«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

24

期数:

2014年06期

页码:

110-113

栏目:

智能、算法、系统工程

出版日期:

2014-06-30

文章信息/Info

Title:

esearch on Environmental Audio Classification Based on Active Learning

文章编号:

1673-629X（2014）06-0110-04

作者:

张雁¹; 吕丹桔¹; 王红崧²

1.西南林业大学 计算机与信息学院；2.西南林业大学 生态旅游学院

Author(s):

ZHANG Yan[1]; L Dan-ju[1]; WANG Hong-song[2]

关键词:

主动学习; 环境音分类; 采样; 熵优先采样; 简单不一致采样

Keywords:

active learning; environmental audio classification; sampling; EPS; SDS

分类号:

TP301

文献标志码:

A

摘要:

环境音分类是当前语音识别领域的研究热点。主动学习是利用未标记数据,在少量标记数据代价下提高监督学习算法的分类性能的方法。文中提出了熵优先采样( Entropy Priority Sampling,EPS)方法和简单不一致采样( Simple Disa-greement Sampling,SDS)方法作为主动学习选择样本的策略。针对环境音数据,提取11维的CELP音频特征,采用单一分类器与EPS,SDS方法对不同标记训练样本比例下的分类实验结果进行了比较分析。结果表明,主动学习方法在标记样本数较少的情况下,能取得较好的分类效果,并且EPS方法的性能优于SDS方法。

Abstract:

Environmental audio classification has been the focus in the field of speech recognition. Active learning enhances the perform-ance of supervised learning classification under the case of few labeled data. Propose EPS ( Entropy Priority Sampling) and SDS ( Simple Disagreement Sampling) methods as the selecting sampling strategies in active learning. For the given environmental audio data, the CELP features in 11 dimensions are extracted. The experiments with the single classifier,EPS and SDS on the environmental audio are carried out in order to illustrate the results of the proposed methods and compare their performance under different percent training sam-ple. The experimental results show that active learning can effectively improve the performance of environmental audio data classification, even under the fewer number of the training examples. The EPS method outperforms the SDS.

常用功能

更新日期/Last Update: 1900-01-01