Neuromorphic learning, working memory, and metaplasticity in nanowire networks

Abstract

Nanowire networks (NWNs) mimic the brain’s neurosynaptic connectivity and emergent dynamics. Consequently, NWNs may also emulate the synaptic processes that enable higher-order cognitive functions such as learning and memory. A quintessential cognitive task used to measure human working memory is the n -back task. In this study, task variations inspired by the n -back task are implemented in a NWN device, and external feedback is applied to emulate brain-like supervised and reinforcement learning. NWNs are found to retain information in working memory to at least n = 7 steps back, remarkably similar to the originally proposed “seven plus or minus two” rule for human subjects. Simulations elucidate how synapse-like NWN junction plasticity depends on previous synaptic modifications, analogous to “synaptic metaplasticity” in the brain, and how memory is consolidated via strengthening and pruning of synaptic conductance pathways.

Self-Assembled Monolayer and Nanoparticles Coenhanced Fragmented Silver Nanowire Network Memristor

“…Among various NW materials, silver nanowires (AgNWs) have been extensively used in flexible, stretchable, and on-skin electronics due to their merits of easy synthesis, excellent flexibility, simple processing, and tunable conductivity. , Because of the self-organized deposition behavior and good electrochemical activity of silver, the research on AgNW network-based memristors has noticeably grown recently. − In general, memristors are characterized by sandwich structures in metal–insulator–metal (MIM) configurations to achieve resistance modulation. Therefore, Ag core-insulative shell NWs are generally used to fabricate random AgNW networks with high-density MIM junctions.…”

“…Furthermore, AgNW networks with neuromorphic-like connectivity exhibit synapse plasticity-like properties such as long-term potentiation (LTP) and structural plasticity. ,− The memory states can be encoded in “winner-takes-all” conductive pathways, which can be revisited and reshaped using spatiotemporal signals . These characteristics enable higher-order brain functions like memory and learning …”

“…20 These characteristics enable higher-order brain functions like memory and learning. 17 Despite considerable progress, very limited studies have demonstrated the integration of artificial neurons (TS behavior) and synapses (synaptic plasticity) in a single AgNW network-based device, which is essential for achieving neuromorphic electronics with high-level cognitive functions. 4 It is reported that both the volatile (for TS) and nonvolatile (for synaptic plasticity) switching in AgNW network memristors arises from the formation and fracture of the Ag conductive filaments (CFs) at the cross-point junctions throughout the network.…”

Tomography of memory engrams in self-organizing nanowire connectomes

“…[16][17][18][19][20] With the aim of mimicking biological neuronal circuits where the principle of selforganization regulates both structure and functions, hardware architectures based on self-organized memristive networks of nano objects have attracted a growing attention. [21][22][23][24][25][26][27][28][29][30][31][32] The emerging spatiotemporal dynamics of these arti cial connectomes, where an emerging behavior arises from complexity similarly to what happens in our brain, make these complex networks versatile physical substrates for hardware implementation of brain-inspired computing paradigms. [33][34][35][36][37][38][39] Despite on one hand devices based on designless nanonetworks have been demonstrated as platforms for hardware implementation of advanced synaptic functions and unconventional computing paradigms, the potentiality of these devices by exploiting their functional connectivity still have to be explored.…”

“…Indeed, self-organizing memristive networks require a radical change of paradigm in implementing neuromorphic functionalities to take full advantage of their intrinsic multiterminal capability beyond the concept of two-terminal devices, posing at the same time new challenges for characterizing and exploiting their emergent behavior that require a shift in thinking and designing neuromorphic circuits. Despite visualization of internal dynamics in NW networks have been proposed through simulations, 22,29,32,40 the crucial coexistence of short-term and long-term plasticity (alternatively called weight and wiring plasticity) on the same physical substrate was just postulated or proved at single unit level (nanowire and nanowire junction) 21 , while static indirect representation of the formation of conductive paths have been analyzed at the nano/microscale from scanning electron microscopy 30 or through thermographic images 28 .…”