Ioannis Anagnostides, Gabriele Farina, Tuomas Sandholm: Near-Optimal Φ-Regret Learning in Extensive-Form Games. ICML 2023: 814-839