Artificial cognitive memory—changing from density driven to functionality driven

Shi, Luping; Yi, Kiyoon; Ramanathan, Kiruthika; Zhao, Rong; Ning, Ning; Ding, Dawei; Chong, T. C.

doi:10.1007/s00339-011-6297-0

Cited by 21 publications

(9 citation statements)

References 83 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This function has been experimentally observed in a wide variety of neural circuits in different species, ranging from locusts to humans, and is regarded as the basis for learning and memory181920. Among currently available devices,chalcogenide device is a competitive candidate for combining data storage and processing in a manner similar to neural cognitive processing2122. In addition to being used as a non-volatile memory because of the reversible switching between chalcogenide's amorphous and crystalline states232425, chalcogenide device is also capable of acting as an arithmetic processor26, as a result of the natural energy accumulation property.…”

mentioning

confidence: 99%

Ultrafast Synaptic Events in a Chalcogenide Memristor

Zhong

et al. 2013

Sci Rep

363

240

View full text Add to dashboard Cite

Compact and power-efficient plastic electronic synapses are of fundamental importance to overcoming the bottlenecks of developing a neuromorphic chip. Memristor is a strong contender among the various electronic synapses in existence today. However, the speeds of synaptic events are relatively slow in most attempts at emulating synapses due to the material-related mechanism. Here we revealed the intrinsic memristance of stoichiometric crystalline Ge2Sb2Te5 that originates from the charge trapping and releasing by the defects. The device resistance states, representing synaptic weights, were precisely modulated by 30 ns potentiating/depressing electrical pulses. We demonstrated four spike-timing-dependent plasticity (STDP) forms by applying programmed pre- and postsynaptic spiking pulse pairs in different time windows ranging from 50 ms down to 500 ns, the latter of which is 105 times faster than the speed of STDP in human brain. This study provides new opportunities for building ultrafast neuromorphic computing systems and surpassing Von Neumann architecture.

show abstract

mentioning

confidence: 99%

Ultrafast Synaptic Events in a Chalcogenide Memristor

Zhong

et al. 2013

Sci Rep

363

240

View full text Add to dashboard Cite

show abstract

“…Thus it would be interesting to investigate how working memory is presented in the polychronous network, and simulate a bioplausible working memory system with increased memory capacity. To this end, we are working towards the development of artificial cognitive memory with the objective of developing a novel function-driven memory technology in comparison to conventional density-driven storage technology [32]. The models of persistent firing neuron and spiking network presented in this paper can be used in the simulation of cognitive memory architectures.…”

Section: Discussionmentioning

confidence: 99%

Modeling Neuromorphic Persistent Firing Networks

Ning¹,

Li²,

He³

et al. 2015

IJIS

View full text Add to dashboard Cite

Neurons are believed to be the brain computational engines of the brain. A recent discovery in neurophysiology reveals that interneurons can slowly integrate spiking, share the output across a coupled network of axons and respond with persistent firing even in the absence of input to the soma or dendrites, which has not been understood and could be very important for exploring the mechanism of human cognition. The conventional models are incapable of simulating the important newly-discovered phenomenon of persistent firing induced by axonal slow integration. In this paper, we propose a computationally efficient model of neurons through modeling the axon as a slow leaky integrator, which captures almost all-known neural behaviors. The model controls the switching of axonal firing dynamics between passive conduction mode and persistent firing mode. The interplay between the axonal integrated potential and its multiple thresholds in axon precisely determines the persistent firing dynamics of neurons. We also present a persistent firing polychronous spiking network which exhibits asynchronous dynamics indicating that this computationally efficient model is not only bio-plausible, but also suitable for large scale spiking network simulations. The implications of this network and the analog circuit design for exploring the relationship between working memory and persistent firing enable developing a spiking networkbased memory and bio-inspired computer systems.

show abstract

“…The ability of a neural network to learn depends on the number of degrees of freedom available to the network (for an electronic device it means the number of switchable states). The number of degrees of freedom determines the plasticity of the system, i.e., its capability of approximating the training set (plasticity scales with the number of degrees of freedom) [1][2][3][4][5][6][7][8][9] .…”

Section: Neuron Activation Functionsmentioning

confidence: 99%

Multistability in Bistable Ferroelectric Materials toward Adaptive Applications

Ghosh

Koster

Rijnders

2016

Adv Funct Materials

View full text Add to dashboard Cite

Traditionally thermodynamically bistable ferroic materials are used for nonvolatile operations based on logic gates (e.g., in the form of field effect transistors). But, this inherent bistability in these class of materials limits their applicability for adaptive operations. Emulating biological synapses in real materials necessitates gradual tuning of resistance in a nonvolatile manner. Even though in recent years few observations have been made of adaptive devices using a ferroelectric, the principal question as to how to make a ferroelectric adaptive has remained elusive in the literature. Here, it is shown that by locally controlling the nucleation energy distribution at the ferroelectric–electrode interface multiple‐addressable states in a ferroelectric can be created, which is necessary for adaptive/synaptic applications. This is realized by depositing a layer of nonswitchable ZnO on top of thin film ferroelectric PbZr x Ti(1– x )O3. This methodology of interface‐engineered ferroelectric should enable realising brain‐like adaptive/synaptic memory in complementary metal‐oxide‐semiconductor (CMOS) devices. Furthermore, the temporally stable multistability in ferroelectrics should enable the designing of multistate memory and logic devices.

show abstract

Artificial cognitive memory—changing from density driven to functionality driven

Cited by 21 publications

References 83 publications

Ultrafast Synaptic Events in a Chalcogenide Memristor

Ultrafast Synaptic Events in a Chalcogenide Memristor

Modeling Neuromorphic Persistent Firing Networks

Multistability in Bistable Ferroelectric Materials toward Adaptive Applications

Contact Info

Product

Resources

About