New Materials for Printed Electronics-Organic Semiconductors and Dielectrics

使用微电子电路来模拟生物神经元的操作的大脑启发设备将成为未来计算系统的核心，因为它们可以提高计算速度并降低功耗。生物神经网络由不精确的异构计算单元组成，具有非常密集的连接，以缓慢（毫秒）的时间尺度处理时空模拟信号，但提供巨大的计算能力。目前的微电子制造已经优化，可以提供数十亿个以非常高的速度运行的同质单元，但这些单元不容易与许多神经形态设计兼容，或者可以在小型设备运行中进行定制。虽然类似大脑的机器可以强行适应数字技术，但另一种方法是使用印刷的电子神经元和突触。这些器件可以单独写入，并提供易于定制的软、柔性电路，成本很低。

Brain-inspired devices using microelectronic circuits to mimic the operation of biological neurons will be at the core of future computing systems as they can increase the speed of computation and reduce the power consumption. Biological neural networks are composed of imprecise, heterogeneous computational units, with very dense connectivity, that process spatiotemporal analogue signals at a slow (millisecond) timescale but deliver immense computational power. Current microelectronic fabrication has been optimised to provide billions of homogeneous units that operate at very high speeds but these are not easily compatible with many neuromorphic designs or can be customised in small device runs. While brain-like machines can be force-fit into digital technologies, an alternative approach is to use printed electronic neurons and synapses. These devices can be individually written and provide soft, flexible circuits that are readily customised at little cost.