Yuchen Sun Email and Phone Number

Proposed a graph-to-sequence neural network to find the solution of the Boolean satisfiability problem.Proposed a strategy to mitigate the issue of assignment dependency in Boolean satisfiability problem.

Worked with Qingwei LinUsed several binary classification algorithms, SVM, Random Forest, XGBoost, to predict hardware failures on the cloud platform.Transformed binary classification into Learning to Rank problem since online hardware failure prediction is extreme imbalanced problem and applied LambdaRank and LambdaMART for solving the problem.Proposed a neural network architecture based on Wide & Deep Learning.Deployed a pipeline on the distributed platform Cosmos for feature extraction and training.Improved the pipelines of several projects on time and space complexity