Towards human-centered artificial intelligence: learning from demonstration and meta-learning for model adaptation