Neural microelectronics

Ohmi, Tadahiro

doi:10.1109/iedm.1997.650395

Cited by 17 publications

(9 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It has been known for some time that arrays of analog resistive memory elements are ideally suited for the multiply-accumulate operations at the heart of DNN forward-inference and training [66,67]. The multiply operation is per- Figure 2, NVM devices with asymmetric conductance responses (e.g.…”

Section: Vector-matrix Multiplication For Neuromorphic Applicationsmentioning

confidence: 99%

Neuromorphic computing using non-volatile memory

Burr

Shelby

Sebastian

et al. 2016

Advances in Physics: X

850

701

View full text Add to dashboard Cite

Dense crossbar arrays of non-volatile memory (NVM) devices represent one possible path for implementing massively-parallel and highly energy-efficient neuromorphic computing systems. We first review recent advances in the application of NVM devices to three computing paradigms: spiking neural networks (SNNs), deep neural networks (DNNs), and 'Memcomputing'. In SNNs, NVM synaptic connections are updated by a local learning rule such as spike-timing-dependent-plasticity, a computational approach directly inspired by biology. For DNNs, NVM arrays can represent matrices of synaptic weights, implementing the matrix-vector multiplication needed for algorithms such as backpropagation in an analog yet massively-parallel fashion. This approach could provide significant improvements in power and speed compared to GPU-based DNN training, for applications of commercial significance. We then survey recent research in which different types of NVM devices-including phase change memory, conductive-bridging RAM, filamentary and nonfilamentary RRAM, and other NVMs-have been proposed, either as a synapse or as a neuron, for use within a neuromorphic computing application. The relevant virtues and limitations of these devices are assessed, in terms of properties such as conductance dynamic range, (non)linearity and (a)symmetry of conductance response, retention, endurance, required switching power, and device variability.

show abstract

Section: Vector-matrix Multiplication For Neuromorphic Applicationsmentioning

confidence: 99%

Neuromorphic computing using non-volatile memory

Burr

Shelby

Sebastian

et al. 2016

Advances in Physics: X

850

701

View full text Add to dashboard Cite

show abstract

“…Therefore in the primary visual cortex where the information from the retina first enters, edges with various orientations are detected at every location of the image on the retina. The fact was discovered by the groundbreaking work of Hubel and Wiesel first in the cat (6) and then in the macaque monkey (7). Anyway, this is the associative principle working at the very primitive level of brain processing.…”

Section: Associative Principlementioning

confidence: 98%

(Invited) A Brain-Inspired VLSI Architecture for Nano Devices and Circuits

Shibata¹

2010

ECS Trans.

View full text Add to dashboard Cite

What kind of systems would be most appropriate to build with nanoelectronics technologies? A human-like intelligent system is proposed as one of the most plausible and promising candidates. In order to achieve human-like recognition performance, various types of sensing devices are required to gather information from environment, a large capacity of memories for learning from experience, and huge computational powers for recognition and understanding. Multifunctional device integration in nanometer scales would certainly provide a sound platform for such system implementation. However, regarding the computational powers, enhancing the integration density alone will not be a solution, because the computational principle in the brain is not yet known. A brain-inspired VLSI architecture based on the associative principle is presented in this paper, in which the non-linear I-V characteristics of nano functional devices are directly utilized as the very bases of computation. Due to the nature of mind processing, the hardware algorithms are tolerable to elemental-device level variability, and in this sense, the approach is most suited to building systems in nanoelectronics.

show abstract

“…Since the biological details about the higher cognitive functions of the brain are not yet known, we call it psychological inspiration, by which we mean the model has been built by observing our mental activities by ourselves. The model is based on the premise that human intelligence is produced by the vast amount of past memories accumulated through our life long experience, and that the most similar event to the current event is automatically retrieved from the past memories and is utilized to understand the situation and make a decision (13). The associative principle is not a mere pattern matching but the most fundamental principle, we believe, that governs the entire processing in the brain.…”

Section: Vlsi Brain Modelmentioning

confidence: 99%

Bio Inspired Architectures in the Nanoscale Integration Era

Shibata¹

2009

ECS Trans.

View full text Add to dashboard Cite

With the advent of the nanoscale integration era, we will be able to build super-high performance CPU chips with more than 10 10 transistors on single chips. The number is equivalent to the total number of neurons in the neocortex of the human brain. However, is it really possible to build systems as intelligent as humans? Digital computers today have enormous computational powers and will have more in the future. However, they are not very smart in such tasks like naïve perception, intuitive understanding, and flexible decision making. Such functionalities are in great demand to establish seamless interfacing between the human society and the cyber world. In this paper we will discuss what we can learn from biology of its computing principles and how we can implement them in the VLSI hardware by best utilizing the sate-ofthe-art ULSI integration technology.

show abstract

Neural microelectronics

Cited by 17 publications

References 36 publications

Neuromorphic computing using non-volatile memory

Neuromorphic computing using non-volatile memory

(Invited) A Brain-Inspired VLSI Architecture for Nano Devices and Circuits

Bio Inspired Architectures in the Nanoscale Integration Era

Contact Info

Product

Resources

About