Human-ML Decision Making

In this project, students developed interpretable machine learning tools that can help with human tasks spanning everything from decision making to art. They studied multiple tasks in interpretable machine learning using methods that can incorporate domain-based constraints and other types of domain knowledge using efficient discrete optimization techniques and Bayesian hierarchical modeling. Students created open source software derived from theoretical insight into specific domain problems.

The tasks for this summer included: 

  1. Create computer generated poetry: incorporate semantic meaning into computer generated poetry.
  2. De-noise low-resolution images: how to upsample a low resolution image while maintaining a high signal to noise ratio.
  3. Provide interpretable image recognition: use neural nets to dissect images into prototypical parts for interpretable image classification, for applications in medicine and beyond.

Participating Students

McCourt Hu

McCourt Hu
I am from Dalian China and majoring in CS and statistics. I am working with prof Cynthia Rudin on super resolution this summer and hope to enjoy my last year at Duke and go to graduate school after graduation.


Sachit Menon

Sachi Menon
Hi! My name is Sachit; I’m a student at Duke University studying math and computer science as an A.B. Scholar. My primary interests lie in deep learning and AI more generally (both fundamentals and applications, in that order). I’m particularly interested in unsupervised learning, especially representation learning and generative modeling. This summer, I'll be working on a project on image super-resolution, or the conversion of images from low resolutions to higher resolutions, via deep learning -- think 'zoom and enhance' from CSI, but real! I’m especially excited about this project as it presents an area where unsupervised methods have a clear, direct use case. In the future, I hope to pursue a PhD in machine learning, ideally centered around advancing unsupervised/semi-supervised methods and their application to tasks in supervised and reinforcement learning.


Bhrij Patel

Bhrij Patel
My name is Bhrij Patel, a rising sophomore. I am from Cary, North Carolina and intending to double major in Math and Computer Science. I am interested in my project for machine learning because I am interested in learning the various applications of it with different models. After Duke, I hope to go to some form of graduate school and go further into the field of data analytics.


Poster

The undergraduate summer program concluded August 2, 2019 with a research showcase and projects fair which featured posters demonstrating what our students have been doing all summer.

ML Decision Making Project Poster at Summer 2019 Research ShowcaseSummer Research Showcase: Human-ML Decision Making Project, with participating student.

Human-ML Decision Making Project Poster Picture2-Summer 2019 Research Showcase
Summer Research Showcase: Human-ML Decision Making Project, with participating students and guests.