摘要
在語音辨識中，鑑別式訓練(discriminative training)常被用來增加語音模型的鑑別度，進而提升語音辨識率，而最小分類錯誤訓練(MCE，minimum classification error)則是鑑別式訓練中最具有代表性的一項技術，但是，因為語音辨識包含了三種錯誤，分別為插入(insertion)、刪除(deletion)與混淆(substitution)錯誤，這三種錯誤對於鑑別式訓練會產生不一樣的影響，因此，我們提出了加強型的最小分類錯誤訓練技術(E-MCE, enhanced minimum classification error)來平衡三種錯誤的影響，本技術是由最小插入錯誤(MIE, minimum insertion error)、最小刪除錯誤(MDE, minimum deletion error)與最小混淆錯誤(MSE, minimum substitution error)訓練技術所組合而成， 以交替使用的方式，利用以上三種訓練技術來進行鑑別式訓練，最後可以得到比傳統的最小分類錯誤訓練更佳之錯誤改善效果。
並列摘要
In continuous speech recognition substitution, insertion and deletion errors usually not only vary in numbers but also have different degrees of impact on optimizing a set of acoustic models. To balance their contributions to the overall error, an enhanced minimum classification error (E-MCE) learning framework is developed. The basic idea is to partition acoustic model optimization into three subtasks, i.e., minimum substitution errors (MSE), insertion errors (MIE) and deletion errors (MDE), and select/generate three corresponding sets of competing hypotheses, one for each individual sub-problem. MSE, MIE and MDE are then sequentially executed to gradually reduce the overall word error rates.
延伸閱讀
- 余文德、廖珗洲、蕭文達、張憲寬、吳定餘、林楨中（2020）。混合式機器學習方法於施工人員個人安全裝備即時辨識之應用。技術學刊，35(4)，155-165。https://www.airitilibrary.com/Article/Detail?DocID=10123407-202012-202012180006-202012180006-155-165
- 張育瑞（2016）。一種人體姿態命令辨識及其身份識別的強化式方法之研究〔碩士論文，國立虎尾科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0028-1608201616302200
- 徐照揚（2023）。基於強化式學習之服務意向辨識系統〔碩士論文，中原大學〕。華藝線上圖書館。https://doi.org/10.6840/cycu202301105
- 林逢春（2009）。Improving Learning Efficiency by Attribute Reduction〔碩士論文，亞洲大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0118-1511201215461937
- Wu, S. Y. (2009). A Methodology to Develop Pervasive Applications [master's thesis, National Chiao Tung University]. Airiti Library. https://doi.org/10.6842/NCTU.2009.00659