Abstract
This study uses advanced modeling and simulation to explore the effects of external events such as radiation interactions on the synaptic devices in an electronic spiking neural network. Specifically, the networks are trained using the spike-timing-dependent plasticity (STDP) learning rule to recognize spatio-temporal patterns (STPs) representing 25 and 100-pixel characters. Memristive synapses based on a TiO2 non-linear drift model designed in Verilog-A are utilized, with STDP learning behavior achieved through bi-phasic pre- and post-synaptic action potentials. The models are modified to include experimentally observed state-altering and ionizing radiation effects on the device. It is found that radiation interactions tend to make the connection between afferents stronger by increasing the conductance of synapses overall, subsequently distorting the STDP learning curve. In the absence of consistent STPs, these effects accumulate over time and make the synaptic weight evolutions unstable. With STPs at lower flux intensities, the network can recover and relearn with constant training. However, higher flux can overwhelm the leaky integrate-and-fire post-synaptic neuron circuits and reduce stability of the network.
Highlights
The concepts, components, and terminologies regarding the neural networks and models used throughout the remainder of the paper are given
The electrical properties of a memristor depend on the physical mechanism or resistance switching, which could be due to metal ions forming a conductive bridge (CBRAM), movement of oxygen ions (RRAM), phase change of the active material (PCM) or self-directed channel (SDC) [16]
The second topology is represented by a spiking neural network in Fig. 1b is a single layer perceptron network with either 25 or 100 pre-synaptic afferents (N1 to N25 or 100), each connected to a single post-synaptic neuron (LIF post N) via single memristors (M1 to M 25 or 100)
Summary
Interactions between memristive synapses and radiation alters the spike-timing-dependent plasticity (STDP) learning mechanism. Without presentation of spatio-temporal patterns (training), radiation effects build up and destabilize the spiking neural network. Customized hardware implementations will enable these spiking neural networks (SNNs) to be highly efficient and incredibly robust and fault-tolerant As such, they may find numerous applications in harsh, radiation-filled environments such as space or at nuclear and military installations. This article aims to analyze the effect of radiation on the spatio-temporal pattern recognition (STPR) capability of SNNs with memristive synapses. It is likely possible for the network to overcome larger amounts of radiation exposure when undergoing continuous on-line training or periodic re-training
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have