Memristors with tunable resistance states are emerging building blocks of artificial neural networks. However, in situ learning on a large-scale multiple-layer memristor network has yet to be demonstrated because of challenges in device property engineering and circuit integration. Here we monolithically integrate hafnium oxide-based memristors with a foundry-made transistor array into a multiple-layer neural network. We experimentally demonstrate in situ learning capability and achieve competitive classification accuracy on a standard machine learning dataset, which further confirms that the training algorithm allows the network to adapt to hardware imperfections. Our simulation using the experimental parameters suggests that a larger network would further increase the classification accuracy. The memristor neural network is a promising hardware platform for artificial intelligence with high speed-energy efficiency.
Early processing of visual information takes place in the human retina. Mimicking neurobiological structures and functionalities of the retina provides a promising pathway to achieving vision sensor with highly efficient image processing. Here, we demonstrate a prototype vision sensor that operates via the gate-tunable positive and negative photoresponses of the van der Waals (vdW) vertical heterostructures. The sensor emulates not only the neurobiological functionalities of bipolar cells and photoreceptors but also the unique connectivity between bipolar cells and photoreceptors. By tuning gate voltage for each pixel, we achieve reconfigurable vision sensor for simultaneous image sensing and processing. Furthermore, our prototype vision sensor itself can be trained to classify the input images by updating the gate voltages applied individually to each pixel in the sensor. Our work indicates that vdW vertical heterostructures offer a promising platform for the development of neural network vision sensor.
Threshold switches with Ag or Cu active metal species are volatile memristors (also termed diffusive memristors) featuring spontaneous rupture of conduction channels. The temporal dynamics of the conductance evolution is closely related to the electrochemical and diffusive dynamics of the active metals which could be modulated by electric field strength, biasing duration, temperature, and so on. Microscopic pictures by electron microscopy and quantitative thermodynamics modeling are examined to give insights into the underlying physics of the switching. Depending on the time scale of the relaxation process, such devices find a variety of novel applications in electronics, ranging from selector devices for memories to synaptic devices for neuromorphic computing.is applied due to the formation of a conduction channel(s) with Ag or Cu atoms. Unlike the ECM cells, the resistance recovers back spontaneously upon cessation of the external bias, yielding a superior I-V nonlinearity [4][5][6][7][8][9][10][11] and unique temporal conductance evolution dynamics. [7,12,13] Such a relaxation process is due to the physical dissolution of the metallic conduction channel under driving forces such as minimization of interfacial energy. In case active metals are used as electrodes, these metals may be doped into the dielectrics eventually under the combined effect of electric fields, thermal diffusion, which may lead to a reduced threshold voltage for the subsequent switching, similar to the process called "electroforming." The unique delay and relaxation dynamics of Ag and Cu-based threshold switches make them suitable for innovative applications in circuits and systems.Threshold switches with Ag or Cu active metals are also termed as "diffusive memristors" [7] to emphasize the underlying nature of the diffusive dynamics of the metal species. Factors including bias amplitude, biasing duration, as well as ambient temperature have been observed to have an impact on such a process, showing a wide range of dynamical properties, which could be exploited as access devices for memories with fast transition (e.g., <100 ns) or synaptic emulators with a relatively slower evolution (e.g., >1 µs). We survey the recently developed material systems which have exhibited this kind of threshold switching. New evidences by electron microscopy and quantitatively thermodynamic modeling are examined to give insights into the underlying physics of the mechanisms. We also discuss applications enabled by the advent of such threshold switches. Temporal Response of the SwitchingThe dynamical response of threshold switching is a critical property for many applications but has been characterized to a lesser extent. The temporal responses could be probed by applying voltage pulses and measuring the resulting currents in the time domain. It is a general observation in both ECM and threshold switches that the conductance experienced a transition from insulating state to conducting state after a finite time duration (delay time) under the external bias, as ill...
Recent breakthroughs in recurrent deep neural networks with long short-term memory (LSTM) units has led to major advances in artificial intelligence. State-of-the-art LSTM models with significantly increased complexity and a large number of parameters, however, have a bottleneck in computing power resulting from limited memory capacity and data communication bandwidth. Here we demonstrate experimentally that LSTM can be implemented with a memristor crossbar, which has a small circuit footprint to store a large number of parameters and in-memory computing capability that circumvents the 'von Neumann bottleneck'. We illustrate the capability of our system by solving real-world problems in regression and classification, which shows that memristor LSTM is a promising low-power and low-latency hardware platform for edge inference.
Neuromorphic computing based on spikes offers great potential in highly efficient computing paradigms. Recently, several hardware implementations of spiking neural networks based on traditional complementary metal-oxide semiconductor technology or memristors have been developed. However, an interface (called an afferent nerve in biology) with the environment, which converts the analog signal from sensors into spikes in spiking neural networks, is yet to be demonstrated. Here we propose and experimentally demonstrate an artificial spiking afferent nerve based on highly reliable NbOx Mott memristors for the first time. The spiking frequency of the afferent nerve is proportional to the stimuli intensity before encountering noxiously high stimuli, and then starts to reduce the spiking frequency at an inflection point. Using this afferent nerve, we further build a power-free spiking mechanoreceptor system with a passive piezoelectric device as the tactile sensor. The experimental results indicate that our afferent nerve is promising for constructing self-aware neurorobotics in the future.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.