The presentations are 25 min long, including questions. So plan for 20 minutes to present and 5 min for questions. You should convey the major idea in the paper, you do not (and should not) explain every single little detail.
Some useful points:
- Give the paper title and the authors on the first slide
- Clearly state the problem the paper tackles
- Explain the prior work on this problem and its limitation (from the prior work section in the paper
- Briefly state which limitations of the prior work the paper proposes to solve and which computational techniques will be used
- Explain the main technical details (the "meat" of the paper)
- Experimental results
- Conclusions (which issues are left to solve, future work, etc...)
- Try to use as many pictures as possible instead of text/formulas
Possible papers for presentations
Presentations on December 2
-
- Islam: "Probabilistic expression analysis on manifolds" by
Ya Chang Changbo Hu Turk, M.
- Wei: "Learning Object Detection from a Small Number of Examples: the Importance
of Good Features" by K. Levi and Y. Weiss.
- Alex
"A linear time algorithm for computing exact Euclidean distance transforms of binary images in arbitrary dimensions" by Maurer, C.R., Jr., Rensheng Qi, Raghavan, V.
- Tharindu:
"A Discriminatively Trained, Multiscale, Deformable Part Model" by
P. Felzenszwalb, D. McAllester, D. Ramanan, In Proc. IEEE CVPR 2008.