Information-Theoretic Constraints for Continual Vision-Language-Action Alignment

@misc{zhao2026informationtheoretic,
  title = {Information-Theoretic Constraints for Continual Vision-Language-Action Alignment},
  author = {Libang Zhao and Qixin Zeng and Hongyin Zhang and Donglin Wang},
  year = {2026},
  eprint = {2603.13335},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2603.13335},
}