From 6351232ed1e74519e765009fbe1914ff0b145050 Mon Sep 17 00:00:00 2001 From: Dane Sabo Date: Mon, 12 May 2025 11:32:08 -0400 Subject: [PATCH] vault backup: 2025-05-12 11:32:08 --- ...ation of neural network controlled autonomous systems.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/Literature Notes/Formal verification of neural network controlled autonomous systems.md b/Literature Notes/Formal verification of neural network controlled autonomous systems.md index dffd08d54..67cc9180d 100644 --- a/Literature Notes/Formal verification of neural network controlled autonomous systems.md +++ b/Literature Notes/Formal verification of neural network controlled autonomous systems.md @@ -79,13 +79,13 @@ Published: 2019-04 >[!example] >we utilize a Satisfiability Modulo Convex (SMC) encoding to enumerate all the possible assignments of different ReLUs. - >- [ ] #Follow-Up + >- [x] #Follow-Up ✅ 2025-05-12 >[!example] >Therefore, recent works focused en-tirely on verifying neural networks against simple input-output specifications [28–33]. Such input-output techniques compute aguaranteed range for the output of a deep neural network givena set of inputs represented as a convex polyhedron. - >- [ ] #Follow-Up + >- [x] #Follow-Up ✅ 2025-05-12 >[!example] >For example, by using binary variables to en-code piecewise linear functions, the constraints of ReLU functions are encoded as a Mixed-Integer Linear Programming (MILP). Com-bining output specifications that are expressed in terms of Linear Programming (LP), the verification problem eventually turns to aMILP feasibility problem [32, 34]. - >- [ ] #Follow-Up + >- [x] #Follow-Up ✅ 2025-05-12