Advadnoun

Advadnoun

About

I go by Advadnoun on Twitter but my given name is Ryan Murdock.

Most of my recent work is in Machine Learning with a focus on text-to-image systems where I’ve done foundational research – and worked on large-scale text-to-image projects that you may have heard of.

Formerly affiliated with the University of Utah, my Patreon (https://www.patreon.com/patronizeme), & Adobe, I’m currently working as a researcher at Canva / Leonardo :)

I have also been in residency at Stochastic Labs & have done consulting work for Kaiber.

I’m especially interested in projects that:

  • Mitigate representational bias & other potential harms from generative models at the levels of data, training, inference, & UI,
  • Improve the expressivity & text alignment of text-to-image models,
  • Lower barriers to access to ML systems posed by data-efficiency issues and compute-requirements, and
  • Adapt ML models for creative uses in ways that are process-focused, weird, &/or focus on individual aesthetics.

I’m also interested in & have worked in the areas of Art, Psychology, Linguistics, Visual Search, & Materials Informatics.

For inquiries, contact rynmurdock(at)gmail[dot]com

Feel free to read a slice of my thoughts in the Blog Posts section above – it’s sparse right now, but follow with RSS or on Twitter if you’d like :) I’m also planning on updating my Portfolio for art eventually.

Selected links of Note

Also see my Google Scholar profile on the side panel for more of my research; the links below are largely unlisted there due to being on Twitter or in media.

BigSleep Citations

LatentVisions Acknowledgement

Seen in