Abzu’s QLattice, A New Explainable AI
summary
Abzu releases a remote QLattice that a Feyn Python client queries to search millions of symbolic equations; runs on a high‑performance cluster while raw data stays local (data privacy by design). On a 300‑patient heart‑failure set with 10 features it returns simple formulas — e.g. age + creatinine + inverse ejection‑fraction term — whose AUC improves over a logistic baseline and generalizes better than a complex gradient booster that overfits. Demo shows iterative function composition and explicit, human‑readable equations (symbolic model discovery).
AI-written summary · about
chapters (8)
- 0:00Intro
- 1:22QLattice: Secure, high-performing, easy-to-use simulator
- 4:28Heart failure case study
- 5:33Problem 1: How do we predict a fatal outcome?
- 6:11Problem 2: How do we explain the diagnosis?
- 7:05Demo
- 22:31Why is inverse important?
- 33:25Demo
related talks
20mWhat can AI do for me?: evaluating machine learning interpretations in cooperative play
15mIntroduction to Explainable AI
16mBuilding Trust through Explainable AI
33mExplainable AI in the LIME-light
44mAI in the Clinic: Prediction is Better Than Cure
46mCan you trust your AI?
47mML: Explainable Artificial Intelligence explained
15mAccelerQ: Accelerating Quantum Eigensolvers With Machine Learning on Quantum Simulators