AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Safety Verification of Decision-Tree Policies in Continuous Time

Christian Schilling, Anna Lukina, Emir Demirović, Kim Guldstrand Larsen

NeurIPS 2023 · Learning theory and algorithms › Prediction and decision losses · approach: Optimal decision tree learning · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
verify decision-tree policy directly instead of a neural-network policyset-based reachability · Quadrotorverification time: 10 and 59 minutes -> 6.5 minutesImproved
exploit decision-tree structure instead of encoding as a general hybrid automatonOurs vs JuliaReach · Cart-poleanalysis completion: 22 secondsImproved
add fixpoint check after bisectionAlgorithm 1 · Cart-poleunbounded-time safetyImproved
ignore complement branch and skip bisection when leaves share an actionAlgorithm 3 · hybrid-automaton comparisonprecision and efficiencyImproved
subdivide initial set into smaller subsetsreachability analysis · Quadrotorapproximation errorImproved
handle state transformation before the tree (acrobot)Algorithm 3 · Acrobotverification time: 100 -> 112No clear effect
increase tree size from depth 2 to depth 3 policydecision-tree policy · Acrobotverification time: 101 sec -> 113 secHurt
use DTCS larger tree (177 nodes) vs small treesdecision-tree policy · Quadrotor vs othersverification time: 84 secNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy