混合量子衍生神經(jīng)網(wǎng)絡模型及算法
doi: 10.11999/JEIT150444
基金項目:
國家自然科學基金(61170132), 黑龍江省自然科學基金(F2015021), 黑龍江省教育廳科學技術研究項目(12541059)
Hybrid Quantum-inspired Neural Networks Model and Algorithm
Funds:
The National Natural Science Foundation of China (61170132), The Natural Science Foundation of Heilongjiang Province, China (F2015021), The Scientific and Technological Research Project of the Education Department of Heilongjiang Province, China (12541059)
-
摘要: 為提高人工神經(jīng)網(wǎng)絡的逼近能力，該文從研究隱層神經(jīng)元的映射機制入手，提出基于量子比特在Bloch球面的繞軸旋轉(zhuǎn)構(gòu)造神經(jīng)網(wǎng)絡模型的新思想。首先將樣本線性變換為量子比特的相位，并使量子比特在Bloch球面上分別繞著3個坐標軸旋轉(zhuǎn)，旋轉(zhuǎn)角度即為網(wǎng)絡參數(shù)。然后通過投影測量可以得到量子比特的球面坐標，將這些坐標值提交到隱層激勵函數(shù)，可得隱層神經(jīng)元的輸出。輸出層采用普通神經(jīng)元?；贚-M(Levenberg-Marquardt)算法設計了該模型的學習算法。實驗結(jié)果表明，該文提出的模型在逼近能力、泛化能力、魯棒性能方面，均優(yōu)于采用L-M算法的普通神經(jīng)網(wǎng)絡。
-
關鍵詞:
- 量子計算 /
- 量子比特旋轉(zhuǎn) /
- 量子衍生神經(jīng)元 /
- 量子衍生神經(jīng)網(wǎng)絡
Abstract: To enhance the mapping ability of artificial neural networks, by studying the mapping mechanism of hidden layer neurons, a new idea of designing neural networks model based on rotation of qubits in the Bloch sphere is proposed in this paper. In the proposed approach, the samples are linearly transformed to quantum bit phase, and the qubits are rotated about three coordinate axes, respectively. The network parameters of hidden layer are the rotation angles. The spherical coordinates of qubits can be obtained by the projection measurement. The output of hidden layer neurons can be concluded by submitting these coordinates to excitation functions in hidden layer. The general neurons are applied to the output layer. The learning algorithms of the proposed model are designed based on the Levenberg-Marquardt (L-M) algorithm. The experimental results show that the proposed model is superior to the classical (L-M) algorthm in approximation ability, generalization ability, and robust performance. -
KAK S. On quantum neural computing[J]. Information Sciences, 1995, 83(3): 143-160. GOPATHY P and NICOLAOS B. Quantum Neural Networks (QNNs) inherently fuzzy feed-forward neural networks[J]. IEEE Transactions on Neural Networks, 1997, 8(3): 679-693. VENTURA D and TONY M. Quantum associative memory with exponential capacity[C]. Proceedings of the IEEE International Joint Conference on Computational Intelligence, Piscataway, NJ, 1998: 509-513. AJIT N and TAMMY M. Quantum artificial neural network architectures and components[J]. Information Sciences, 2000, 128(3): 231-255. 解光軍, 莊鎮(zhèn)泉. 量子神經(jīng)網(wǎng)絡[J]. 計算機科學, 2001, 28(7): 1-6. XIE G J and ZHUANG Z Q. Quantum neural network[J]. Computer Science, 2001, 28(7): 1-6. 解光軍, 范海秋, 操禮程. 一種量子神經(jīng)計算網(wǎng)絡模型[J]. 復旦學報(自然科學版), 2004, 43(5): 700-703. XIE G J, FAN H Q, and CAO L C. A quantum neural computational network model[J]. Journal of Fudan University (Natural Science), 2004, 43(5): 700-703. 解光軍, 周典, 范海秋. 基于量子門組單元的神經(jīng)網(wǎng)絡及其應用[J]. 系統(tǒng)工程理論與實踐, 2005, 25(5): 113-117. XIE G J, ZHOU D, and FAN H Q. A neural network model based on quantum gates cell and its applications[J]. Systems Engineering Theory Practice, 2005, 25(5): 113-117. MAEDA M, SUENAGA M, and MIYAJIMA H. Qubit neuron according to quantum circuit for XOR problem[J]. Applied Mathematics and Computation, 2007, 185(2): 1015-1025. LI P C and LI S Y. Learning algorithm and application of quantum BP neural networks based on universal quantum gates[J]. Journal of Systems Engineering and Electronics, 2008, 19(1): 167-174. 李盼池. 一種量子神經(jīng)網(wǎng)絡模型學習算法及應用[J]. 控制理論與應用, 2009, 26(5): 531-534. LI P C. Learning algorithm and applications of the quantum neural networks model[J]. Control Theory Application, 2009, 26(5): 531-534. LI P C, SONG K P, and YANG E L. Model and algorithm of neural networks with quantum gated nodes[J]. Neural Network World, 2010, 20(2): 189-206. LI P C and XIAO H. A hybrid quantum-inspired neural networks with sequence inputs[J]. Neurocomputing, 2013, 117: 81-90. LI P C and XIAO H. Model and algorithm of quantum- inspired neural network with sequence input based on controlled rotation gates[J]. Application Intelligence, 2014, 40(1): 107-126. 李盼池, 周紅巖. 基于受控Hadamard門的量子神經(jīng)網(wǎng)絡模型及算法[J]. 計算機研究與發(fā)展, 2015, 52(1): 211-220. LI P C and ZHOU H Y. Model and algorithm of quantum neural network based on the controlled Hadamard gates[J]. Journal of Computer Research and Development, 2015, 52(1): 211-220. 張翼鵬, 陳亮, 郝歡. 一種改進的量子神經(jīng)網(wǎng)絡訓練算法[J]. 電子與信息學報, 2013, 35(7): 1630-1635. doi: 10.3724/SP.J. 1146.2012.01417. ZHANG Y P, CHEN L, and HAO H. An improved training algorithm for quantum neural networks[J]. Journal of Electronics Information Technology, 2013, 35(7): 1630-1635. doi: 10.3724/SP.J.1146.2012.01417. 李楠, 侯旋. 自適應量子前向?qū)魉惴ㄑ芯縖J]. 電子與信息學報, 2013, 35(11): 2778-2783. doi: 10.3724/SP.J.1146.2013. 00101. LI N and HOU X. Research on adaptive quantum forward counter propagation algorithm[J]. Journal of Electronics Information Technology, 2013, 35(11): 2778-2783. doi: 10. 3724/SP.J.1146.2013. 00101. 郭通, 蘭巨龍, 李玉峰. 基于量子自適應粒子群優(yōu)化徑向基函數(shù)神經(jīng)網(wǎng)絡的網(wǎng)絡流量預測[J]. 電子與信息學報, 2013, 35(9): 2220-2226. doi: 10.3724/SP.J.1146.2012.01343. GUO T, LAN J L, and LI Y F. Network traffic prediction with radial basis function neural network based on quantum adaptive particle swarm optimization[J]. Journal of Electronics Information Technology, 2013, 35(9): 2220-2226. doi: 10.3724/SP.J.1146.2012.01343. 張鈴, 張鈸. M-P神經(jīng)元模型的幾何意義及其應用[J]. 軟件學報, 1998, 9(5): 334-338. ZHANG L and ZHANG B. A geometrical representation of M-P neural model and its applications[J]. Journal of Software, 1998, 9(5): 334-338. 張鈴, 張鈸, 殷海風. 多層前向網(wǎng)絡的交叉覆蓋設計算法[J]. 軟件學報, 1999, 10(7): 737-742. ZHANG L, ZHANG B, and YIN H F. An alternative covering design algorithm of multi-layer neural networks[J]. Journal of Software, 1999, 10(7): 737-742. GIULIANO B, GIULIO C, and GIULIANO S. Principles of Quantum Computation and Information Volume I: Basic Concepts[M]. Singapore: World Scientific, 2004: 108-112. -
計量
- 文章訪問數(shù): 1337
- HTML全文瀏覽量: 133
- PDF下載量: 394
- 被引次數(shù): 0
下載:
