomena need a high frequency input of meteorological data fields to work accuracy, which can directly be provided by an online coupled model system. Based on the COSMO trajectory module the model LAPASI was developed that integrates an online coupled Lagrangian particle transport into the default COSMO version. It supports any kind of simulation that are possible with the COSMO including idealized cases and can handle a couple Million particles with individual start times, start locations and dry depositions velocities. Thus, LAPASI is a useful extension for COSMO and a necessary addition to the previously existing LPDM's.
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: German
26:33 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) German 2017

Long-Term Accessibility of Software through Web Archives

Software is widely used and referenced in research and scientific publications. Hence, just like the results published in an article, associated software should be preserved and made long-term accessibly as well. Due to the nature of software and its dynamic aspects, this is rather challenging though. Very commonly, only related materials, such as a software's documentation, parts of its source code or change logs, are freely available. However, these can be very valuable to comprehend or reproduce experiments described in literature. We found that a big portion of this data is provided on the Web. Around 60% of the software webpages we analyzed link to documentation, while another 50% even contain some artifacts of the actual software [1, 2]. Web Archives are a way to preserve this information and allow for long-term accessibility, even if the software and corresponding information change. Although many of those websites are already archived, our study shows that the evolution of a software is not always well captured. Therefore, we are working towards a pro-active approach to archive software on the Web in the future as part of the scientific process. [1] Holzmann, H., Runnwerth, M., Sperber, W.: Linking Mathematical Software in Web Archives. 5th International Congress on Mathematical Software, ICMS 2016. Berlin, Germany (2016). [2] Holzmann, H., Sperber, W., Runnwerth, M.:?Archiving Software Surrogates on the Web for Future Reference.20th International Conference on Theory and Practice of Digital Libraries, TPDL 2016, Hannover, Germany(2016).
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: German
18:34 Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB) German 2017

Spatial Segmentation Algorithm for Stratified Random Sample considering geometric circumstances

We developed a Segmentation Algorithm, which is based on Region Growing, for Stratified Random Sample. But instead of raster data our algorithm is using vector data! By the use of accurate vector data we can easily implement barriers, like rivers or railways, to control the growing we need to. Also by splitting polygons we achieve a kind of density gradient to slow the grow in unwanted regions, like bridges. As well we had to remove a lot of bottlenecks to reduce runtime and make it performant and practicable. The result are approximately optimal segmented regions, which are met certain conditions regarding the sample design and take consideration to geometric circumstances. In our case it is useful for reduction of costs on Germany-wide field survey for random sample of non-residential building, but is also applicable to any other wanted segmentation.
  • Published: 2017
  • Publisher: Weierstraß-Institut für Angewandte Analysis und Stochastik (WIAS), Technische Informationsbibliothek (TIB)
  • Language: German
out of 1 pages
Loading...
Feedback

Timings

   51 ms - page object
   40 ms - search
    2 ms - highlighting
    1 ms - highlighting/21900
    1 ms - highlighting/21910
    2 ms - highlighting/21899
    1 ms - highlighting/21905
    1 ms - highlighting/21913
    1 ms - highlighting/21902
    2 ms - highlighting/21904

Version

AV-Portal 3.8.0 (dec2fe8b0ce2e718d55d6f23ab68f0b2424a1f3f)