Speaker "Joan Xiao" Details Back



Natural Language Processing Is Fun: Generate Lyrics in the Style of Your Favorite Artist


Deep Learning has shown amazing promise due to the rapid rise in computational power and big data, and Recurrent Neural Networks are popular models that have generated a significant amount of excitement across multiple domains, including NLP, Speech and Computer Vision. In particular, RNNs have been used as powerful language models, at the levels of both words and characters. In this talk, we explain how to create RNN based character level language model, and use it to generate lyrics in the style of a given artist. We start with a review of how RNNs work, explore a lyrics dataset, train a RNN based character level language model, and finally, generate lyrics in the style of your favorite artist. PyTorch code will be used during the presentation.


Joan Xiao is a Lead Machine Learning Scientist at Figure Eight, a human-in-the-loop machine learning and artificial intelligence company. In her role, she leads research innovation and applies novel technologies to a broad range of real word problems. Previously she led the data science team at H5, a leading data search and analytics service company in e-Discovery industry. Prior to that, she led a Big Data Analytics team at HP. Joan received her Ph.D in Mathematics and MS in Computer Science from University of Pennsylvania.