2024 Machine Learning Marathon

In collaboration with MadAI and the Open Source Program Office, the ML+X community invites both beginner and advanced machine learning (ML) practitioners to join the 2024 Machine Learning Marathon (MLM24)!

Running from September 12 to December 12, this 12-week hackathon offers real-world ML projects (“Challenges“) catering to different skill levels, application areas, and ML/AI methods. Many of the challenges are inspired by or directly sourced from research applications here at UW-Madison! Participants will form local or remote teams of 2-5 members, dedicating at least 3 hours/week to complete their chosen challenge, with the potential to win prizes. Weekly “sprint” events will facilitate knowledge exchange and discussion on ML tools and strategies. Each team will also be paired with an advisor to guide their progress. Explore the sections below for detailed information on the available challenges, event schedule, registration, rules, code of conduct, and more.

Registration Deadline: Ready to join as a participant or team advisor? Please register by 9/10 so we can plan our catering needs. If you have any questions or require additional information, please contact ml-marathon-organizers@g-groups.wisc.edu.

Quick Links

Challenges & Teams

Select a challenge

There are a variety of challenges available catering to different skill levels, application areas, and ML/AI methods. Beginners should check out the “Getting Started” challenges, while more experienced practitioners should check out the “Reproduce & Enhance” or “Active” challenges. Click each of the accordion menu options below see which category best suits your learning and event goals. From each challenge page, we recommend you review the “Overview”, “Rules”, “Discussion” , and “Code” tabs to become more familiar with each challenge and see how others are beginning to approach the problem.

Form a team!

Once you have a favorite challenge in mind, visit the MLM24 Slack channel (you may need to join the Data Science Hub Slack first) to begin searching for team members to work with. Please share in this channel:

Whether you’ll be participating remotely or in-person (Madison, WI)
Your preferred challenge(s)
Your previous experience working on ML projects (all experience levels welcome!)
Your goal(s) for participating in the event
How many hours/week you realistically expect to commit to the project, including adjustments for fall semester

Team formation deadline: Once your team is ready, have one person fill out the Team Registration form. The deadline to form a team is Thursday, 9/19, 11:59PM. However, we recommend you try to form a team sooner rather than later!

Getting Started

Description: The “Getting Started” challenges are intended to teach ML fundamentals in an accessible and hands-on manner. If you’re relatively new to the world of ML or want to explore one of the below topics in detail, you may want to pick one of these challenges.

Deadline: There is no official end-date to these since they are ongoing. However, for the purposes of the ML Marathon event, you will be expected to submit a solution by 12/11.

Award: Learn machine learning fundamentals with others! Since these challenges are primarily intended to help you learn and obtaining 100% accuracy is sometimes possible, participants in the “Getting Started” challenges will not be eligible to win the $50 gift card.

Digit Recognizer (Computer Vision)
Predict CO2 Emissions in Rwanda (Time-Series Forecasting)
Use GANs to Create Art (Generative Art)
Natural Language Processing with Disaster Tweets (NLP)
Detecting Contradiction and Entailment in Multilingual Text (NLP)
Predict Sales Prices and Practice Feature Engineering, Random Forests, and Gradient Boosting (High-Dimensional Modeling)

Reproduce & Enhance

Description: The “Reproduce & Enhance” challenges are intermediate/advanced challenges which have already taken place within the global Kaggle community. The benefit of choosing a challenge which has already transpired is that you can review past submissions and discussions to try to reproduce the solution yourself — then build from there! This might be a good fit for you if there’s an application you really want to reproduce or you lack the time to develop new solutions yourself.

Deadline: There is no official end-date to these challenges have already been completed. However, for the purposes of the ML Marathon event, you will be expected to submit an alternative solution by 12/11.

Award: Learn how to apply the best techniques towards a given ML problem/dataset! Since the solution is provided, participants in the “Reproduce & Enhance” category will not be eligible to win the $50 gift card award.

Active

Description: The “Active” challenges are intermediate/advanced challenges which are ongoing on the Kaggle platform. This means that a cash prize may be available (check challenge rules) for those that want to shoot for it! These projects will require dedication and a time commitment of at least 3-5 hours/week (per member) to meet the challenge deadline.

Global deadline (Kaggle prize): Each challenge has its own deadline to be considered for the grand prize offered by Kaggle. Visit the Kaggle page for your challenge to learn more.

Local deadline ($50 Gift Card): Only participants that choose an “Active” challenge and submit a solution by 12/11 will be eligible to win the $50 gift card award. The team with the highest leaderboard ranking (across all “Active” challenges) will receive a $50 Amazon gift card for each member of the team. Some restrictions may apply.

Kaggle periodically posts new challenges. Check this webpage again in July to see if additional projects may be available! If you are interested in hosting a challenge yourself, please contact ml-marathon-organizers@g-groups.wisc.edu.

Schedule

All times are listed in US central time. All events will also have a Zoom option for remote participants to join. Events that start at 5:30PM will be catered. Please feel free to bring food to the events that start earlier at 4:30PM.

9/10 (Thur), 11:59PM: Deadline to register
9/12 (Thur), 5:30-7:30PM; Rm. 1240, Computer Sciences: Kickoff
1. Overview of projects; Resources for getting started; Meet other participants
9/19 (Thur), 11:59PM: Deadline to form team
9/19 (Thur), 4:30-6:30PM; Rm. 1240, Computer Sciences: Sprint 1
1. Coworking time; Demo presentation
9/26 (Thur), 4:30-6:30PM; Rm. 1240, Computer Sciences: Sprint 2
1. Coworking time; Demo presentation
10/3 (Thur), 4:30-6:30PM; Orchard View, Discovery Building: Sprint 3
1. Exploratory data analysis presentations; Coworking time; Demo presentation
10/10 (Thur), 4:30-6:30PM, Orchard View, Discovery Building: Sprint 4
1. Coworking time; Demo presentation
10/17 (Thur), 4:30-6:30PM; Rm. 1240, Computer Sciences: Sprint 5
1. Progress report presentations; Coworking time; Demo presentation
11/21 (Thur), 4:30-6:30PM, Rm. 1240, Computer Sciences: Sprint 6
1. Draft of solution presentations; Coworking time; Demo presentation
12/10 (Tue): Optionally present solutions (20-min) at ML+X forum or MadAI event
1. 12-1PM, Orchard View, Discovery Building; 20-min presentation @ ML+X forum
2. 5:30-7:30PM, Room TBD; 20-min presentation @ MadAI event
12/11 (Wed), 11:59PM: Project submissions due
12/12 (Thur), 5:30-7:30PM, Orchard View, Discovery Building: Lightning talks & project evaluations

Rules

All participants are expected to follow the rules detailed below.

Team Size: The maximum team size is 5. There is no minimum team size. However, in the Kaggle spirit of collaboration, we recommend you find at least one other person to work with.
Virtual Participation: Members may participate remotely/virtually but only with other remote participants. This is primarily to prevent communication problems during the live “sprint” events.
One Challenge Per Participant: To ensure efforts are not spread too thin, each participant may only compete in one challenge/team.
$50 Gift Card Award: Only participants/teams that choose an “Active Challenge” (see Available Challenges) will be eligible to win a $50 gift card. The team with the highest leaderboard ranking (across all ongoing global challenges) will receive a $50 Amazon gift card for each member of the team.
Challenge Specific Rules: Be sure to review the “Rules” section of your chosen challenge to see if any additional rules may apply.
Code of Conduct: All participants are expected to abide by the Code of Conduct to maintain a collaborative and friendly learning environment.

Code of Conduct

In order to maintain a collaborative and friendly learning environment, all event participants and advisors are expected to adhere to the following Code of Conduct.

Respect and Inclusivity: Encourage a diverse range of perspectives and backgrounds. Avoid offensive language or comments that might make others uncomfortable.
Active Listening: Aim to talk 20% of the time and listen the other 80%. Avoid interrupting or dominating conversations; allow others to share their thoughts.
Ask Questions and Share Knowledge: Share your knowledge and insights generously, but in a way that is understandable to those with less experience.
Be Patient and Supportive: Be patient with individuals who may not have as much experience.Offer assistance and mentorship when appropriate and requested.

Mindful Communication: Explain terms that others may be less familiar with
Feedback and Critique: Provide constructive feedback when necessary, but do so in a respectful and considerate manner. Be open to receiving feedback on your own work and ideas.
Embrace the Kaggle Spirit! Kaggle thrives on open-source collaboration. Participants should actively contribute to the community by sharing their work, kernels, and insights.

Registration

To help cover catering and room reservation costs, we are charging a nominal fee of $15 (+$2.85 processing fee) to participate in the 2024 Machine Learning Marathon. However, to encourage additional knowledge dissemination during and following the event, we are happy to grant a refund (minus the processing fee) to participants that complete one of the following qualifying activities:

Present a 10-30 minute demo on a machine learning tool, strategy, or resource at one of the “sprint” events (see Schedule)
Present your methods and results at December’s ML+X forum; 20 min. presentation (12/10, 12-1pm CT, in-person or Zoom)
Present your methods and results at December’s MadAI meetup; 20 min. presentation (12/10, 5:30-7:30pm CT, in-person or Zoom)
Write a short blog about your experience working on the Kaggle challenge, including a detailed report of your methods and results (further instructions will be provided at the kickoff event)
Contribute a useful machine learning resource/tutorial/blog/etc. to ML+X Nexus, a resource sharing platform for ML practitioners (further instructions will be provided at the kickoff event)

Please make sure to review the Schedule, Rules, and Code of Conduct before registering. Once you are ready to register, click the button below to purchase your ticket via Eventbrite.

Volunteer to Share ML Insights!

During each of the sprint events, participants are encouraged to share their machine learning insights and knowledge through short, low-stakes presentations lasting 10-30 minutes. These presentations can cover a specific tool, model, workflow, or any other ML resource that has proven useful. They can also focus on past experiences or learning moments. Any knowledge that may be valuable to others is welcome! If you have a topic in mind, please email ml-marathon-organizers@g-groups.wisc.edu with your idea and preferred presentation date(s).

Calling ML Experts! Join as Team Advisor

Are you already an ML expert and want to help give back to community? Join the ML Marathon as a team advisor! Advisors will be expected to attend at least half of the sprints to help guide their teams towards a working solution. In addition, advisors are strongly encouraged to present at least one demo/tutorial during the 12-week event. Registration is free for advisors given that they will be volunteering their time and effort to assist with the hackathon.

If you would like to join as an advisor, please email ml-marathon-organizers@g-groups.wisc.edu to tell us a little bit about yourself and your background in ML. If there are any specific challenges you prefer to advise, please also share this information. Finally, to help us plan catering, advisors should still register (select the free “Advisor” ticket) by the 9/10 deadline.