Research

DMIS Lab focuses on Biomedical AI, NLP, Graph ML and other topics. Below are some exemplary works produced by lab members. More papers related to each of these topics can be found in the publications section.

Biomedical AI

Keywords: Drug Discovery, Personalized Medicine, Systems Biology

Biomedical data and the field of bioinformatics holds much potential to benefit human life. Computers have been used in order to take advantage of this data gold mine. Despite advances in computer and information technology, tasks that use biomedical data, like drug prescription and discovery, are complicated and influenced by many factors not accountable by human beings. DMIS Lab aims to address this issue by developing full-scale machine learning algorithms that may further advance existing biomedical data mining and bioinformatics methods and techniques in order to facilitate this process. Such methods provide much practical value in real-world applications.

Dr. Sunkyu Kim's VAECox paper was accepted to Bioinformatics and presented at the 28th Conference on Intelligent Systems for Molecular Biology (ISMB 2020).
Dr. Minji Jeon's paper was accepted to Nature Communications 2019. This paper highlights the achievements of DMIS Lab in AstraZeneca's DREAM Challenge.
More news and papers can be found here.

Natural Language Processing (NLP)

For prospective students

News & Research

Quick links

Keywords: Biomedical NLP, Question Answering, Named Entity Recognition

Natural language processing (NLP) is the field of studying how computers may understand and utilize human language. DMIS Lab conducts research in various subfields of NLP including question answering (QA), named entity recognition (NER), neural machine translation (NMT), etc. This research effort also spills over to biomedical AI due to the fact that many bioinformatics tasks rely on information from various biomedical corpora. DMIS Lab is currently developing such tools including biomedical NER, biomedical QA, etc. Below are some examples from our recent work:

1 paper accepted to ACL 2022, and 1 paper accepted to NAACL 2022.
1 paper accepted to Bioinformatics 2022.
BioBERT (co-first authored by Dr. Jinhyuk Lee and Wonjin Lee; Bioinformatics 2020) has been ranked as the most read papers in Bioinformatics which is one of the top-tier journals in the domain (over 2,400 citations as of July, 2022). It also has been selected as one of the top 3 BioNLP papers of the year by IMIA (International Medical Informatics Association) year book.
More news and papers can be found here.

Graph ML & Machine Learning Applications (Graph & ML)

Keywords: Graph Representation Learning, Computer Vision, Recommender Systems

Machine learning methods that are based on graph structures have gained much popularity as they provide utility for data in the form of graphs. As objects like drugs or chemicals may be represented by graphs of various forms, DMIS Lab focuses on develop machine learning algorithms that utilize such graphical structures. In addition to graph-based models, our lab also conducts research on more fundamental aspects of the field of ML in order to fine tune the effectiveness of our current tools and methods and enhance our understanding of the algorithms we use.

Two papers have been accepted to ECCV 2020.
Graph Transformer Network (GTN) and Self-Attention Graph (SAG) Pooling have been accepted to NeurIPS 2019 and ICML 2019, respectively.
More news and papers can be found here.