IBM OpenPOWER Developer Challenge

From Rethink Wiki
Jump to: navigation, search

Introduction

About OpenPOWER Developer Challenge: Cognitive Cup

The IBM OpenPOWER Developer Challenge is to exploit IBM’s hardware super-computing resources and develop a technically innovative solution to a real world problem.

There are 3 tracks:

  1. ArtNet – Use Deep Learning to train Neural Networks to analyze paintings
  2. TuneNet – Guide programmer optimization using Deep Learning
  3. YourNet – define and solve your own Deep Learning problem

Application deadline

Apply here

About OpenPOWER

The OpenPOWER Foundation is an open technical community based on the POWER architecture by IBM, Nvidia and others to build open data center technologies.

Reward

1st Prize:

  • One representative from the winning team or winning individual will receive an all-expenses paid trip to Supercomputing 2016 in Salt Lake City, UT.
  • The entire winning team (up to 5 people) will win an Apple Watch Sport. Approximate Retail Value: US$3,500 for trip US$299/watch

2nd Prize:

  • iPad Air 2 16Gb (up to 5 per winning team)
  • Approximate Retail Value: US$499/ iPad

3rd Prize:

  • Apple Watch Sport (up to 5 per winning team).
  • Approximate Retail Value: US$299/ watch

Eligibility

Individuals who are at least 18 years of age as of the time of entry, teams of eligible individuals, and organizations with no more than 50 employees from the following countries: Australia, Canada (excluding Quebec), China (including Hong Kong), France, Germany, India, Israel, Japan, Mexico, Netherlands, South Korea, United Kingdom, and the United States.

Judging Criteria

  • Demonstrated ability to maximize accuracy (50% weighting)
    • End-to-end accuracy achieved for the problem
  • Spark Usage (25% weighting)
    • Does the solution incorporate Spark as the framework to scale beyond a single node?
    • How well does the solution incorporate Spark to scale?
  • Technical innovation (25% weighting)
    • How well is Deep Learning exploited as part of the solution?
    • Does the solution incorporate any new approaches to the Deep Learning problem (e.g. did the submission involve creating new code, algorithms, or leverage emerging frameworks)?

Application Process

Sign up on [devpost.com devpost.com]. Submission is to be done through the Devpost website. Upon completion of the project, you are expected to publish your code on Github and share the link. You are also required to create a 1-minute video showcasing the output of your solution.

Previous Winners

Praveen Sridhar

Blog on Praveen's Experience

medium link here.

Resources

  1. Openpower foundation link
  2. Challenge listing on Devpost
  3. Cognitive Cup Listing on Devpost
  4. OpenPOWER Resources
  5. Supercomputing Conference, 2016
  6. Praveen's Pitch Deck at SC16