The following list contains topics we intend to discuss in our seminars:
- Language Models as Knowledge Bases
- Pointer Networks
- Bidirectional LSTM-CRF models for sequence tagging
- Sequence level training with recurrent neural networks: this paper introduces the idea of using reinforcement learning to optimize non-differentiable metrics such as BLEU.
- Neural Architecture Search with Reinforcement Learning
- Linguistic Knowledge and Transferability of Contextual Representations