Difference between revisions of "Scattering Networks for Scene Labeling"
From iis-projects
m (→Status: Available) |
m |
||
| (5 intermediate revisions by the same user not shown) | |||
| Line 4: | Line 4: | ||
Scattering networks [1,2] extract characteristic features from signals by recursively applying the composition of the following three operations: convolution with a set of filter functions, a non-linearity, and a sub-sampling step. Such networks, combined with a classifier (such as, e.g., support vector machines) were successfully employed in a number of classification tasks [1,3]. In contrast to traditional convolutional neural networks which learn the filters from training data [4], scattering networks use pre-defined filters such as wavelets [1], curvelets [2], or shearlets [2]. While a scattering network-based classifier arguably leads to less flexible machine learning models than convolutional network-based classifiers which learn the filters, they potentially allow for faster implementations thanks to the structure of the pre-defined filters (e.g., tensorized wavelet filters). | Scattering networks [1,2] extract characteristic features from signals by recursively applying the composition of the following three operations: convolution with a set of filter functions, a non-linearity, and a sub-sampling step. Such networks, combined with a classifier (such as, e.g., support vector machines) were successfully employed in a number of classification tasks [1,3]. In contrast to traditional convolutional neural networks which learn the filters from training data [4], scattering networks use pre-defined filters such as wavelets [1], curvelets [2], or shearlets [2]. While a scattering network-based classifier arguably leads to less flexible machine learning models than convolutional network-based classifiers which learn the filters, they potentially allow for faster implementations thanks to the structure of the pre-defined filters (e.g., tensorized wavelet filters). | ||
| − | + | This project shall explore the application of scattering network-based classifiers to the problem of scene labeling, whose aim is to assign a class label such as "street", "tree", or "building" to every pixel of an image. Applications of scene labeling include situational awareness systems [4], which often demand low-complexity and low-power scene labeling. | |
The goal of this project is to develop a scattering network-based classifier for scene labeling and to realize a GPU implementation thereof (followed by VLSI/FPGA, if time permits). | The goal of this project is to develop a scattering network-based classifier for scene labeling and to realize a GPU implementation thereof (followed by VLSI/FPGA, if time permits). | ||
| − | ===Status: | + | ===Status: Completed=== |
| + | : Fall Semester 2015 | ||
| + | |||
Contact/Supervision: | Contact/Supervision: | ||
| − | * Michael Tschannen (michaelt ät nari.ee.ethz.ch) | + | * [http://www.nari.ee.ethz.ch/commth/people/show/michaelt Michael Tschannen] (michaelt ät nari.ee.ethz.ch) |
| − | * Thomas Wiatowski (withomas ät nari.ee.ethz.ch) | + | * [http://www.nari.ee.ethz.ch/commth/people/show/withomas Thomas Wiatowski] (withomas ät nari.ee.ethz.ch) |
* [[:User:Lukasc | Lukas Cavigelli]] | * [[:User:Lukasc | Lukas Cavigelli]] | ||
| − | * Michael Lerjen (mlerjen ät nari.ee.ethz.ch) | + | * [http://www.nari.ee.ethz.ch/commth/people/show/mlerjen Michael Lerjen] (mlerjen ät nari.ee.ethz.ch) |
This project is a collaboration between the communication theory group and the digital circuits and systems lab. | This project is a collaboration between the communication theory group and the digital circuits and systems lab. | ||
| − | + | [[Category:System Design]] [[Category:Completed]] [[Category:Semester Thesis]] [[Category:Lukasc]] [[Category:2015]] | |
<!-- | <!-- | ||
===Status: {Available, Reserved, In Progress, Completed}=== | ===Status: {Available, Reserved, In Progress, Completed}=== | ||
| Line 37: | Line 39: | ||
===Professor=== | ===Professor=== | ||
| − | + | [http://www.nari.ee.ethz.ch/commth/people/show/boelcskei Helmut Bölcskei] | |
| − | + | ||
| + | [http://www.iis.ee.ethz.ch/portrait/staff/lbenini.en.html Luca Benini] | ||
==References== | ==References== | ||
Latest revision as of 11:29, 5 February 2016
Contents
Description
Scattering networks [1,2] extract characteristic features from signals by recursively applying the composition of the following three operations: convolution with a set of filter functions, a non-linearity, and a sub-sampling step. Such networks, combined with a classifier (such as, e.g., support vector machines) were successfully employed in a number of classification tasks [1,3]. In contrast to traditional convolutional neural networks which learn the filters from training data [4], scattering networks use pre-defined filters such as wavelets [1], curvelets [2], or shearlets [2]. While a scattering network-based classifier arguably leads to less flexible machine learning models than convolutional network-based classifiers which learn the filters, they potentially allow for faster implementations thanks to the structure of the pre-defined filters (e.g., tensorized wavelet filters).
This project shall explore the application of scattering network-based classifiers to the problem of scene labeling, whose aim is to assign a class label such as "street", "tree", or "building" to every pixel of an image. Applications of scene labeling include situational awareness systems [4], which often demand low-complexity and low-power scene labeling.
The goal of this project is to develop a scattering network-based classifier for scene labeling and to realize a GPU implementation thereof (followed by VLSI/FPGA, if time permits).
Status: Completed
- Fall Semester 2015
Contact/Supervision:
- Michael Tschannen (michaelt ät nari.ee.ethz.ch)
- Thomas Wiatowski (withomas ät nari.ee.ethz.ch)
- Lukas Cavigelli
- Michael Lerjen (mlerjen ät nari.ee.ethz.ch)
This project is a collaboration between the communication theory group and the digital circuits and systems lab.
Prerequisites
- Matlab, C programming, linear algebra
Character
- 0%-20% Theory
- 80%-100% Programming
Professor
References
- J. Bruna and S. Mallat, “Invariant scattering convolution networks,” IEEE Trans. Pattern Anal. Mach.Intell., vol. 35, no. 8, pp. 1872-1886, 2013. link
- T. Wiatowski and H. B¨olcskei, “Deep convolutional neural networks based on semi-discrete frames,” Proc. of IEEE International Conference on Information Theory (ISIT), pp. 1212-1216, 2015. link
- J. Andén and S. Mallat, “Deep scattering spectrum,” IEEE Trans. on Signal Process., vol. 62, no. 16, pp. 4114-4128, 2014. link
- Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proc. of the IEEE, pp. 2278-2324, 1998. link
- L. Cavigelli, M. Magno, and L. Benini, “Accelerating real-time embedded scene labeling with convolutional networks,” Proc. of ACM/EDAC/IEEE Design Automation Conference (DAC), pp. 1-6, 2015 link
