Minus AI — Research Lab
A research lab for reinforcement learning and interactive visualization.
We study how agents learn from interaction — and build tools that make those dynamics legible.
News
Paper
September 2026
RLHF without human labels
at ECCV 2026 in Malmö
RLHF works, but human feedback is slow and expensive. Our framework lets models check themselves, redrawing the image a caption describes and measuring the drift: faithful image description for search and accessibility, on about 20% of the usual data.
Exhibition
July 2026
Exhibiting A Walled City
at SIGGRAPH 2026 in Los Angeles
Our interactive AI installation turns images uploaded by visitors into the architectural chambers of an evolving virtual city, modeled on Hong Kong’s Kowloon Walled City. One of 12 works selected from 211 submissions for the SIGGRAPH 2026 Art Gallery.
Exhibition
July 2026
Presenting ReCollection
at ZKM
Our interactive installation on how agents form, forget, and revisit memories is on view at ZKM Center for Art and Media, Karlsruhe from July 4, 2026 to February 21, 2027, turning personal memories into synthetic ones, assembled from the fragments of many people’s pasts.
Our Approach
Three threads of research, one feedback loop.
We study how agents learn, build tools to see what they learn, and publish so others can build on it.
Reinforcement Learning
Agents that learn from interaction — exploration, credit assignment, and stable policy optimization.
Interactive Visualization
Tools that make training dynamics, value landscapes, and agent behavior legible in real time.
Open Research
Methods, tooling, and findings shared openly, so the community can build on our work from day zero.
Join Us
Help us build learning agents.
We’re a small team of researchers and engineers who love hard problems at the intersection of reinforcement learning and visualization.