Nature Inspired Computational Architectures

<strong>Seminar of the School of Physical Sciences -------------------------------------------------</strong> Title: <strong>Nature Inspired Computational Architectures</strong> Speaker: <strong>Saurabh Bose</strong> (University of Canterbury, Christchurch) Date: <strong>April 6, 2016 (Wednesday)</strong> <strong>Abstract: </strong>Man-made computers are based on circuits of functional units that follow given design rules. On the other hand, natural computers exploit the emergent properties and massive parallelism of interconnected networks of locally active components to perform highly complex computational tasks like navigation, recognition, and decision-making. In this talk, I will summarize my efforts to understand the dynamics of disordered networks of nanoscale components to realize computational architectures. In Bose et al., Nature Naotechnology report, we have shown that interconnected metal nanoparticles acting as strongly nonlinear single-electron transistors can be configured in situ into any Boolean logic gate. For this purpose, the electrical properties were artificially evolved, using genetic algorithm, to perform computational tasks reconfigurably. The system meets the criteria for the physical realization of (cellular) neural networks: universality, compactness, robustness and evolvability, which implies scalability to perform more advanced tasks. I will also discuss our current efforts to emulate biological neuronal structures with percolating nanoparticle network. These are similar to the brain-architecture and naturally incorporate the necessary complexity and criticality. These self-assembled switching devices are promising for brain-inspired neuromorphic architecture suitable for fault-tolerant computation and real-time on-chip pattern recognition.