Pre-requisites for SEE.4C hackathon (Toulon April 22, 2017)

First name

This is a required question

Last name

This is a required question

Codalab username from https://competitions.see4c.eu/

Please register to this server to complete your pre-requisites.

This is a required question

Proof of skills

One or two relevant publications you made related to the workshop topic or another evidence of your skill level.

This is a required question

"Free lunch" compulsory check list

In machine learning there is NO free lunch. To win your free food complete ALL the following prerequisites. This will also guarantee you WiFi access.

1. Signed up at https://competitions.see4c.eu/
2. From the BETA TEST WEBSITE (1), download the sample submission and submit it.
3. If you did this correctly, you should see your results on the leaderboard (2).
4. Download and install Anaconda for Python 2.7 http://continuum.io/downloads (3).
5. Make a submission generated with README.ipynb in the starting kit (4).
6. Make a small change in simple_model.py. Resubmit your code (5).

This is a required question

Comments

(1) The BETA TEST WEBSITE is found at: https://competitions.see4c.eu/competitions/7. Go to the "Participate" tab, accept the rules, go to "Get data" in the left side menu. Download the SAMPLE SUBMISSION. In the "Participate" tab, go to "View/Submit Results" to upload it. You can refresh to check the status of your submission. (2) Go to the "Results" tab to see your results on the leaderboard. (3) If you already have Anaconda, update it: conda update conda conda update anaconda older versions do not include h5py, which is needed. Eventually run conda install -c anaconda h5py=2.6.0 (4) Go to the "Participate" tab then go to "Get data". Download the STARTING KIT. Type at a command prompt jupyter-notebook README.ipynb. Execute all the cells (shift + return). The last cell generates a new sample submission. (5) Here is one way you can do that. The Spyder editor coming with Anaconda can be started with the Anaconda launcher. You can open in the editor the file sample_code/simple_model.py. Modify the method predict.py: instead of submitting the last frame, submit the average of the past few last frames. Go back to the jupyter-notebook to check your change works fine and generate your new submission. If you are smart, you can cheat by looking around in the starting kit: you will find one other example of prediction models that beat the baseline called "linear" see whether you can get it to work, modify it, understand it!

RMSE obtained

The score you got with your new modified model.py

This is a required question

Time taken

Time taken to do all the tasks above:

This is a required question

To better get ready

Read: https://arxiv.org/abs/1605.08104