引用本文: | 刘培磊,唐晋韬,谢松县,等.增量式神经网络聚类算法[J].国防科技大学学报,2016,38(5):137-142.[点击复制] |
LIU Peilei,TANG Jintao,XIE Songxian,et al.Incremental clustering algorithm of neural network[J].Journal of National University of Defense Technology,2016,38(5):137-142[点击复制] |
|
|
|
本文已被:浏览 9621次 下载 7429次 |
增量式神经网络聚类算法 |
刘培磊1,2, 唐晋韬1, 谢松县1, 王挺1 |
(1.国防科技大学 计算机学院, 湖南 长沙 410073;2.
2.国防信息学院 信息化建设系 信息资源管理教研室, 湖北 武汉 430010)
|
摘要: |
神经网络模型具有强大的问题建模能力,但是传统的反向传播算法只能进行批量监督学习,并且训练开销很大。针对传统算法的不足,提出全新的增量式神经网络模型及其聚类算法。该模型基于生物神经学实验证据,引入新的神经元激励函数和突触调节函数,赋予模型以坚实的统计理论基础。在此基础上,提出一种自适应的增量式神经网络聚类算法。算法中引入“胜者得全”式竞争等学习机制,在增量聚类过程中成功避免了“遗忘灾难”问题。在经典数据集上的实验结果表明:该聚类算法与K-means等传统聚类算法效果相当,特别是在增量学习任务的时空开销方面具有较大优势。 |
关键词: 神经网络 增量学习 聚类算法 时间开销 |
DOI:10.11887/j.cn.201605021 |
投稿日期:2015-09-28 |
基金项目:国家自然科学基金资助项目(61532001,61472436) |
|
Incremental clustering algorithm of neural network |
LIU Peilei1,2, TANG Jintao1, XIE Songxian1, WANG Ting1 |
(1. College of Computer, National University of Defense Technology, Changsha 410073, China;2.
2. Teaching and Research Section of Information Resource Management, Department of Information Construction, Academy of National Defense Information, Wuhan 430010, China)
|
Abstract: |
Neural network model is powerful in problem modelling. But the traditional back propagating algorithm can only execute batch supervised learning, and its time expense is very high. According to these problems, a novel incremental neural network model and the corresponding clustering algorithm were put forward. This model was supported by biological evidences, and it was built on the foundation of novel neuron’s activation function and the synapse adjusting function. Based on this, an adaptive incremental clustering algorithm was put forward, in which mechanisms such as “winner-take-all” were introduced. As a result, “catastrophic forgetting” problem was successfully solved in the incremental clustering process. Experiment results on classic datasets show that this algorithm’s performance is comparable with traditional clustering models such as K-means. Especially, its time and space expenses on incremental tasks are much lower than traditional clustering models. |
Keywords: neural network incremental learning clustering algorithm time expense |
|
|
|
|
|