Novel Nano-Machine Architecture for Machine Learning in the IoBNT
In this work, we propose a novel nano-scale architecture that performs matrix multiplications. Matrix multiplications are the basic operations of machine learning (ML) algorithms and, thus, the presented approach enables their application at the nano-scale, for example inside the human body in the Internet of Bio-Nano-Things (IoBNT). It is based on the molecule exchange between connected compartments and introducing chemical reactions in some of them. The matrix entries are solely defined by the volumes of the compartment. We provide a detailed mathematical description of the stochastic and dynamic behavior of the system. Moreover, we derive design guidelines for the proposed architecture. Finally, we validated the proposed approach through particle-based simulations.
Silicon Austria Labs
Email Address of Submitting AuthorStefan.email@example.com
ORCID of Submitting Author0009-0008-5759-8660
Submitting Author's InstitutionInstitute for Communications Engineering and RF Systems, JKU LIT SAL eSPML Lab, Johannes Kepler University (JKU)
Submitting Author's Country