The Laboratory for Interactive Human-Computer Analytics works in a variety of application domains, including medical informatics, journalism, data and bioengineering.

LIHCA: A Software Platform for Human-in-the-Loop Analytics Systems

This software platform makes it easy to build interactive machine learning tools.  You can follow the simple instructions to install the open-source server-side software and then you are ready to connect from your JavaScript front-end with our API.  LIHCA provides a JavaScript object that lets you run machine learning algorithms on the server using python, with several scikitlearn implementations wrapped plus extras like metric learning.  The library also makes dealing with sessions, data and logging user interactions and model updates simple.

Check it out at platform.lihca.io .

Journalism in the Age of Wikileaks

Working with MuckRock.com we are tackling the problem of developing stories from massive piles of text.  Using interactive machine learning we help journalists and other users to discover threads of their stories in document collections.

Transcriptomics Workstation

IRIS – Integrated Radiological Image Search

ModelSpace – Visualizing the Trails of Data Models in Visual Analytics
Systems

Taking advantage of how human-in-the-loop analytics systems create machine learning models from user inputs, we use vector representations of these models to chart each user’s path through the space of models.  Read about it in ModelSpace at MLUIVA.

Related Previous Work on Interactive Machine Learning

Dis-function: Learning distance functions interactively. ET Brown, J Liu, CE Brodley, R Chang. Visual Analytics Science and Technology (VAST), 2012 IEEE Conference on, 83-92 138 2012

Human-Machine-Learner Interaction: The Best of Both Worlds. ET Brown, R Chang, A Endert. Workshop on Human-Centered Machine Learning at ACM CHI 2016.

Finding waldo: Learning about users from their interactions. ET Brown, A Ottley, H Zhao, Q Lin, R Souvenir, A Endert, R Chang. IEEE Transactions on visualization and computer graphics 20 (12), 1663-1672 60 2014.

Visualization by demonstration: An interaction paradigm for visual data exploration. B Saket, H Kim, ET Brown, A Endert. IEEE transactions on visualization and computer graphics 23 (1), 331-340 24 2017.

Podium: Ranking data using mixed-initiative visual analytics. E Wall, S Das, R Chawla, B Kalidindi, ET Brown, A Endert. IEEE transactions on visualization and computer graphics 24 (1), 288-297 16 2018.

Dynamic difficulty using brain metrics of workload. D Afergan, EM Peck, ET Solovey, A Jenkins, SW Hincks, ET Brown, R Chang, R JK Jacob. Proceedings of the 32nd annual ACM conference on Human factors in computing … 84 2014.

Bitnami