Antique Painting Authentication

Classify antique paintings as authentic or replica from five numeric features, using a training set where most labels are unknown.

Tabular

Binary classification (semi-supervised)

The task

A family friend who is an archaeologist and art critic wants an algorithm that tells authentic antique paintings from replicas. Expert authentication is costly, so only a small share of the paintings are labelled and the rest are unknown.

Each painting is described by five digital features, which the statement says show strong structural patterns. The training CSV adds a sixth column: 1 for authentic, -1 for replica and 0 for unknown. The validation and test sets have the same features without labels, with 500 samples each, like the training set.

You are expected to use all samples, labelled or not, to train a model and predict -1 or 1 for every painting in the hidden validation and test sets.

Abridged by SOTA from the official materials. The official statement has the exact rules, and it wins wherever this summary differs.

At a glance

You get

training_set.csv with 500 rows of five feature columns and a label column (1 authentic, -1 replica, 0 unknown); hidden validation and test CSVs of 500 rows each, without labels.

You submit

submission.ipynb producing submission.zip with submissionA.csv (validation) and submissionB.csv (test), one prediction (-1 or 1) per line and no header.

Scoring

Classification accuracy (proportion of correctly predicted samples). The Scientific Committee's highest score (0.98) and baseline score (0.46) on Leaderboard B were used for score normalisation.

Rules

Notebook must finish within 10 minutes on the test machine
At most 15 successful submissions; up to 2 can be selected and the higher Leaderboard B score counts
Only a notebook may be submitted (no self-made mounted datasets or .pth files)
Test machine has no internet access; pretrained models may be used only if they import offline

Format

Individual Contest, Day 2 (6 August 2025); six hours for the day's three tasks

Links

Antique Painting Authentication

The task

At a glance

Links

Notebook

Data

Solution

Results

Source files

Details