A distributed cellular approach of large scale SOM models for hardware implementation

Laurent Rodriguez (1) , Lyes Khacef (1) , Benoît Miramond (1)

1 LEAT - Laboratoire d'Electronique, Antennes et Télécommunications

Self-organization is a powerful bio-inspired feature that has been poorly investigated in the context of hardware computing architectures. The neural foundations of brain selforganization enable structural plasticity, continuous learning,and tolerance to lesions but at the cost of massive lateral synaptic connections. A cellular formulation of the related neural models would be able to tackle this limitation by iterating the propagation of the information in the network. This paper introduces the SOMA hardware architecture, which relies on this principle to develop a Self-Organizing Machine Architecture. We show the benefits of our cellular implementation compared to the existing neural models. This contribution leads to a hardware compliant algorithm thanks to decentralized operations and communications. A generalization of the technique is proposed to implement different Kohonen-like self-organizing models. This particular cellular implementation reaches the same behavior as the centralized models but drastically reduces their computing complexity. We present the performance results and discuss the benefits of cellular computing approaches in the perspective of designing hardware neural processors for embedded applications

Mots clés: particular cellular implementation; centralized models; computing complexity; cellular computing approaches; hardware neural processors; distributed cellular approach; neural models; large scale SOM models; hardware compliant algorithm; self-organizing machine architecture; Neurons; Computer architecture; Self-organizing feature maps; Hardware; Computational modeling; Synapses; Biological neural networks; self-organizing maps; cellular computing; neuromorphic architectures; hardware neural networks; cellular formulation; related neural models; SOMA hardware architecture; decentralized operations; self-organizing models; computational complexity; learning (artificial intelligence); neural net architecture; self-organising feature maps; hardware implementation; powerful bio-inspired feature; hardware computing architectures; neural foundations; brain self-organization; structural plasticity; continuous learning; massive lateral synaptic connections

Domaines: Architectures Matérielles [cs.AR]

Laurent Rodriguez, Lyes Khacef, Benoît Miramond. A distributed cellular approach of large scale SOM models for hardware implementation. IEEE Conference on Image Processing and Signals (IPAS), Dec 2018, Nice (sophia-antipolis), France. pp.250-255, ⟨10.1109/IPAS.2018.8708904⟩. ⟨hal-01923134⟩