A Brain-Inspired Homeostatic Neuron Based on Phase-Change Memories for Efficient Neuromorphic Computing

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.

Nonvolatile Memristive Materials and Physical Modeling for In‐Memory and In‐Sensor Computing

“…These neurons, when integrated with plastic synapses, have exhibited sophisticated computational tasks. [400,401] Moreover, GST is an important material system for examining the intrinsic stochastic character of phase-change neurons, which is regulated by crystallization processes and may be altered by varying the chemical bond strength and amorphous topology. By doping with chemically strong but geometrically mismatched dopants, e.g., carbon or nitrogen, GST discloses enhanced stochastic character.…”

A survey and perspective on neuromorphic continual learning systems

“…The setup uses a custom circuit that incorporates an OxRAM crossbar array for emulating the neuron circuit. Muñoz-Martin et al ( 2021 ) exploited the conductance drift of the phase change memory (PCM) device for active forgetting. The authors devise a neuron with internal homeostatic and plastic regulation, inherently achieving stability-plasticity balance.…”

From Oxides to 2D Materials: Advancing Memristor Technologies for Energy-Efficient Neuromorphic Computing

“…Compared with other unconventional computing schemes, neuromorphic computing has garnered significant attention owing to its unprecedented energy efficiency, scalability, and potential to replace traditional von Neumann computing paradigms. , Neuromorphic computing is inspired by the biological concepts of human brain processing, demonstrating remarkable energy efficiency through an innovative chip architecture based on spiking neurons, synapses, and neural networks. − In addition, neuromorphic computing offers several advantages in terms of bandwidth and power consumption in electronic systems. − By contrast, hardware realizations of other unconventional computing approaches, such as spintronics and quantum computing, have faced challenges related to fragility, decoherence, and technical difficulties in maintaining qubits for sustained periods. , Although these approaches are promising, they have not yet demonstrated the same levels of energy efficiency and scalability as neuromorphic computing. , …”