SICIM Machine Learning Challenge
Ready for a new challenge?
SICIM is hosting an exciting Machine Learning Challenge — your opportunity to dive deep into AI, sharpen your skills, and see how you measure up against your peers.
What's the Challenge About?
You'll be provided with datasets containing characteristics of value documents. There are two datasets, each representing different types of documents with slightly varying feature vectors. Your task is to identify counterfeit documents using machine learning techniques. Two different tasks with different challenges await you.
Task A: Smart Classification
Train a model to classify documents as genuine or counterfeit. Performance evaluation of your model will consider both prediction performance measured as F0.5-score and model efficiency, number of adjusted parameters during training, which are necessarily to proceed the inference procedure.
Task B: Anomaly-detection
In this task, you'll only receive data from genuine documents — no examples of counterfeits are available for training. Despite that, your model should still be able to identify counterfeit documents when evaluated later. This task requires creative use of unsupervised or one-class learning learning techniques. Performance will again be measured by considering prediction performance in terms of F0.5-score .
How to participate?
Think you're up for the challenge? Compete solo or form a small team with fellow students or members at Hochschule Mittweida.
- Register by emailing: Marika Kaden Upon registration, you'll receive the datasets along with detailed evaluation criteria and submission guidelines.
- Submission format: Submit your solution via git@HSMW
- Programming Language:
- Python (prefered)
- MATLAB (also accepted)
- Your submission must include:
- The trained model
- A script explaining how to run your model
- The number of learned parameters, which are necessary for the inference together with a short model explanation
Key Details
- Where to register? Marika Kaden
- Who can participate? Anyone with an HSMW email address (individuals or small teams welcome)
- When is the deadline? August 15, 2025
- Where to submit? git.hs-mittweida.de
Challenge yourself. Learn something new. Show what your AI can do. We're looking forward to your participation!