Dat Duong Email & Phone Number

My work focuses on understanding the interaction between human clinicians and deep learning image classifiers and image (and text) generators on different medical datasets such as skin disorders, eye scans, and face images.

UCLA

Increased recall rates for rare and unobserved labels in a multi-label classification problem by modeling the vector representations for all the labels based on their text descriptions and hierarchy.Built a deep learning model based on the Transformer neural network to predict gene ontology labels for proteins based on their amino acid sequences and the.

Greater Seattle Area

Implemented multilevel regression models to determine factors associated with drastic changes in the time series of the inventory level.

Greater Seattle Area

Detected outliers in the time series of out-of-stock values for many product lines and the types of events that trigger these outliers; for example, level shifts in the series indicate business decisions that set the expected values to new standard.

Waltham, MA

Implemented special requests and executed R code for discrete choice model that computes the probability of a person choosing an item from a set. Each item is given an utility value, and the model estimates the probability that the utility value of one item is better than that of another.

Berkeley

Estimated the number of individuals answering a certain combination of questions in the 2008 American Community Survey-Public Use Microdata. The output was used to construct a map displaying the percentage of people within a community who are benefited by Obama or Romney in the 2012 election.

UCLA

Built model to find which variants in the genome that are associated with a specific disease. My method included prior knowledge about the variants like their distances to binding sites of transcription factors.

Doctor Of Philosophy (Ph.D.), Computer Science

Master Of Science (M.S.), Statistics

B.S, Applied Mathematics

B.S, Molecular, Cell & Developmental Biology