The ORA Project
Embedded Files
The ORA Project (short for Ontological Regression Algorithm) seeks to discover novel forms of machine learning using yet-unexplored concepts from information theory, category theory, graph theory, and cognitive science.
The primary goal of this project is to create a highly efficient ML architecture capable of finding patterns in any form of information with little to no manual tuning. A key use-case of this technology would be in brain-computer interfaces, which are complex, noise-susceptible systems that can vary drastically from person to person.
Current Progress and Past Work
The Evolving Micro-Forest
The Evolving Micro-Forest
The Evolving Micro-Forest was the Lab's first attempt at a completely self-tuning machine learning model, which is based upon the Random Forest technique.
The model creates a population of tiny random forest models (called micro-forests) which compete, "reproduce", and evolve over many generations, automatically tuning themselves to the input dataset.
Page updated
Google Sites
Report abuse