Paradox ML Challenge

The Paradox ML Challenge is a two-stage competition organized by Google Developers Group JMI. Participants, including students and professionals, will have the opportunity to test their machine learning skills on a specially curated dataset based on a real-world problem statement.

Eligibility

Open to all individuals over the age of 18 at the time of entry.

Each person can be part of only one team at a time.

Competition Format

Round 1: Technical Round

Dataset Release: Available on Kaggle at 12:01 a.m. on 26/10/24.

Submission: Only the team leader is allowed to submit on Kaggle.

Code of Conduct: Strict adherence to the code of conduct is required; avoid using external APIs or prompting any pre-trained large language models during training or prediction phases.

Submission Format: All submissions must be made in a properly formatted CSV file. Each team is allowed three submissions per day, with a maximum of six attempts throughout the competition.

Public Solutions: Teams must make their solutions public on Kaggle. The top 10 submissions on the leaderboard at the end of 27/10/2024 will be selected for Round 2.

Round 2: Presentation Round

Date: Selected teams will present their solutions to the organizing team and judges on 07/11/2024.

Evaluation Criteria: Final evaluation based on the novelty of the solution, accuracy of results, and computation time.

Competition Rules

One Account Per Participant: Participants cannot sign up with multiple Kaggle accounts; only the team leader’s submission will be evaluated.

No Private Sharing: Code or data cannot be privately shared outside of the team. Sharing on forums is acceptable.

Public Dissemination: The competition host reserves the right to publicly disseminate entries or models.

Evaluation Metric: Scoring and ranking metrics will be displayed once the problem statement is live.

Open Licensing: Winning solutions must be available under an open-source license to qualify for recognition and prizes. Winners must link to their solutions within seven days of notification.

Team Size Limit: Maximum team size is two members; mergers are not allowed.

Dataset Manipulation: No manual or human effort is permitted in the test set. Only the provided data on Kaggle can be used for developing and testing models.

Tie Resolution: Ties will be resolved in favor of the submission that was submitted first.

Submission Constraints

Prohibited APIs: Use of APIs like OpenAI's GPT, Hugging Face, Gemini, or any other pre-trained language models during training or prediction is not allowed. Fine-tuning LLMs is allowed, but prompting for results is not.

Eligibility for Prizes: Submissions developed with code relying on open-source software licenses are ineligible for prizes.

By entering a submission, participants agree to the following:

All information provided is true and complete.

The submission is original work or used with permission, with proper credit given.

The submission does not contain confidential information, trade secrets, or violate any intellectual property rights.

The submission is free of malicious code.

The submission complies with all applicable laws and does not trigger any reporting obligations.

The submission has not been previously published or won any other award.