Memristive Extreme Learning Machine: A Neuromorphic Implementation

Neuromorphic computation has been a hot research area over the past few years. Memristor, as one of the neuromorphic computation materials memorizes the conductance value and is able to adapt it according to changing voltages. This paper pioneers a neuromorphic computing paradigm implementation (through memristor) for Extreme Learning Machine (ELM), which is one of most popular machine learning methods. By simulating the biological synapses with memristors and combining the memory property of memristor with high-efficient processing ability in ELM, a three-layer ELM model for classification is constructed. We represent the ELM network weights through memristive conductance values. The conductance values (network weights) are updated through tuning the voltages. Experimental results over the Iris dataset show that the memristor-based ELM achieves the same level performance as the one implemented via traditional software, and exhibits great potential that ELM can be implemented in neromorphic computation paradigms.