Wei-Lun, LUO

Research Assistant  •  Research Center for Information Technology Innovation, Academia Sinica

九月 2020 - Present

Multiperiod Corporate Default Prediction

• Provide a consistent term structure of cumulative default probabilities through a carefully designed neural network.
• Tailor neural networks through economic domain knowledge to prevent our model from overfitting.
• Outperform the state-of-art statistical model on AR(10%) and RMSE (20%) for US public companies from 1990-2017.
• Part of this work was published at the 2022 SIAM International Conference on Data Mining (SDM).

Importance Sampling in Reinforcement Learning

• Implement Approximate Bayesian Computation(ABC) algorithm on Multi Armed Bandits (MAB) problems for faster computation.

Team leader, Recommendation Algorithms for KKStream

• Collaborate with team members and others from KKStream to construct a knowledge graph for items to improve performance.

Intern  •  E.SUN BANK

七月 2019 - 二月 2020

• Developed a trading platform with functions of optimization algorithms.
• Implemented reinforcement learning to solve backward stochastic differential equations.
• Designed an introduction lecture about reinforcement learning for a research group in the company (about 15 persons).
• Applied machine learning approaches to Forex forecasting.

Quantitative trader  •  Bincentive Inc, Taiwan

八月 2018 - 四月 2019

• Developed trading strategies on cryptocurrency markets.
• Crawled cryptocurrency markets data from different exchanges.
• Developed a part of trading API.