Abstract
One of the main goals of neuromorphic computing is the implementation and design of systems capable of dynamic evolution with respect to their own experience. In biology, synaptic scaling is the homeostatic mechanism which controls the frequency of neural spikes within stable boundaries for improved learning activity. To introduce such control mechanism in a hardware spiking neural network (SNN), we present here a novel artificial neuron based on phase change memory (PCM) devices capable of internal regulation via homeostatic and plastic phenomena. We experimentally show that this mechanism increases the robustness of the system thus optimizing the multi-pattern learning under spike-timing-dependent plasticity (STDP). It also improves the continual learning capability of hybrid supervised-unsupervised convolutional neural networks (CNNs), in terms of both resilience and accuracy. Furthermore, the use of neurons capable of self-regulating their fire responsivity as a function of the PCM internal state enables the design of dynamic networks. In this scenario, we propose to use the PCM-based neurons to design bio-inspired recurrent networks for autonomous decision making in navigation tasks. The agent relies on neuronal spike-frequency adaptation (SFA) to explore the environment via penalties and rewards. Finally, we show that the conductance drift of the PCM devices, contrarily to the applications in neural network accelerators, can improve the overall energy efficiency of neuromorphic computing by implementing bio-plausible active forgetting.
Highlights
The field of artificial intelligence (AI) has recently seen significant breakthroughs in the research, showing high performance in several tasks such as image recognition, natural language processing and playing games (Collobert et al, 2011; Krizhevsky et al, 2012; Mikolov et al, 2012; Silver et al, 2016)
We propose the use of phase change memory (PCM)-based homeostatic neurons for achieving continual learning in standard convolutional neural network
In this work we introduced a novel artificial neuron based on phase change memory (PCM) devices capable of internal regulation via homeostatic and plastic procedures
Summary
The field of artificial intelligence (AI) has recently seen significant breakthroughs in the research, showing high performance in several tasks such as image recognition, natural language processing and playing games (Collobert et al, 2011; Krizhevsky et al, 2012; Mikolov et al, 2012; Silver et al, 2016). The combination of the benefits introduced by homeostatic mechanism and reinforcement learning would improve the artificial intelligence systems toward the ability to autonomously interact with the environment in real life situations In this framework, several neuromorphic spiking neural networks (SNNs) based on CMOS technology have been proposed, demonstrating VLSI synaptic circuits with homeostatic neurons (Bartolozzi and Indiveri, 2006; Chicca et al, 2014; Qiao et al, 2017) and reward-based decision-making circuits (Wunderlich et al, 2019; Yan et al, 2019). The gradual crystallization of a PCM device enables the continual tuning of the internal threshold of the neuron as a function of the level of firing excitation This adaptation process improves the learning capability and directly translates in hardware the homeostatic control mechanism that manages the synaptic weight update during STDP. This work highlights the importance of PCM devices as key elements to achieve adaptation, learning and autonomous navigation exploiting the benefits of local edge computing
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
