PI-H2T: Enhancing Long-Tailed Visual Recognition with Permutation-Invariant and Head-to-Tail Feature Fusion

[system]

---

This is a companion discussion topic for the original entry at https://arxiv.org/abs/2506.00625