Sequential Labeling Using Deep-Structured Conditional Random Fields

By Dong Yu, Shizhen Wang, Li Deng. 2010

60 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Deep Learning: Methods and Applications (2014, methodology): “For example, the deep-structured CRF, which stacks many layers of CRFs, have been usefully applied to the task of language identification [429], phone recognition [410], sequential labeling in natural language processing [428], and confidence calibration in speech recognition [423].”
Deep Learning of Representations: Looking Forward (2013, methodology): “With the less common supervised pre-training (Bengio et al., 2007; Yu et al., 2010; Seide et al., 2011b), each additional layer is trained with a supervised objective (as part of a one hidden layer network).”
Column Networks for Collective Classification (2016, methodology): “There has been recent work that combines deep learning with structured prediction but the main learning and inference problems for general multi-relational settings remain open (Belanger and McCallum 2016; Do, Arti, and others 2010; Tompson et al. 2014; Yu, Wang, and Deng 2010; Zheng et al. 2015).”
Roles of Pre-Training and Fine-Tuning in Context-Dependent DBN-HMMs for Real-World Speech Recognition (2010, methodology): “Deep learning techniques have been successfully applied to applications such as natural language processing [12][13], language recognition [14], image processing [15], vision [16], audio encoding [17], audio classification [18], phone recognition [19], and semantic tagging [20] in the last several y”
Deep-structured hidden conditional random fields for phonetic recognition (2010, background): “proposed the deep-structured conditional random fields (DCRFs) [17] and obtained better than the state-of-the-art results on the query labeling [4] tasks.”
Deep Learning and Its Applications to Signal and Information Processing (2011, methodology): “For example, the deep-structured CRF, which stacks many layers of CRFs, have been successfully used in the speech-related task of language identification, phone recognition, sequential labeling [ 15 ], and confidence calibration.”
An Overview of Deep-Structured Learning for Information Processing (2011, methodology): “Various versions of deep-structured CRFs are usefully applied to phone recognition (Yu and Deng, 2010), spoken language identification (Yu, Wang, Karam, and Deng, 2010), and natural language processing (Yu, Wang, and Deng, 2010).”
Scalable stacking and learning for building deep architectures (2012, methodology): “Similarly stacked, deep-CRF was used in speech processing [8].”
Learning in the Deep-Structured Conditional Random Fields (2009, methodology): “Given their discriminative nature and their high flexibility in choosing features, CRFs have been widely and successfully used to solve sequential labeling problems, notably those in natural language processing [1] [2] and speech processing [3].”
Convolutional neural network based triangular CRF for joint intent detection and slot filling (2013, methodology): “In [12], a multi-layer CRF was used for sequence labeling, but the feature extraction was also different, and no joint modeling was performed.”
Language recognition using deep-structured conditional random fields (2010, methodology): “In the deep-structured CRF, both model parameter estimation and state sequence inference are carried out layer-by-layer in a bottom-up manner so that the computational complexity is limited to at most linear to the number of layers used [13][14].”
Deep Convex Net: A Scalable Architecture for Speech Pattern Classification (2011, methodology): “This particular way of serially connecting any adjacent modules in the DCN has been motivated partly by our earlier work on the deep-structured conditional random field [11].”
Porting concepts from DNNs back to GMMs (2013, background): “Likewise, the combination of log-linear models to form deep structures has been investigated before [24].”