Welcome! Please contribute your ideas for what challenges we might aspire to solve, changes in our community that can improve machine learning impact, and examples of machine learning projects that have had tangible impact.
Lacking context for this site? Read the original paper: Machine Learning that Matters (ICML 2012). You can also review the slides.
Look at kaggle.com
  • Kaggle.com is a place where machine learning meets the real world. Jeremy Howard of kaggle has a few talks up on youtube about this.
  • 4 Comments sorted by
  • Interesting site!  It is a clearinghouse for organizations to post data analysis challenges and for researchers to compete to find the best solution.

    Have you participated in a competition?  What was it like?

    I wonder if there are articles written up anywhere about the benefits of the solutions.  Obviously, many would be proprietary results, but some could be published.  The website claims "Kaggle is proud to have achieved extraordinary results that have outperformed betting markets and advanced the state of the art in HIV research and chess ratings." so I would be interested in more details.  (The links just go to the project sites.)

    This anecdote from the site's blog is a fascinating tale of classic overfitting (and how easy it is to do).
  • I haven't personally participated yet, however, the youtube talk be jeremy howard gives a good overview  (each challenge has given better than state of the art results, there is private competition component).

    One idea is that of combining kickstarter and kaggle - raise funds on kickstarter and launch competitions on kaggle (for high impact machine learning applications).
  • I have participated in one competition from that site. Usually they give you straight those feature vectors and ask you to build a classifier/regressor that fits best their test set (+ for winners, another test set to verify). I still think this may be too focused on that type of data but at least it is solving some interesting and useful problems (fraud detections in car selling announcements, coming back patient prediction based on their medical history, hand written text recognition and so on).
    A missing part is that the obtained knowledge is not communicated (even though there is a forum for discussions) and mainly stays with the experimenter or the organizer of the competition. At least a report or an article showing the main conclusions could be welcome for wider public.
  • To link this to the ml impact paper, kaggle should be required practice for every ml student, and maybe even ml professors/practitioners should be required to show kaggle credentials as a requirement for promotion/tenure.


To post or add a comment, please sign in or register.

Tip: click the star icon to bookmark (follow) a discussion. You will receive email notifications of subsequent activity.
If search doesn't work, try putting a + in front of your search term.