Bio-inspired signal processor consisting of coupled quantum-dot devices

在这项工作中，我们提出了一个生物启发的信号处理设备，模仿反应扩散系统的动力学结合神经网络。反应扩散系统（RD系统）是化学反应和物质扩散共存于非平衡状态的化学复杂系统。它在自然界中产生各种动态现象。构建一个与神经网络相结合的反应扩散系统的电模拟将使我们能够在LSI芯片上生成人工生物动力学并开发生物启发的信息处理系统。我们提出构建一个RD系统的电模拟，即，由量子点电路组成的电RD系统。RD系统可以看作是一系列相互作用的化学非线性振子的集合，因此我们可以用电振子代替化学振子来构造电RD系统。我们使用的电振荡器，量子点电路产生的库仑阻塞现象引起的非线性振荡。RD系统中的扩散作用可以通过振荡器之间的电容耦合来模仿。通过将耦合振荡器排列成网络，我们设计了一个量子点RD系统。我们还提出了构建一个由量子点电路组成的电神经网络。电神经网络由阈值器件和加权耦合突触组成，加权耦合突触由单稳态单电子振荡器和耦合电容组成。我们通过计算机模拟表明，RD系统与神经网络相结合，产生电耗散结构，或动画的节点电位在电路中的时空模式，这是一个类似于化学RD系统的特性。

In this work, we proposed a bio-inspired signal processing device that imitated the dynamics of reaction-diffusion systems combined with neural networks. A reaction-diffusion system (RD system) is a chemical complex system in which chemical reactions and material diffusion coexist in a nonequilibrium state. It is producing various dynamic phenomena in the natural world. Constructing an electrical analog of reaction-diffusion systems combined with neural networks will enable us to generate artificial biodynamics on a LSI chip and develop bio-inspired information-processing systems.We proposed constructing an electrical analog of RD systems, i.e., an electrical RD system consisting of quantum-dot circuits. An RD system can be considered an aggregate of chemical nonlinear oscillators interacting with one another, so we can construct electrical RD systems by using electrical oscillators instead of chemical ones. We used, as the electrical oscillator, a quantum-dot circuit that produced nonlinear oscillation caused by the Coulomb blockade phenomenon. The action of diffusion in RD systems can be imitated by capacitive coupling between the oscillators. By arranging coupled oscillators into a network, we designed a quantum-dot RD system. We also proposed constructing an electrical neural network consisting of quantum-dot circuits. The electrical neural network is composed of threshold devices and weighted coupling synapses consisting of monostable single-electron oscillators and coupling capacitors. We showed through computer simulation that the RD system combined with neural networks produced electrical dissipative structures, or animated spatiotemporal patterns of node potential in the circuit, which was a characteristic similar to that in chemical RD systems.