-
Predicting Up and Down Probability of Equity Index by Technical Indicators: Demostrate the idea to using TI as input of predictive models
-
Anormaly email detection by unspuervised learning approach (with BERT): Demostrate to use BERT pre-trained model to replace traditional feature engineering and how to deal with zero fraud dataset
-
Speech emotion detection by TRILL pre-trained model: Demostrate the feasibility of using pre-trained model by Google to distill features of a speech to detect speaker emotion
-
Chineses Name Entity Recognition (NER) by SpaCy: Demostrate to use SpaCy for Name Entity recognition from Chinese passage