Tiotto Thomas F, Goossens Anouk S, Borst Jelmer P, Banerjee Tamalika, Taatgen Niels A
Groningen Cognitive Systems and Materials Center, University of Groningen, Groningen, Netherlands.
Artificial Intelligence, Bernoulli Institute, University of Groningen, Groningen, Netherlands.
Front Neurosci. 2021 Feb 19;14:627276. doi: 10.3389/fnins.2020.627276. eCollection 2020.
Memristors have attracted interest as neuromorphic computation elements because they show promise in enabling efficient hardware implementations of artificial neurons and synapses. We performed measurements on interface-type memristors to validate their use in neuromorphic hardware. Specifically, we utilized Nb-doped SrTiO memristors as synapses in a simulated neural network by arranging them into differential synaptic pairs, with the weight of the connection given by the difference in normalized conductance values between the two paired memristors. This network learned to represent functions through a training process based on a novel supervised learning algorithm, during which discrete voltage pulses were applied to one of the two memristors in each pair. To simulate the fact that both the initial state of the physical memristive devices and the impact of each voltage pulse are unknown we injected noise into the simulation. Nevertheless, discrete updates based on local knowledge were shown to result in robust learning performance. Using this class of memristive devices as the synaptic weight element in a spiking neural network yields, to our knowledge, one of the first models of this kind, capable of learning to be a universal function approximator, and strongly suggests the suitability of these memristors for usage in future computing platforms.
忆阻器作为神经形态计算元件引起了人们的兴趣，因为它们有望实现人工神经元和突触的高效硬件实现。我们对界面型忆阻器进行了测量，以验证它们在神经形态硬件中的应用。具体而言，我们通过将掺铌的SrTiO忆阻器排列成差分突触对，将其用作模拟神经网络中的突触，连接权重由两个配对忆阻器之间归一化电导值的差异给出。该网络通过基于一种新型监督学习算法的训练过程学习表示函数，在此过程中，离散电压脉冲被施加到每对中的两个忆阻器之一。为了模拟物理忆阻器件的初始状态和每个电压脉冲的影响都是未知的这一事实，我们在模拟中注入了噪声。然而，基于局部知识的离散更新被证明能产生稳健的学习性能。据我们所知，在脉冲神经网络中使用这类忆阻器件作为突触权重元件，产生了首批此类模型之一，能够学习成为通用函数逼近器，并强烈表明这些忆阻器适用于未来的计算平台。
