量子侧抑制强化竞争算法研究  

Research on quantum lateral inhibition reinforcement competitive algorithm

作　　者：侯旋[1] 

机构地区：[1]空军工程大学航空航天工程学院,陕西西安710038

出　　处：《电子设计工程》2015年第6期37-40,共4页Electronic Design Engineering

摘　　要：深入研究了量子竞争网络的原理,基于量子竞争算法,通过引入神经元侧抑制功能与强化学习机制,建立了量子侧抑制强化竞争网络模型,提出了量子侧抑制强化竞争算法。量子侧抑制强化竞争网络竞争层神经元所具有的侧抑制功能提高了网络精度,其强化学习机制加速了网络学习。通过对具有四类样本的数据集进行模式识别实验,量子侧抑制强化竞争算法的平均正确率为98.89%,量子竞争算法的平均正确率为95.56%。结果表明量子侧抑制强化竞争算法相对量子竞争算法具有更高的模式识别能力。The principle of Quantum Competitive Neural Network（QCNN） has been intensively studied. Basing on Quantum Competitive Algorithm（QCA）, it establishes the model of Quantum Lateral Inhibition Reinforcement Competitive Neural Network（QLIRCNN） by introducing lateral inhibition function and reinforcement learning mechanism of neuron. Quantum Lateral Inhibition Reinforcement Competitive Algorithm（QLIRCA） is proposed. Lateral inhibition function improves the network accuracy and reinforcement learning mechanism accelerates network learning in competitive layer neurons of QLIRCNN. In pattern recognition experiment with four samples data set, average accuracy rate of QLIRCA is 98.89% and QCA is 95.56%. Experiment results show that QLIRCA has higher pattern recognition capabilities to QCA.

关 键 词：量子神经网络 侧抑制联结 强化学习机制 量子竞争 模式识别 

分 类 号：TN919.5[电子电信—通信与信息系统]