International Machine Learning Competition

Solve machine learning problems across three rounds, from the basics to current research, and compete with students from all over the world.

190
Countries
$1,500
Prize money
3
Rounds
fig. 1 Support Vector Machine
maximum-margin classifier the evolution of machine learning
About

Understand AI.
Don't just use it.

IMLC is a competition about how machine learning really works. You do not need to be an expert to participate: the first round needs no more than curiosity, school math, and a drive to learn. Each round builds on the previous one, so digging deeper into the topics is rewarded! Starting from the fundamentals, you work your way up to cutting-edge machine learning research.

  • a.

    Learn by solving

    The problems are the lessons. Each round builds on the last and we guide you through the material to make you succeed in each round. You do not need a powerful computer or a paid course.

  • b.

    A real competition

    Each round decides who advances, and the final ranking earns gold, silver and bronze awards. Participants come from over 190 countries.

  • c.

    For your future

    What you learn here lays the groundwork for studying or working in machine learning, and for understanding the systems that will soon be part of everyday life.

Looks hard? Good. Solving problems like this is how you learn it. And every point counts.
Two attempts to train the same model, on training data and on new data.

The panels follow two training runs, A and B. On the left, both end up looking equally good. On the right, they do not. Which run would you trust with new data, and what happened to the other one?

Q4.A bank uses a model to spot fraudulent card payments. Only 1 in 200 payments is fraudulent, so a model that simply answers "not fraud" every single time is correct 99.5% of the time.

(a)Why is this model useless despite its high accuracy?

(b)Propose a way of scoring fraud models so that this trick cannot win.

"…storing the model's weights with 4 bits instead of 16 cuts memory use to a quarter, with almost no loss in accuracy…"

P2.From a recent research paper. What does "almost no loss" leave open, and what experiment would you run to check the claim?

Participate

The clock is already running.

Solutions for the qualification round are due by Friday, 6 November 2026.

00
Days
:
00
Hours
:
00
Min
:
00
Sec
Download Problems Submit Solutions

no GPUs required · open to students worldwide · financial aid available

The spectrum

From SVMs to state of the art.

Six areas, one spectrum: the fundamentals that never change and the research that changes every month. Every round draws on all of it.

T·01

Fundamentals

SVMs, logistic regression, decision trees, kernels: the classical core, derived by hand. This is the math everything else stands on.

svm · logistic regression · kernels
T·02

Optimization & Generalization

How machines actually learn: gradient descent, loss landscapes, regularization, and why models that merely memorize fail to generalize.

gradient descent · bias–variance · overfitting
T·03

Deep Learning

Backpropagation by hand, then the architectures that took over: CNNs, RNNs and the transformer, understood layer by layer.

backprop · cnn · transformers
T·04

Frontier Models

The current research edge: LLMs, fine-tuning, RAG, and the efficiency work, from quantization to distillation, that makes them deployable.

llms · rag · distillation
T·05

Applications

Where it all lands: vision, healthcare, recommender systems, science. The same methods, translated into systems people rely on.

vision · healthcare · recommenders
T·06

Trustworthy AI

Adversarial attacks, robustness, alignment and fairness: how models fail, how they get attacked, and what responsible deployment demands.

adversarial · alignment · fairness
Rounds

Three rounds over five months.

Each round raises the stakes: first the fundamentals, then a real research paper, then one final exam where everything comes together. See the full rounds & process →

R1
Qualification until 6 November 2026

Qualification Round

Start here. A set of questions on the core ideas behind machine learning, from how models make decisions to how they learn from data. Score 15/17/20 points by age group to move on to the next round.

online · solo/group work · at your own pace
R2
Pre-Final Friday, 18 December 2026

Pre-Final Round

You receive a recent research paper on a topic in machine learning. Study it, then sit a supervised 60-minute exam: questions about the paper, plus two calculation problems. Registration is 12 EUR, and financial aid is available for anyone who needs it.

research paper · supervised exam · three problems
R3
Final Sunday, 17 January 2027

Final Exam

One timed exam of 30 questions that brings everything together: the foundations, the modern models, and the judgement to reason about both. The top finishers earn awards. Results announced 23 January 2027. See all awards →

supervised exam · timed questions · multiple-choice/short answers
Teachers & Schools

Bring IMLC to your classroom.

Register your students, follow their progress, and put your school on the leaderboard. Partner schools get reduced registration for every participant.

Teacher Account

Manage and track your students through every round, from qualification to the final exam, in one dashboard.

Register as a teacher →

School Partnership

Partner schools receive reduced registration fees for all students and a certificate shipment to the school.

Become a partner school →

Classroom Materials

Posters, problem sheets and training material to introduce machine learning thinking to your students.

Get the materials ↓
Ambassadors

Represent the IMLC in your country.

Ambassadors bring the competition to their schools, universities and communities. You spread the word, mentor newcomers, and grow a local ML community. We support you with materials, a certificate and an international network.

  • a.

    Promote

    Share the competition at your school or university and on local channels. Every registration through you counts toward your ambassador level.

  • b.

    Support

    Help participants in your region with questions about rounds, registration and preparation.

  • c.

    Grow

    Advance through ambassador levels, earn certificates and references, and join an international network of ML enthusiasts.

Newsletter

Stay in the loop.

Round openings, deadlines and results: a handful of emails a year, nothing else.