Traceoid is revisiting the theory that underlies machine learning. Our approach opens up the door towards scalable energy-based models, faster and cheaper training and inference and interpretability. If we succeed, our approach will mark a significa…

We are working on making energy-based models (EBMs) viable by revisiting some of the math that underlies machine learning. Ideal candidate would understand how one goes about writing high perf code as well as be able to explain the concept of a Hami…

We are working on making energy-based models (EBMs) viable by revisiting some of the math that underlies machine learning. Here is what Yann LeCun had to say about EBMs https://x.com/ylecun/status/1380066315600343042 If you have background in any o…

We're trying to apply the insights of category theory, dependent type theory, and functional programming to deep learning. How do we best equip neural nets with strong inductive biases from these fields to help them reason in a structured way? Our u…

We are building an AI system that can accurately represent knowledge and handle uncertainty, to enable the discovery of insights and solve problems based on explainable reasoning. We envision applications to automate analysis and speed up research i…

We are building a system which accurately represents knowledge and uncertainty, to enable the discovery of insights and transparent problem solving.