Planning with tensor networks based on active inference DOI Creative Commons
Samuel T. Wauthier, Tim Verbelen, Bart Dhoedt

и другие.

Machine Learning Science and Technology, Год журнала: 2024, Номер 5(4), С. 045012 - 045012

Опубликована: Авг. 29, 2024

Abstract Tensor networks (TNs) have seen an increase in applications recent years. While they were originally developed to model many-body quantum systems, their usage has expanded into the field of machine learning. This work adds growing range by focusing on planning combining generative modeling capabilities matrix product states and action selection algorithm provided active inference. Their ability deal with curse dimensionality, represent probability distributions, dynamically discover hidden variables make specifically interesting choice use as inference, which relies ‘beliefs’ about within environment. We evaluate our method T-maze Frozen Lake environments, show that TN-based agent acts Bayes optimally expected under

Язык: Английский

A Survey on Cyber-Physical Security of Autonomous Vehicles Using a Context Awareness Method DOI Creative Commons
Aydin Zaboli, Junho Hong, Jaerock Kwon

и другие.

IEEE Access, Год журнала: 2023, Номер 11, С. 136706 - 136725

Опубликована: Янв. 1, 2023

Autonomous vehicles face challenges in ensuring cyber-physical security due to their reliance on image data from cameras processed by machine learning. These algorithms, however, are vulnerable anomalies the imagery, leading decreased recognition accuracy and presenting concerns. Current learning models struggle predict unexpected vehicular situations, particularly with unpredictable objects anomalies. To combat this, scholars focusing active inference, a method that can adapt based human cognition. This paper aims incorporate inference into autonomous vehicle systems. Multiple studies have delved this approach, showing its potential address gaps field. Specifically, these frameworks proven effective handling unforeseen

Язык: Английский

Процитировано

4
Dynamical Perception-Action Loop Formation with Developmental Embodiment for Hierarchical Active Inference DOI
Kanako Esaki,

Tadayuki Matsumura,

Shunsuke Minusa

и другие.

Communications in computer and information science, Год журнала: 2023, Номер unknown, С. 14 - 28

Опубликована: Ноя. 15, 2023

Язык: Английский

Процитировано

2
Planning with tensor networks based on active inference DOI Creative Commons
Samuel T. Wauthier, Tim Verbelen, Bart Dhoedt

и другие.

Machine Learning Science and Technology, Год журнала: 2024, Номер 5(4), С. 045012 - 045012

Опубликована: Авг. 29, 2024

Abstract Tensor networks (TNs) have seen an increase in applications recent years. While they were originally developed to model many-body quantum systems, their usage has expanded into the field of machine learning. This work adds growing range by focusing on planning combining generative modeling capabilities matrix product states and action selection algorithm provided active inference. Their ability deal with curse dimensionality, represent probability distributions, dynamically discover hidden variables make specifically interesting choice use as inference, which relies ‘beliefs’ about within environment. We evaluate our method T-maze Frozen Lake environments, show that TN-based agent acts Bayes optimally expected under

Язык: Английский

Процитировано

0