Neural conditional random fields

By Trinh Minh Tri Do, Thierry Artières. 2010

142 citations recorded by the source corpus.

Subfields: Model Architectures, Information Extraction

How researchers see this paper

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

In Model Architectures

In Information Extraction

Citations in the corpus

Conditional Random Fields as Recurrent Neural Networks (2015, neutral): “For example, Do et al .”
Fully Connected Deep Structured Networks (2015, background): “More recently [5, 27, 22, 6, 28, 25] incorporate non-linearities into unary potentials but generally assume exact inference to be tractable.”
A Primer on Neural Network Models for Natural Language Processing (2015, methodology): “Dos Santos and Gatti (2014), dos Santos and Zadrozny (2014) and Kim et al.”
Effect of Non-linear Deep Architecture in Sequence Labeling (2013, methodology): “It was termed Conditional Neural Fields by Peng et al. (2009), and later Neural Conditional Random Fields by Do and Artieres (2010).”
Convolutional neural network based triangular CRF for joint intent detection and slot filling (2013, methodology): “In [8, 9], conditional neural field and neural CRF were introduced which combined the NN based feature learning and the CRF models.”
Cocktail Party Processing via Structured Prediction (2012, unknown): “Other related studies include [21] and [22].”
Semantic Role Labeling with Neural Network Factors (2015, methodology): “A similar use of non-linear potential functions in a structured model was proposed by Do and Artières (2010) for speech recognition, and by Durrett and Klein (2015) for syntactic phrase-structure parsing.”
DyNet: The Dynamic Neural Network Toolkit (2017, methodology): “In some cases, it is desirable to use sophisticated inference algorithms to compute or approximate a diﬀerentiable quantity such as the marginal likelihood or Bayes risk during learning [26, 15, 25, 38].”
Position-Aware Tagging for Aspect Sentiment Triplet Extraction (2020, methodology): “The loss function L for the training data D is de-ﬁned as: The overall model is analogous to that of a neural CRF (Peng et al., 2009; Do et al., 2010; Lam-ple et al., 2016); hence the inference and decoding follow standard marginal and MAP inference procedures.”
Hidden-Unit Conditional Random Fields (2011, methodology): “As an alternative, some studies have trained CRFs on feature representations learned by unsupervised learners (Do and Artieres, 2010; Mohamed et al., 2009); however, features discovered by unsupervised learning are not necessarily optimized for classification.”
Toward Robustness against Label Noise in Training Deep Discriminative Neural Networks (2017, background): “Recently, they have shown state-of-the-art results in segmentation [15, 16] when combined with deep neural networks [17, 18, 19].”
Sequence Classification Restricted Boltzmann Machines With Gated Units (2020, methodology): “…in this article to adapt to the sequence labeling task), and state-of-the-art models: 1) Structured Learning Ensemble (SLE) [7]: An optimized ensemble of 7 models: SVM multiclass , SVM struct , M3N, Perceptron, SEARN, CRF, and HMM. 2) Neural CRF [15]: A combination of CRF and deep networks.”