There were some tutorials I took two of them:
They basically apply the machine learning algorithms along with NLP techniques to cancer cure detection task. They borrow sequence tagging, dependency parsing, word embedding and apply to tackle their current research. They talked usual about graph LSTMs and biLSTMs. (The presenter)
They briefly introduce end to end conversation based personal assitant like Apple's siri, Amazon's alexa etc. In other words, this tutorial was a detailed overview on how to apply dialogue systems on top of deep learning. The emphasize was on RL and different structured LSTM like architectures. Their slides (at least take a look at references in btw slides 100 and 120) are quite explanatory, and the cited papers try to catch up with the current research on that area. Finally, emptathy oriented task, Fung et al., 2016, was interesting part for me (multimodal application)