Automatic perturbation analysis for scalable certified robustness and beyond
Linear relaxation based perturbation analysis (LiRPA) for neural networks, which computes
provable linear bounds of output neurons given a certain amount of input perturbation, has …
provable linear bounds of output neurons given a certain amount of input perturbation, has …
The convex relaxation barrier, revisited: Tightened single-neuron relaxations for neural network verification
C Tjandraatmadja, R Anderson… - Advances in …, 2020 - proceedings.neurips.cc
We improve the effectiveness of propagation-and linear-optimization-based neural network
verification algorithms with a new tightened convex relaxation for ReLU neurons. Unlike …
verification algorithms with a new tightened convex relaxation for ReLU neurons. Unlike …
Function-dependent neural-network-driven state feedback control and self-verification stability for discrete-time nonlinear system
Deep learning significantly impacts neural network controller synthesis. Despite the higher
efficiency of deep learning algorithms compared to traditional model-based controller design …
efficiency of deep learning algorithms compared to traditional model-based controller design …
AI-Facilitated Dynamic Threshold-Tuning for a Maritime Domain Awareness Module
S Chan - 2024 IEEE International Conference on Industry 4.0 …, 2024 - ieeexplore.ieee.org
This paper presents a Decision Support Engine (DSE), which utilizes a unique Lower
Ambiguity, Higher Uncertainty (LAHU) and Higher Ambiguity, Lower Uncertainty (HALU) …
Ambiguity, Higher Uncertainty (LAHU) and Higher Ambiguity, Lower Uncertainty (HALU) …