Suggested Projects

Module 1: Text Generation

The sequence-to-sequence model can not only be used for POS-tagging, but also for text generation where machines generate sentences and documents, such as Tweets, Hacker News, image captions, Yelp reviews. For an open-sourced collection of text generators based on neural networks, see TextGenRNN.

• Project 1 Product review generation.
  Li Dong, Shaohan Huang, Furu Wei, Mirella Lapata, Ming Zhou and Ke Xu. Learning to Generate Product Reviews from Attributes, EACL, 2017.
• Project 2 Multi-modal text generation: you will generate Reddit-like comments given a Reddit image.

Module 2: Relation Extraction

• Project 3
  Daojian Zeng, Kang Liu, Yubo Chen, and Jun Zhao. Distant Supervision for Relation Extraction via Piecewise Convolutional Neural Networks, EMNLP, 2015.
• Project 4
  Dongxu Zhang and Dong Wang. Relation Classification via Recurrent Neural Network, EMNLP, 2015.
• Project 5
  DSGAN: Generative Adversarial Training for Distant Supervision Relation Extraction, ACL, 2018.

Module 3: RNN

• Project 6 An early paper on RNN for text generation.
  Ilya Sutskever, James Martens, and Geoffrey Hinton. Generating Text with Recurrent Neural Networks, ICML, 2011.
• Project 7 Empirical evaluation of RNN.
  Rafal Jozefowicz, Wojciech Zaremba, and Ilya Sutskever. An Empirical Exploration of Recurrent Network Architectures, ICML, 2015.
• Project 8 The first paper using adversarial learning for language generation.
  Lantao Yu, Weinan Zhang, Jun Wang, and Yong Yu. SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient, AAAI, 2017.
• Project 9 Deal with the discrepancy between the generated texts during test time and the training texts.
  Samy Bengio, Oriol Vinyals, Navdeep Jaitly, and Noam Shazeer. Scheduled Sampling for Sequence Prediction with Recurrent Neural Networks, NeurIPS, 2015.

Module 4: Image Captioning

• Project 10 You will need to connect computer vision with NLP.
  Tseng-Hung Chen, Yuan-Hong Liao, Ching-Yao Chuang, Wan-Ting Hsu, Jianlong Fu, and Min Sun Show, Adapt and Tell: Adversarial Training of Cross-domain Image Captioner, ICCV, 2017.