Discord

Checklist AICC 2026 Round 6 · Task 2

Massive Problem

Classify single cells into 12 cell types from the expression levels of 1434 genes, training on nine patients and predicting for a tenth.

  • Tabular
  • Multiclass classification (gene-expression data with patient batches)

The task

Framed as a story set in the 'AICC hospital', the task asks for a classifier that assigns cells to one of 12 types based on the expression levels of 1434 genes.

The training data come from 9 separate patients (one file per patient), each row holding Gene1–Gene1434, the batch (patient) and the label (cell type 1–12). The test set comes from a separate tenth patient and has the gene features and batch only.

The dataset is very large, and the statement warns that memory must be managed carefully.

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
RNA_seq_patient_0 … RNA_seq_patient_8 (training; features Gene1–Gene1434, batch, label) and test.csv (Gene1–Gene1434, batch).
You submit
CSV with columns id (row index in test.csv) and label (predicted cell type).
Scoring
Macro-averaged F1. The baseline scores 0.2413 (rounded to 4 decimal places).
Rules
  • No architectural restrictions apart from the standard no-LLM rule.
  • The task and its data may not be republished, in whole or in part, without permission from the authors.
  • Individual participation (maximum team size 1); at most 5 submissions per day.
  • AICC contest rules (stated on each Kaggle rules page, not enforceable): no use of LLMs for writing code or getting task ideas; no internet use other than official library documentation and the contest platform; no communication with anyone during the contest; clarifications only via the #clarification-requests channel on the AICC Discord server.
Format
AICC Round 6, online on Kaggle, 25 Apr 2026 02:33 UTC – 27 Apr 2026 08:00 UTC.

Details

Year
2026, Online (Kaggle)
Round
Round 6 · Task 2
Language
English
License
Varies by task: Classifier Classifier — CC BY-NC-SA 4.0; Massive Problem — 'Subject to Competition Rules' (republishing the task or its data without the authors' permission is forbidden); Nuclei Reconstruction — CC BY-NC 4.0. Solutions repository: MIT., as stated by the source