Yiyuan Cui

Experience

• 

  Sensagrate

  May 2021 - Sept 2021

  Machine Learning Engineer internship

  • Time series predictions on large datasets for any given date using Arima model and Fbprophet model with 85% accuracy• Constructed Recurrent Neural Networks (RNNs) to perform clustering and compared with t-SNE algorithm, resulting in a 20% accuracy increase• Deployed the time series prediction model to AzureML as a RESTFUL API, Created a Flask application that received input data from the user passed the data to the AzureML model for prediction, and returned the prediction result to the user• Mapped safety solutions for data and provided user recommendations using a collaborative filtering method achieving an MSE of 0.22 Show less

• 

  UC San Diego

  Sept 2021 - Mar 2024

  • Conducted research on non-intrusive breath detection, using a microphone of the subject/patient’s face and IMU to collect data. Used CNN to classify movements, coughing, and talking stages in real time and achieved 95% accuracy in breath detection and 92% accuracy in movement classification.• Performing Exploratory Data Analysis on subjects’ sleep data including in three phases with Tableau and evaluating subject responses to test phases• Using SQL to retrieve data from the UCSD Hillcrest hospital server. Data cleaning and data imputation using forward feeding and missingness flags. Predicting sepsis using GRUs with EMR data as shown in previous papers, and combining Physiology data with EMR data to provide a better prediction which was not done before. The improvement in using only EMR data is around 15%. Show less

  • Student Manager

    Mar 2022 - Mar 2024

  • Scientific Researcher

    Sept 2021 - Mar 2024