KDD 2017 : Knowledge Discovery and Data Mining
Conference Series : Knowledge Discovery and Data Mining
Call For Papers
KDD Cup 2017 Call for Proposals
We invite organization proposals for KDD Cup 2017. Starting from 1997, KDD Cup has been the most prestigious annual data mining competition held in conjunction with the ACM SIGKDD conference on Knowledge Discovery and Data Mining.
KDD 2017 will be held in Halifax, Canada from Aug 13th to Aug 17th, 2017. The competition is anticipated to last for 2~4 months, and the winner is supposed to be notified by mid-June. The winners will be announced during the KDD conference and present their solutions in the KDD Cup workshop during the conference.
This call for proposal solicits industry or academic institutes to submit their proposal as the potential organizer of the 2017 KDD Cup competition. We are looking for a strong proposal that contains most of the following features from a proposed task: a novel and motivated goal, a rigid and fair setup, challenging yet manageable task, and accessibility to the public.
1. A novel and motivated goal: Of particular interests are machine learning tasks that are different from a traditional (competition) setup, in which in the end a supervised learner is on demand given a set of training data with the goal to optimize typical prediction quality in the testing. We encourage organizers to ponder on a novel yet practical challenge that has broad real-world application scenarios and can lead to actionable knowledge. Examples such as learning with incrementally arrival data and evaluation on the accumulated error; prediction given limited amount of resources; learning with mostly unlabeled data; or addressing cold-start issues in learning, or learning with different types of data, etc. are highly recommended.
2. A rigid and fair setup: The organizers should guarantee the accessibility of the data and the confidentiality of the ground truth. The evaluation metrics should be both meaningful for the application and statistically sound for objective performance comparison. The baseline should be established with evidence to show that non-trivial performance can be achieved, and an estimate of what constitutes a significant difference in performance is preferable.
3. A challenging yet manageable task: The task should be challenging in the sense that there is decent room for improvement from the basic solutions, and novel ideas are required to succeed in the competition. The task should be manageable in about 3 months such that the competitors can mainly focus on solving the core challenges.
4. Accessibility: The competition should be accessible to the majority of the machine learners and data miners without excessive domain knowledge or powerful computational infrastructure.
Your proposal should answer the following questions:
1. How does the proposed challenge meet the four above-mentioned requirements?
2. Which competition infrastructure do you plan to use (e.g. Kaggle, or building your own)?
3. How many resources (including people, time, award money) do you plan to invest?
4. What is the time table for the competition?
5. Is there any concern of the privacy about the released data? Have you obtained the right to release the data for the competition from the legal department of your institute?
6. Do you require the winners to submit the source code of their winning solutions?
7. How would you handle Q&A and possible revision during the competition?
8. What is your basic solution?
and also include:
9. Names, affiliations, postal addresses, phone numbers, and short biographies of the organizers.
10. An endorsement letter from the higher-level management of the organization.
Please keep the proposal concise. Please send your proposals in PDF format to email@example.com by December 9th, 2016.
October 9, 2016 - CFP Release
December 9, 2016 - Proposal Submission Deadline
December 31, 2016 - Decision Notification
February - March, 2017 - KDD Cup starts
June 2017: Announcement of the KDD Cup Winner