Spatiotemporal Information Processing Emulated by Multiterminal Neuro‐Transistor Networks

Abstract

All external sensory stimuli produce a spatiotemporal pattern of action potentials, which is transmitted to the biological neural system to be processed. The relative timing of synaptic spikes from different presynaptic neurons represents the features of the stimuli. A fundamental prerequisite in cortical information processing is the discrimination of different spatiotemporal input sequences. Here, capacitively coupled multiterminal oxide‐based neuro‐transistors are proposed for spatiotemporal information processing, mimicking the dendritic discriminability of different spatiotemporal input sequences. The experimental results demonstrate that such multiterminal neuromorphic devices can act as spatiotemporal information processing compartments for fundamental cortical computation. Also, as an example of spatiotemporal information processing, sound location functionality of the human brain is also emulated by constructing a simple artificial neural network based on such oxide‐based multiterminal neuro‐transistors.

The Design of 3D‐Interface Architecture in an Ultralow‐Power, Electrospun Single‐Fiber Synaptic Transistor for Neuromorphic Computing

“…Synapses are the basic unit of a human brain for information processing 9. With the technology based on complementary metal‐oxide semiconductor (CMOS), at least 10 transistors are needed to implement one synapse, which is too large for an artificial neuromorphic system 10–12. Therefore, developing a single device to achieve synaptic behavior is an important improvement in the basic hardware for the artificial neuromorphic system.…”

Recent Advances in Transistor‐Based Artificial Synapses

“…In addition to CNTs, oxide semiconductors have also been used as channel materials in electrolyte‐gated synaptic transistors, where indium‐zinc oxide (IZO) and indium‐gallium‐zinc oxide (IGZO) are the most widely used oxide materials . High intrinsic carrier mobility, excellent optical transparency, decent flexibility, and good compatibility with large‐area processing technologies make these metal oxides promising candidates for artificial synaptic transistors.…”

“…The ability of synapse to strengthen or weaken its information transmission efficiency over time provides the physiological substrate for a variety of synaptic computing and learning . Although the operating speed of a single neuron or synapse is usually much lower than that of a single modern transistor, neural networks can concurrently perform learning and spatiotemporal information processing in analogy parallel, which makes the human brain more efficient at processing complex and unstructured information than von Neumann computers . The outstanding computational performance of the human brain inspired scientists around the world to develop brain‐like computers.…”

Vertical 0D‐Perovskite/2D‐MoS2 van der Waals Heterojunction Phototransistor for Emulating Photoelectric‐Synergistically Classical Pavlovian Conditioning and Neural Coding Dynamics

“…When an electrical pulse is applied to the coplanar gate, the positive mobile ions in the polymer gel electrolyte move laterally toward the MoS 2 channel interface, which induces a large number of free electrons in the MoS 2 channel and a significant EPSC. [2,67] When this electrical stimulus has stopped, the free electrons remaining in the channel rapidly recombine with the accumulated holes at the heterojunction interface, resulting in typical STM behavior. As a counterpart, the optical approach can flexibly control the photoelectric performance of bionic devices by changing the carrier transport characteristics at the heterojunction interfaces.…”