Bioinformatics is an interdisciplinary field of biology, computer science, and statistics. As biomedical data has been amassed, machine learning algorithms became important for novel biological knowledge discovery. We research on developing machine learning algorithms that can aid diagnosis, treatment, and prevention with patient data with various diseases.

Biomedical Entity Search

To have an advanced search tool that directly returns relevant biomedical entities such as genes, drugs, diseases, and mutations rather than a long list of articles is desirable. Our system, BEST, a biomedical entity search tool, processes free text queries and returns up-to-date results (entities) quickly and accurately. Neural network based named-entity recognition (NER) and normalization will be used to identify known entities and new entities.

Biomedical Image Processing

Deep learning has rapidly become a dominant methodology for analyzing medical images. In an era of big data, significant features hidden in the data can be automatically discovered without laborious feature engineering thanks to deep learning. We are interested in broad medical image types(neuro, retinal, pulmonary, digital pathology, breast, cardiac) and apply deep learning to image classification, detection, segmentation, and other tasks.

Machine Reading at Scale

Introduction of large scale machine comprehension datasets (SQuAD, CNN/Daily, etc) has produced powerful neural models that can answer questions given a short passage. However, understanding longer texts is still a challenging task as it requires more effective and scalable methodologies. We aim to tackle this problem using multi-passage QA datasets (TriviaQA, Quasar-T, etc) with more scalable techniques.

Social Network Data Mining

With a growing popularity of mobile-based social media platforms such as Twitter, Instagram, Facebook, and Foursquare, a large amount of data has been generated by users on these social networks. Based on machine learning approaches, we focus on understanding user behavior patterns based on large-scale social network data. We also attempt to solve real-world problems such as POI(Point of Interest) recommendation and POI data imputation.

Stock Market Data Mining

Stock data mining employs machine learning algorithms such as deep learning, reinforcement learning and human insight to gather and analyze extensive financial data. Our aim is to find viable signals with predictibility in future market trends and develop algorithms for efficient business trades.


FoodInformatics employs deep-learning and human insight to gather and analyze extensive food-related data. We analyze a large amount of human-generated food recipes and utilize human-curated food databases, which include chemical properties of food, to better understand the world of food. Related Topics - Food Ingredient Embedding, Drink Recipe Analysis, Food & Drink Pairing based on Artificial Intelligence, Personalized Food Recommendation System

Other Data