We are standing at the threshold of the information age, the financial services industry becomes the perfect playing field for machine learning technologies. We dedicate to develop applications of deep learning in finance, study the causal relationship between the data, and make progress on related computer science theory.
The project adopts deep learning models for financial data visualization. In financial trading, techniques like pattern recognition could advance the efficiency of strategic decisions. The professional traders could identify specific patterns among the candlestick charts with their long years of experience. At the same time, a convolutional neural network (CNN) in deep learning extracts spatial features from data. Compared to the model which designed to work on time series problem, the training process of the CNN model tends to be more similar to how human learns to study financial data.
- Encoding candlesticks as images for patterns classification using convolutional neural networks [Arxiv]
- Chen, J., Tsai, Y. Encoding candlesticks as images for pattern classification using convolutional neural networks. Financ Innov 6, 26 (2020). https://doi.org/10.1186/s40854-020-00187-0
- Explainable Deep Convolutional Candlestick Learner [Arxiv]
- Accepted by The 32nd International Conference on Software Engineering & Knowledge Engineering (SEKE 2020), KSIR Virtual Conference Cener, Pittsburgh, USA, July 9--July 19, 2020.
- Data Augmentation For Deep Candlestick Learner [Arxiv]
- Adversarial Robustness of Deep Convolutional Candlestick Learner [Arxiv]
- Funder & Principal Investigator (PI): Yun-Cheng Tsai Google Scholar
- Consultant: Samuel Yen-Chi Chen Google Scholar
- Researcher: Jun-Hao Chen
- Researcher: Chia-Ying Tsao
- Researcher: Chih-Shiang Shur