L1 regularization is better than L2 for learning and predicting chaotic systems

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.