Segmental Dynamic Time Warping Research Articles

Assisted keyword indexing for lecture videos using unsupervised keyword spotting

Many students use videos to supplement learning outside the classroom. This is particularly important for students with challenged visual capacities, for whom seeing the board during lecture is difficult. For these students, we believe that recording the lectures they attend and providing effective video indexing and search tools will make it easier for them to learn course subject matter at their own pace. As a first step in this direction, we seek to help instructors create an index for their lecture videos using audio keyword search, with queries recorded by the instructor on their laptop and/or created from video excerpts. For this we have created an unsupervised within-speaker keyword spotting system. We represent audio data using de-noised, whitened and scale-normalized Mel Frequency Cepstral Coefficient (MFCC) features, and locate queries using Segmental Dynamic Time Warping (SDTW) of feature sequences. Our system is evaluated using introductory Linear Algebra lectures from instructors with different accents at two U.S. universities. For lectures produced using a video camera at RIT, laptop-recorded queries obtain an average Precision at 10 of 71.5%, while 79.5% is obtained for within-lecture queries. For lectures recorded using a lapel microphone at MIT, using a similar keyword set we obtain a much higher average Precision at 10 of 89.5%. Our results suggest that our system is robust to changes in environment, speaker and recording setup.

Similarity Search in Data Stream with Adaptive Segmental Approximations

PDF HTML阅读 XML下载 导出引用 引用提醒 基于适应性分段估计的数据流相似性搜索 DOI: 作者: 作者单位: 作者简介: 通讯作者: 中图分类号: 基金项目: Supported by the National High-Tech Research and Development Plan of China under Grant Nos.2006AA01Z451, 2007AA01Z474 (国家高技术研究发展计划(863)) Similarity Search in Data Stream with Adaptive Segmental Approximations Author: Affiliation: Fund Project: 摘要 | 图/表 | 访问统计 | 参考文献 | 相似文献 | 引证文献 | 资源附件 | 文章评论 摘要:相似性搜索在股票交易行情、网络安全、传感器网络等众多领域应用广泛.由于这些领域中产生的数据具有无限的、连续的、快速的、实时的特性,所以需要适合数据流上的在线相似性搜索算法.首先,在具有或不具有全局约束条件下,分别提出了没有索引结构的DTW(dynamic time warping)下限函数LB_seg_WFglobal和LB_seg_WF,它们是一种分段DTW技术,能够处理数据流上的非等长序列间在线相似性匹配问题.然后,为了进一步提高LB_seg_WFglobal和LB_seg_WF的近似程度,提出了一系列的改进方法.最后,针对流上使用LB_seg_WFglobal或LB_seg_WF可能会出现连续失效的情况,分别提出了DTW的下限函数LB_WFglobal(具有全局约束条件)和上限函数UB_WF、下限函数LB_WF(不具有全局约束条件).通过增量方式快速估计DTW,极大地减少了估计DTW的冗余计算量.通过理论分析和统计实验,验证了该方法的有效性. Abstract:Similarity search has attracted many researchers from various communities (real-time stock quotes, network security, sensor networks). Due to the infinite, continuous, fast and real-time properties of the data from these communities, a method is needed for online similarity search in data stream. This paper first proposes the lower bound function LB_seg_WFglobal for DTW (dynamic time warping) in the presence of global warping constraints and LB_seg_WF for DTW without global warping constraints, which are not applied to any index structures. They are segmented DTW techniques, and can be applied to sequences and queries of varying lengths in data stream. Next, several tighter lower bounds are proposed to improve the approximate degree of the LB_seg_WFglobal and LB_seg_WF. Finally, to deal with the possible continuously non-effective problem of LB_seg_WFglobal or LB_seg_WF in data stream, it is believed that lower-bound LB_WFglobal (in the presence of global warping constraints) and lower-bound LB_WF, upper-bound UB_WF (without global warping constraints) can fast estimate DTW and hence reduce a lot of redundant computations by incrementally computing. The theoretical analysis and statistical experiments confirm the validity of the proposed methods. 参考文献 相似文献 引证文献