Multimodal Representation Learning for Place Recognition Using Deep Hebbian Predictive Coding.

Autor:	Pearson MJ; Bristol Robotics Laboratory, University of The West England Bristol, Bristol, United Kingdom., Dora S; Department of Computer Science, Loughborough University, Loughborough, United Kingdom.; Center for Mathematics and Informatics, Amsterdam, Netherlands., Struckmeier O; Intelligent Robotics Group, Aalto University, Helsinki, Finland., Knowles TC; Bristol Robotics Laboratory, University of The West England Bristol, Bristol, United Kingdom., Mitchinson B; Department of Computer Science, University of Sheffield, Sheffield, United Kingdom., Tiwari K; Intelligent Robotics Group, Aalto University, Helsinki, Finland., Kyrki V; Intelligent Robotics Group, Aalto University, Helsinki, Finland., Bohte S; Center for Mathematics and Informatics, Amsterdam, Netherlands.; Department of Cognitive and Systems Neuroscience, University of Amsterdam, Amsterdam, Netherlands., Pennartz CMA; Department of Cognitive and Systems Neuroscience, University of Amsterdam, Amsterdam, Netherlands.
Jazyk:	angličtina
Zdroj:	Frontiers in robotics and AI [Front Robot AI] 2021 Dec 13; Vol. 8, pp. 732023. Date of Electronic Publication: 2021 Dec 13 (Print Publication: 2021).
DOI:	10.3389/frobt.2021.732023
Abstrakt:	Recognising familiar places is a competence required in many engineering applications that interact with the real world such as robot navigation. Combining information from different sensory sources promotes robustness and accuracy of place recognition. However, mismatch in data registration, dimensionality, and timing between modalities remain challenging problems in multisensory place recognition. Spurious data generated by sensor drop-out in multisensory environments is particularly problematic and often resolved through adhoc and brittle solutions. An effective approach to these problems is demonstrated by animals as they gracefully move through the world. Therefore, we take a neuro-ethological approach by adopting self-supervised representation learning based on a neuroscientific model of visual cortex known as predictive coding. We demonstrate how this parsimonious network algorithm which is trained using a local learning rule can be extended to combine visual and tactile sensory cues from a biomimetic robot as it naturally explores a visually aliased environment. The place recognition performance obtained using joint latent representations generated by the network is significantly better than contemporary representation learning techniques. Further, we see evidence of improved robustness at place recognition in face of unimodal sensor drop-out. The proposed multimodal deep predictive coding algorithm presented is also linearly extensible to accommodate more than two sensory modalities, thereby providing an intriguing example of the value of neuro-biologically plausible representation learning for multimodal navigation. Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. (Copyright © 2021 Pearson, Dora, Struckmeier, Knowles, Mitchinson, Tiwari, Kyrki, Bohte and Pennartz.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu