Binary Classification¶

Convert a regression dataset into binary labels and train a classifier — for example, metal vs. non-metal based on a bandgap threshold.

Command¶

train_alignn.py \
  --root_dir "alignn/examples/sample_data" \
  --classification_threshold 0.01 \
  --config "alignn/examples/sample_data/config_example.json" \
  --output_dir=temp

Values in id_prop.csv are thresholded: 1 if above --classification_threshold, 0 otherwise.

Supported tasks¶

Binary only

The current training script supports binary classification only. For multi-class problems, open a GitHub issue — it is being considered.

Typical classification targets (JARVIS-DFT)¶

Classifier	Threshold
Metal / non-metal (OPT / MBJ bandgap)	0.01 eV
Magnetic / non-magnetic	0.05 µB
Stable / unstable (ehull)	0.1 eV
High / low SLME	10 %
High / low spillage	0.1
High / low Seebeck	±100 µV K⁻¹
High / low power factor	1000 µW (mK²)⁻¹

See the Performance page for accuracy numbers.

Metrics¶

The training loop reports AUC and accuracy on the validation set and writes prediction_results_test_set.csv with per-structure predicted probabilities.