Semiconductor Memory Devices for Hardware-Driven Neuromorphic Systems
Cho, Seongjae
Semiconductor Memory Devices for Hardware-Driven Neuromorphic Systems - Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute 2021 - 1 electronic resource (81 p.)
Open Access
This book aims to convey the most recent progress in hardware-driven neuromorphic systems based on semiconductor memory technologies. Machine learning systems and various types of artificial neural networks to realize the learning process have mainly focused on software technologies. Tremendous advances have been made, particularly in the area of data inference and recognition, in which humans have great superiority compared to conventional computers. In order to more effectively mimic our way of thinking in a further hardware sense, more synapse-like components in terms of integration density, completeness in realizing biological synaptic behaviors, and most importantly, energy-efficient operation capability, should be prepared. For higher resemblance with the biological nervous system, future developments ought to take power consumption into account and foster revolutions at the device level, which can be realized by memory technologies. This book consists of seven articles in which most recent research findings on neuromorphic systems are reported in the highlights of various memory devices and architectures. Synaptic devices and their behaviors, many-core neuromorphic platforms in close relation with memory, novel materials enabling the low-power synaptic operations based on memory devices are studied, along with evaluations and applications. Some of them can be practically realized due to high Si processing and structure compatibility with contemporary semiconductor memory technologies in production, which provides perspectives of neuromorphic chips for mass production.
Creative Commons
English
books978-3-0365-1733-9 9783036517346 9783036517339
10.3390/books978-3-0365-1733-9 doi
Technology: general issues
Energy industries & utilities
leaky integrate-and-fire neuron vanadium dioxide neural network pattern recognition a-IGZO memristor Schottky barrier tunneling non filamentary resistive switching gradual and abrupt modulation bimodal distribution of effective Schottky barrier height ionized oxygen vacancy energy consumption hardware-based neuromorphic system synaptic device Si processing compatibility TCAD device simulation benchmarking neuromorphic HW neuromorphic platform spiNNaker spinMPI MPI for neuromorphic HW Boyer-Moore DNA matching algorithm flexible electronics neuromorphic engineering organic field-effect transistors synaptic devices short-term plasticity neuromorphic system on-chip learning overlapping pattern issue spiking neural network 3-D neuromorphic system 3-D stacked synapse array charge-trap flash synapse
Semiconductor Memory Devices for Hardware-Driven Neuromorphic Systems - Basel, Switzerland MDPI - Multidisciplinary Digital Publishing Institute 2021 - 1 electronic resource (81 p.)
Open Access
This book aims to convey the most recent progress in hardware-driven neuromorphic systems based on semiconductor memory technologies. Machine learning systems and various types of artificial neural networks to realize the learning process have mainly focused on software technologies. Tremendous advances have been made, particularly in the area of data inference and recognition, in which humans have great superiority compared to conventional computers. In order to more effectively mimic our way of thinking in a further hardware sense, more synapse-like components in terms of integration density, completeness in realizing biological synaptic behaviors, and most importantly, energy-efficient operation capability, should be prepared. For higher resemblance with the biological nervous system, future developments ought to take power consumption into account and foster revolutions at the device level, which can be realized by memory technologies. This book consists of seven articles in which most recent research findings on neuromorphic systems are reported in the highlights of various memory devices and architectures. Synaptic devices and their behaviors, many-core neuromorphic platforms in close relation with memory, novel materials enabling the low-power synaptic operations based on memory devices are studied, along with evaluations and applications. Some of them can be practically realized due to high Si processing and structure compatibility with contemporary semiconductor memory technologies in production, which provides perspectives of neuromorphic chips for mass production.
Creative Commons
English
books978-3-0365-1733-9 9783036517346 9783036517339
10.3390/books978-3-0365-1733-9 doi
Technology: general issues
Energy industries & utilities
leaky integrate-and-fire neuron vanadium dioxide neural network pattern recognition a-IGZO memristor Schottky barrier tunneling non filamentary resistive switching gradual and abrupt modulation bimodal distribution of effective Schottky barrier height ionized oxygen vacancy energy consumption hardware-based neuromorphic system synaptic device Si processing compatibility TCAD device simulation benchmarking neuromorphic HW neuromorphic platform spiNNaker spinMPI MPI for neuromorphic HW Boyer-Moore DNA matching algorithm flexible electronics neuromorphic engineering organic field-effect transistors synaptic devices short-term plasticity neuromorphic system on-chip learning overlapping pattern issue spiking neural network 3-D neuromorphic system 3-D stacked synapse array charge-trap flash synapse