Advances in statistical natural language processing using stochastic processes

2014 Fiscal Year Final Research Report

• Project/Area Number: 24700152
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: The Institute of Statistical Mathematics
• Principal Investigator: MOCHIHASHI Daichi 統計数理研究所, 大学共同利用機関等の部局等, 准教授 (80418508)
• Research Collaborator: OHISHI Yasunori NTTコミュニケーション科学基礎研究所 ; YOSHII Kazuyoshi 産業技術総合研究所/京都大学 ; UCHIUMI Kei ; TSUKAHARA Hiroshi デンソーITラボラトリ ; NOJI Hiroshi NII/総合研究大学院大情報学専攻
• Project Period (FY): 2012-04-01 – 2015-03-31
• Keywords: 統計的自然言語処理 / ノンパラメトリックベイズ / 統計的機械学習 / ベイズ統計学

Outline of Final Research Achievements

While unsupervised speech recognition was superseded by Glass et al. (ACL 2012) as a statistical natural language processing research leveraging stochastic processes, we could show new models in music information processing and signal processing in this area. Specifically, we proposed a novel non-negative matrix factorization for music information processing utilizing periodic frequencies with Gaussian processes, and an unsupervised multi-labeling infinite HMM using Markov Indian buffet processes.
In a statistical natural processing area, we could propose a novel language nonparametric Bayesian n-gram language models adapting to topical contexts, and an unsupervised morphological analysis with latent word classes.

Free Research Field

統計的自然言語処理