Albert Ho

• Utilized clustering techniques such as UMAP and conducted differential expression analysis using *CAT data. Retrieved data revealed additional gene expression patterns that were not as prevalent in previous research.

• Analyzed scRNA-sequencing algorithms (*CAT) using stem cell data. Data analysis revealed the algorithm could be used to reverse-engineer data originally found through consensus non-negative matrix factorization.

• Processed large datasets and SEURAT objects with over 57,000 samples utilizing tools like Pandas and R.

• Enhanced diagnostic recall scoring for ventricular arrhythmias by developing a machine learning model. Compared algorithms like Logistic Regression, Random Forest, and CNNs, achieving highest recall with XGBoost.

• Boosted recall scores from 80% to 99% by implementing a convolutional neural network for sequence-to-sequence prediction. Utilized hyper-parameterization to optimize model performance.

• Processed and labeled large datasets (over 16,000 samples) using Pandas dataframes for training and testing

• Developed an Unreal Engine 3D simulation application of an archaeological site in Central Valley, California

• Implemented .fbx scan models, realistic lighting textures, and object interactions utilizing C++/Unreal modules

• Presented finalized application on UC Merced’s Wide Area Visualization Environment