Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yoshihara, Sota, Yamamoto, Ryosuke, Kusumoto, Hiroyuki, Shimura, Masanari
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2505.17696
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911091293945856
author Yoshihara, Sota
Yamamoto, Ryosuke
Kusumoto, Hiroyuki
Shimura, Masanari
author_facet Yoshihara, Sota
Yamamoto, Ryosuke
Kusumoto, Hiroyuki
Shimura, Masanari
contents This paper proposes a novel theoretical framework for guaranteeing and evaluating the resilience of long short-term memory (LSTM) networks in control systems. We introduce "recovery time" as a new metric of resilience in order to quantify the time required for an LSTM to return to its normal state after anomalous inputs. By mathematically refining incremental input-to-state stability ($δ$ISS) theory for LSTM, we derive a practical data-independent upper bound on recovery time. This upper bound gives us resilience-aware training. Experimental validation on simple models demonstrates the effectiveness of our resilience estimation and control methods, enhancing a foundation for rigorous quality assurance in safety-critical AI applications.
format Preprint
id arxiv_https___arxiv_org_abs_2505_17696
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Enhancing AI System Resiliency: Formulation and Guarantee for LSTM Resilience Based on Control Theory
Yoshihara, Sota
Yamamoto, Ryosuke
Kusumoto, Hiroyuki
Shimura, Masanari
Artificial Intelligence
Systems and Control
This paper proposes a novel theoretical framework for guaranteeing and evaluating the resilience of long short-term memory (LSTM) networks in control systems. We introduce "recovery time" as a new metric of resilience in order to quantify the time required for an LSTM to return to its normal state after anomalous inputs. By mathematically refining incremental input-to-state stability ($δ$ISS) theory for LSTM, we derive a practical data-independent upper bound on recovery time. This upper bound gives us resilience-aware training. Experimental validation on simple models demonstrates the effectiveness of our resilience estimation and control methods, enhancing a foundation for rigorous quality assurance in safety-critical AI applications.
title Enhancing AI System Resiliency: Formulation and Guarantee for LSTM Resilience Based on Control Theory
topic Artificial Intelligence
Systems and Control
url https://arxiv.org/abs/2505.17696