搜索

x
中国物理学会期刊

氧化物基忆阻型神经突触器件

Oxide-based memristive neuromorphic synaptic devices

CSTR: 32037.14.aps.68.20191262
PDF
HTML
导出引用
  • 忆阻器具有高密度、低功耗和阻值能够连续可调的特性, 被认为是模拟神经突触最具潜力的候选者. 而金属氧化物, 因其氧离子可迁移, 组分易于调控, 与传统CMOS兼容等优点, 是发展高性能忆阻器件的理想材料. 本文首先介绍了氧化物基忆阻器件阻变行为及其运行机制, 包括数字型和模拟型忆阻器. 主要综述了基于模拟型忆阻器实现的突触器件认知功能模拟, 包括非线性传输特性、时域突触可塑性、经验式学习和联合式学习等. 然后进一步介绍了忆阻型突触器件在模式识别、声音定位、柔性可穿戴和光电神经突触方面的潜在应用. 最后总结展望氧化物基忆阻神经突触在相关领域的可能发展方向.

     

    Memristors are considered to be the potential candidate for simulating synapses due to their high density, low power consumption and continuously adjustable resistance. Metal oxide is an ideal choice for fabricating memristive devices with high performance due to its advantages of oxygen migration, easy adjustment of components and compatibility with traditional CMOS. In this review paper, the memristive behaviors and operation mechanism of oxide-based memristors including digital-type memristors and analog-type memristors are first introduced. We mainly summarize the cognitive functions simulated by analog-type memristive synapse, including nonlinear-transmission characteristic, synaptic plasticity, learning experience, and non-associative/associative learning. Then, the potential applications of memristive synapse in pattern recognition, sound localization, logic operation, flexibility/transferability and optoelectronic memristive synapse are introduced. Finally, we provide an outlook of the future possible studies of oxide-based memristive synapse in the relevant fields.

     

    目录

    /

    返回文章
    返回
    Baidu
    map