Announcing the Winner of the Acumos AI Challenge

Congratulations to the winner of the Acumos AI Challenge - Chris Buonocore of Brookline, MA! Chris took home the top prize of $50,000 for his entry, the Acumos Property Assistant. We also awarded two runner-up prizes of $25,000 each to Rhutvij Savant, for Cancer Classify, and to Jinhe Shi & Yuhua Gong for User Review Prediction. Thank you to everyone who entered the competition!

The Acumos AI Challenge, sponsored by AT&T and Tech Mahindra, is an open source developer competition for the Acumos AI platform seeking innovative artificial intelligence (AI) solutions. The companies, along with The Linux Foundation, encourage students, developers, and data scientists to submit their ground-breaking AI solutions for a chance to win prize money and industry recognition. 

Acumos AI is a platform and open source framework that makes it easy to build, share and deploy AI applications. The Acumos AI platform, hosted by The Linux Foundation's LF Deep Learning Foundation, simplifies development and provides a marketplace for accessing, using and enhancing AI apps.

Contestants will compete for a share of $100,000 in cash prizes, including:

    • The top three teams will win a trip (for two members) to participate in the finals in San Francisco on or about September 10, 2018.
    • The 1st place team will win the $50,000 Grand Prize.
    • The 2nd and 3rd place teams will each be awarded $25,000.
    • Finalists will also have their solution featured in the Acumos Marketplace, receive PR exposure, and meet with AT&T and Tech Mahindra executives during the event.
    • In addition, all eligible AI solutions will be made available as open source in the Acumos Marketplace.

 

What Types of AI Solutions Are We Seeking?

The Acumos AI Challenge is seeking innovative, open source AI models across all use cases, including, but not limited to:

5G & SDN

AI models that improve the overall performance and efficiencies of 5G networks and Software-Defined Networking (e.g. antenna control optimization).

Media & Entertainment

AI models targeting a media or entertainment use case. Examples include solutions for broadcast media, internet, film, social media, ad campaign analysis, video and image recognition, speech & sound recognition, video insight tools, and more.

Security

AI models around network security use cases such as advanced threat protection, cyber security, IoT security, and more.

Enterprise Solutions

AI models targeting an enterprise use case, including solutions for automotive, IoT, home automation, mobile, infrastructure, manufacturing, real estate, retail, travel, transportation & logistics, utilities (energy, water & others) distribution.

Since it is so easy to onboard new models into Acumos, there are nearly an infinite number of use cases to consider that can benefit consumers and businesses across a multitude of disciplines. When submitting your entry, we encourage you to consider all scenarios that you are passionate about.

AT&T and Tech Mahindra created the Acumos AI Challenge, together with The Linux Foundation, to enable and accelerate AI adoption and innovation, while recognizing developers who are paving the future of AI development.

We look forward to your entry and hope to see you at the finals in September!

View full rules

Eligibility

  • Open to individuals or teams of eligible individuals
  • Contest is open only to entrants who are legal U.S. residents and are physically located in one of the 50 United States, District of Columbia, and Puerto Rico
  • Entrants must be at least 18 years old at time of entry (except Alabama and Nebraska residents who must be at least 19, and Mississippi residents must be at least 21)
  • Selected finalists must be able to attend the finals on or about Sept.10, 2018 in San Francisco, CA. Travel and hotel will be provided for two members from each finalist team
  • Employees of AT&T, Tech Mahindra or its affiliates are not eligible to participate
  • See Official Rules for complete submission and eligibility requirements

Requirements

  • Submissions will be accepted between May 31 - August 5, 2018.
  • Teams are required to enter their submission via the Acumos AI Challenge portal at https://acumos-challenge.org
  • Submission must include:
    • Model source code
    • Written description of your solution
    • Summary Video
      • Video must be no more than three minutes long, and must demonstrate your solution in action, including screenshots or video of your solution, and feature a minimum of one voiceover from an Entrant team member. The Video can be uploaded during submission as a .mp4 or .mov file or Entrant can share a Youtube or Vimeo link to their Video.
    • Sample dataset produced by the model
    • Sample application for testing the model
    • Documentation explaining how to deploy the model, how do deploy the demonstration of the model, and all supporting toolkits and programming languages.
      • Documentation must be submitted under Creative Commons Attribution 4.0 International License (CC BY 4.0). Entrants can visit this guide to learn more.
  • Submission code must be provided under Apache 2.0 license, and all referenced or linked code must be clearly attributed and licensed with an open source license that is compatible with Apache 2.0. For instructions on how to implement this, visit this guide
  • To the extent any contributed model includes data, any rights in such data shall be made available under the CDLA-Permissive 1.0 License.
  • Entry must be submitted with the following information: Entrant's (including each team member's ) full name, email address, and physical address.
  • Submission should be new work (created after 1/1/2017), or substantially enhances an existing open source AI model.
  • There is no limit to the number of submissions per entrant.
  • Selected finalists must be able to attend the finals on Sept.10, 2018 in San Francisco, CA. Travel and hotel will be provided for two members from each finalist team.
  • See Official Rules for complete submission and eligibility requirements.

How to enter

Required Steps for the Acumos AI Challenge on Devpost

  1. Register for the challenge at https://acumosaichallenge.devpost.com. If you do not already have a Devpost account, you will be taken through a simple sign up wizard.
  2. Once you have registered for the Challenge on Devpost, an account will be created for you on the Acumos platform at https://acumos-challenge.org and you will receive an email with your login credentials.
  3. On Devpost, you will need to "Enter a submission" by pressing the "Enter a submission" button, and then add a new project, or select an existing project from your Devpost account. This is where you will enter required fields about your project:
    • Enter the name of your project
    • Enter a short description of your project
    • Add team members (optional)
    • Here's the whole story: this is where you'll describe your project in more detail.
    • Accept the Official Rules and submit your project by pressing the "Submit to Acumos AI Challenge" button. You can continue to edit your submission until the submission deadline.

Required Steps on the Acumos Challenge Portal

  1. Once you have registered for the Challenge on Devpost, an account will be created for you on the Acumos platform at https://acumos-challenge.org and you will receive an email with your login credentials.
  2. Once you receive your Acumos platform account credentials, it is recommended that you login and get familiarized with the UI.
  3. Install the client library for the modeling toolkit that you are working with.
  4. Once you are ready to onboard your model, you will go through the Onboarding workflow after which your model will be in your private "My Models" area, and a deployable microservice docker image will have been created and saved in the Acumos Nexus repository (accessible only to you). NOTE: your model is not published or submitted yet, so you can continue to work on it in private until you are ready, repeating the onboarding process etc. as needed.
  5. Use the "My Models" tools to add documentation, a license/copyright notice, test data, and summary video as metadata for your model.
  6. Use the "My Models" tools to deploy/download your model, so that you verify the ability of the Challenge judges to do the same. You may optionally share your model with other devs that you want to work with as a team.
  7. When you are ready to submit your entry with all required items, you will use the Publishing a Model workflow to submit it. Note that the metadata about your model (e.g. documents, description) will then be viewable in the platform Marketplace by other devs, but the code/microservice for your model will not be accessible to others until after the Challenge.
  8. IMPORTANT FINAL STEPS: On our Devpost site  (https://acumosaichallenge.devpost.com), if you haven't already done so, you will need to "Enter a submission" by pressing the "Enter a submission" button, and then add a new project, or select an existing project from your Devpost account. This is where you will enter all required fields about your project:
    • Enter the name of your project
    • Enter a short description of your project
    • Add team members (optional)
    • Here's the whole story: this is where you'll describe your project in more detail
    • Accept the Official Rules and submit your project by pressing the "Submit to Acumos AI Challenge" button. You can continue to edit your submission until the submission deadline.

For a complete list of entry requirements, see Official Rules.

Entries must be received by 11:45pm PT, August 5, 2018 to be considered.

 

Judges

Panel of Judges

Panel of Judges
(coming soon)

Judging Criteria

  • Novelty & Originality (20%)
    Models will be compared to other submissions and judged based on inventiveness.
  • Viability and Impact on Market (20%)
    Judges will evaluate both the practicality of the solution and how significant an impact it will have.
  • Difficulty of Technical Implementation (20%)
    Judges will compare how much effort was done on the model compared to other submissions.
  • Best Supports Package Requirements and Intended Function (20%)
    Model will be evaluated on how well the package supports required items (Classification, Data Transformer, Prediction), containing code that uses a modeling toolkit) and how well the solution meets target functions compared to other models.
  • Performance (20%)
    Models will be evaluated based on performance measures such as accuracy, speed, scalability and quality using known methods used to benchmark optimized models.
  • ------------------------------------
    Acumos, Acumos AI and the Acumos logo are trademarks of LF Projects, LLC.