Jun Yuan 袁军Data Story Teller, passionate about AI Fairness and Human-Centered Decision-Making

I am a PhD candidate in Data Science, with a background in Mathematics and Statistics. My research interest lies in the intersection of human-centered AI and information visualization. Currently, I am working on designing visual analytic interfaces for ranking interpretation on high-stake decision-making (e.g., student course evaluation, AI hiring practice); adapting Learning-to-rank for fairness, accountability, and transparency; adversarial attacks and detection on explainable AI methods; information-seeking in crisis.

Projects & Publications

Ensuring transparency in the recruitment and talent acquisition process is essential. However, the current utilization of automated decision systems has proven to be disadvantageous for recruiters. This endeavor represents an initial effort to establish contextual transparency in online recruiting, with a specific emphasis on promoting candidate diversity and highlighting transferrable skills.

Ranking is a socio-technical artifact that humans interact with daily. This work is an effort to develop principled protocols to understand the existing public rankings (e.g., college rankings), domain-specific rankings (e.g., loan applicant rankings), and rankings that are machine-generated (e.g., social media feed rankings.). Given the spate of new legislations on algorithmic accountability, it is imperative that researchers from social science, human-computer interaction, and data science work in unison to demystify how rankings are produced, who has agency to change them, and what metrics of socio-technical impact one must use for informing the context of use.

Experience

Research Assistant (Thesis Advisor: Prof. Aritra Dasgupta)

NJIT, Department of Informatics · Newark, NJ
Jan 2019 - Present
  • Created a visual analytics workflow and interface to explain ranking models and assist stakeholders in gaining useful insights.
  • Formulated Learning-to-rank metrics for human-in-the-loop model training and evaluation.
  • Designed post-processing formula and sensitivity analysis to adopt Shapley Values for algorithmic rankers.
  • Developed subjectivity analysis workflow on COVID-related scientific communication.
  • Organized procedure for subjectivity annotation team.
  • Planned and took part in interviews and user studies within a multidisciplinary team.
  • Developed data mapping and augmentation rules for student analysis for the College of Computing's Dean's office.
  • Designed Tableau visualizations to aid administrators in strategic planning.

Data Science Intern

Accern · New York, NY
Jun 2021 - Aug 2021
  • Subjectivity modeling with BERT on COVID-19 tweets data.
  • Developed annotation guidelines and collaborated with a four people annotation team.
  • Generated 20 thousand annotation data for model training and achieved above 70% test accuracy.

Data Science Intern

Accern · New York, NY
Jun 2020 - Aug 2020
  • Topic modeling with BERT, T-SNE on finance news data.
  • Evaluated Machine Learning models via collaborative manual annotation.
  • Discovered temporal trend of topics during the COVID-19 breakout.

Education

PhD candidate, Data Science

New Jersey Institute of Technology

MSc, Mathematical Sciences

Clemson University

BSc, Mathematics

Hangzhou Dianzi University, China

Skills

Communication
  • Public Speaking
  • Tableau
  • LaTeX
  • Writing
  • Empathy
Analysis
  • Python
  • R
  • Matlab
  • SAS
  • DevOps

Volunteers

Graduate Student Association of NJIT · Vice President of Finance
Graduate Student Association of NJIT · Vice President of Programming

Teaching

Data Visualization · Teaching Assistant
MATH1010 · Instructor of Record
STAT8010 (SAS) · Lab Instructor

Mentoring

NSF-REU Summer Research Program (2019) · Research Mentor