Energy-Efficient Threshold Circuits for Biological Information Processing, and its Limitations

我们将阈值电路视为神经网络的理论模型，构造具有低触发活性的阈值电路，以计算生物信息处理的简化任务。然后，我们给出了一个计算视觉信息处理的简化任务的阈值电路的显式构造。此外，我们还证明了存在显式任务，任何能效阈值电路都需要一定的深度或扇入来计算这些任务。

We consider a threshold circuit as a theoretical model of a neural network, andconstruct threshold circuits with low firing activity that computes a simplified task for biological information processing. Then we provide an explicit construction of a threshold circuit computing a simplified task for visual information processing. Moreover, we prove that there are explicit tasks which any energy-efficient threshold circuit needs a certain amount of depth or fan-in to compute.