Crowdsourced Image Segmentation
HCOMP 2018, Demo Paper
Developed techniques for evaluating the quality of crowdsourced image segmentation by aggregating across multiple workers.
Doris Jung-Lin Lee
I am a PhD student at UC Berkeley working with Professor Aditya Parameswaran.
My current research interest lies in interactive data analytics, visualization, and human-computer interaction. I am currently working on tools that assist data scientists in the process of exploratory data analysis and machine learning.
Before this, I was at the computer science department at the University of Illinois, Urbana-Champaign. I received my Bachelor's degree in physics and astrophysics from UC Berkeley in 2016.
I am awarded the 2020 Facebook PhD Fellowship in Systems for Machine Learning.
News Updates
- 05/2020: Spoke about automation in data science for the Alter Everything Podcast.
- 04/2020: An extended abstract describing my thesis direction for the CHI Doctoral Consortium is published.
- 03/2020: Some preliminary visualizations for our new project on visualizing the impact of interventions for COVID-19.
- 02/2020: New blogpost reflecting on the state of vis recommendation systems.
- 01/2020: Presented our new system Lux at the RISE retreat in Monterey.
Publications
A Human-in-the-loop Perspective on AutoML: Milestones and the Road Ahead
[PDF]
IEEE Data Bulletin 2019 |
You can't always sketch what you want: Understanding Sensemaking in Visual Query SystemsIEEE Visual Analytics Science & Technology (TVCG Track at VAST'19 at VIS) |
Avoiding Drill-down Fallacies with VisPilot: Assisted Exploration of Data SubsetsACM Intelligent User Interface (IUI) 2019 |
The Case for a Visual Discovery Assistant: A Holistic Solution for Accelerating Visual Data Exploration
[PDF]
IEEE Data Bulletin 2018 |
Crowdclass: Designing classification-based citizen science learning modulesAAAI Human Computation and Crowdsourcing (HCOMP) 2016 |
Skintillates: Design and Prototype Epidermal Interactions
[Project Page] [PDF]
Honorable Mention AwardACM Designing Interactive Systems (DIS) 2016 |
‘I don't want to wear a screen’: Probing perceptions of and possibilities for dynamic displays on clothing
[Project Page] [PDF]
Best Paper AwardACM Computer-Human Interaction (CHI) 2016 |
Past Projects
Social Behavioural Data Mining
Fashion Account Discovery
KDD 2017, ML4Fashion Workshop
Classification and crawling Twitter data to identify fashion accounts in social networks.
Energy Usage Patterns
AAAI 2017, AI for Smart Grids & Buildings Workshop
Comparison of Clustering Techniques for Residential Energy Behavior using Smart Meter Data.
Fabrication tools & Wearables
Shrinky Circuits
Developing a rapid prototyping technique for building circuit boards without involving conventional PCB/chemical etching procedures.
Project Jacquard
Collaboration with Google Advanced Technologies and Projects (ATAP) group on a new type of interactive wearable technology.
Skintillates
Creating low-cost, accessible fabrication technique for on-skin wearable electronics that can integrate with a variety of electronic components.
Cosmological Data Mining
Pattern Discovery and Large-Scale Data Mining on Cosmological Datasets. [Poster]Workshop on Algorithms for Modern Massive Data Sets (MMDS 2016).
RC3 Mosaics
Designing a pipeline for generating scientifically calibrated images of large nearby RC3 galaxies using an adaptive algorithm for positional update.
Halo Finder
Applying unsupervised machine learning algorithms to cosmological simulations for finding dark matter haloes.
SDSS Imaging Systematics
Investigating how systematics affect the imaging data quality from the Sloan Digital Sky Survey.
Astrophysical Fluid Simulations
Star Formation
Magnetohydrodynamics, adaptive mesh refinement simulations for the evolution of a collapsing dense core.
Accretion Disk
Investigating the effects of Papaloizou-Pringle and magnetorotational instabilities in accretion disk torus.
astroSim-tutorial
Jupyter Notebook tutorials on how to build, run, and analyze astrophysical simulations.