用创新计算动力学研究神经网络的组合创新-《计算机技术与发展》

文章信息/Info

Title:: Research on Combinatorial Innovation of Neural Network Using Dynamics for Computational Creativity

Author(s):: LI Qian; WEN Gui-hua; DING Yue-hua; Research Institute of Computer Application, South China Univ. of Technology

Keywords:: dynamics for computational creativity; laws of computational creativity; reminding and combining; neural network

摘要:: 创新计算动力学研究创新的基本规律，不仅为创新的机械化实现奠定基础，也为人类的创新活动提供规则化支持。文中利用创新计算动力学的基本定律研究神经网络技术的组合创新活动，归纳了组合创新的重要前提，同时证明了冲经网络的组合创新过程是符合创新计算的基本定律之联想与组合定律。最后，提出了相似组合、对立组合和信息域关联组合三种具体的组合创新模式。由于创新活动的广泛性和创新计算动力学的一般性，它们对于神经网络的未来发展以及其它理论或技术的创新都具有一定的启发意义

Abstract:: Dynamics for computational creativity （DCC） not only lays a theoretical basis for mechanization of innovation, but also provides the rules for human innovation activities. It is utilized to study the combinatorial innovation of neural network, and several important prerequisites to combinatorial innovation are proposed. At the same time, the innovation of neural network is proved to be in conformity with reminding and combining, which is one of the basic laws of DOC. Then according to the laws, three specific innovative schemas are concluded and illustrated. Because of the popularization of innovation and the generality of DCC, these prerequisites and innovative schemas will provide some procedural guidance for the development of neural network as well as the innovation of other technologies

备注/Memo:: 国家自然科学基金资助项目（60003019）李倩（1982-），女，江西吉安人，硕士研究生，研究方向为创新计算理论、人工智能等；文贵华，博士，副研究员，硕士生导师，研究方向为创新计算埋论、人工智能等；丁月华，高级工程师，教授，硕士生导师，研究方向为中间件技术、计算机应用技术研究和计算机测控