Machine Learning Talk: Support Vector Machines

26.01.18 by Jonas Krug, Tim Schlottmann, Hendrik Sieck

Presentation slides

Presentation Slides for Linear SVM and SVM with Kernels (PDF)

Presentation notebooks (code)

1. Implemented SVM with Gaussian kernel (iPYNB)
2. MNIST handwriting example (iPYNB)
3. Error analysis (iPYNB)

Presentation notebook notes (PDF)

Run presentation notebooks

1. Install Docker
2. Run pull.sh
3. Run start-notebook.sh
4. Locate a browser to the URL and execute the notebooks

Benchmarks

• Intel 5200U @ 2.2 GHz needs 56 minutes for all notebooks executing in parallel
• Intel Xeon E3-1240 v5 @ 3.5 GHz needs ca. 30-40 minutes

The complete variable workspace of all notebooks require ca. 5-6 GB RAM.

Handout/Summary

Handout/Summary (PDF)

License

All code and all presentation slides are licensed with GPLv3, see LICENSE for more information.

Credits: Latex beamer theme by Till Tantau (2007, GPL), modifications by Mijail Guillemard.