Text Classification from Labeled and Unlabeled Documents using EM

By Kamal Nigam, Andrew McCallum, Sebastian Thrun, Tom Michael Mitchell. 2000

3,258 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing

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

In Natural Language Processing

Citations in the corpus

Semi-supervised sequence tagging with bidirectional language models (2017, methodology): “…methods for structured prediction problems include co-training (Blum and Mitchell, 1998; Pierce and Cardie, 2001), expectation maximization (Nigam et al., 2000; Mohit and Hwa, 2005), structural learning (Ando and Zhang, 2005) and maximum discriminant functions (Suzuki et al., 2007; Suzuki…”
Revisiting LSTM Networks for Semi-Supervised Text Classification via Mixed Objective Function (2019, methodology): “Previous work regarding the application of SSL to this task includes training Naive Bayes model on unlabeled data using expectation-maximization algorithm [22], incorporation of multi-view embeddings representation of input words along with its one-hot representation as model’s input [9], use of add”
Understanding Dataset Difficulty with V-Usable Information (2021, neutral): “Uncertainty sampling, for example, picks the instances that the partially trained model is least certain about (Lewis & Gale, 1994; Nigam et al., 2000), which could be interpreted as a measure of difficulty.”
Dropout Training as Adaptive Regularization (2013, background): “MNB with EM [20] failed to give an improvement.”
Identifiability of parameters in latent structure models with many observed variables (2008, methodology): “Finite mixtures of products of finite measure are widely used to model data, for instance in biological taxonomy, medical diagnosis, or classification of text documents [21, 35].”
Toward integrating feature selection algorithms for classification and clustering (2005, background): “…[40], [63], [81], [90], machine learning [6], [41], [43], [44], and data mining [17], [18], [42], and widely applied to many fields such as text categorization [50], [70], [94], image retrieval [77], [86], customer relationship management [69], intrusion detection [49], and genomic analysis [91].”
Predicting sample size required for classification performance (2012, background): “Literature suggests that the predictive power of the classifiers is largely dependent on the quality and size of the training sample [1-6].”
Semi-Supervised Multitask Learning (2007, methodology): “Semi-supervised learning has been an area of significant recent interest in the machinelearning community [2], [3], [4], [5], [6], [7], [8], [9].”
Improving Semi-Supervised Learning with Auxiliary Deep Generative Models (2015, methodology): “There exist many approaches to performing semi-supervised learning, e.g. transductive SVM (TSVM) (Joachims, 1999), EM methods (Nigam et al., 2000), graph-based methods (Blum and Chawla, 2001; Zhu et al., 2003; Pitelis et al., 2014) and deep auto-encoders (Rifai et al., 2011; Ranzato and Szummer,…”
A brief introduction to weakly supervised learning (2018, background): “Generative methods [19,20] assume that both labeled and unlabeled data are generated from the same inherent model.”
Review of "Mining the Web: discovering knowledge from hypertext data" by Soumen Chakrabati. Morgan Kaufman 2003. (2002, neutral): “[164].”
A biterm topic model for short texts (2013, methodology): “[35] modeled each tweet in the way of mixture of unigrams [23].”
P-N learning: Bootstrapping binary classifiers by structural constraints (2010, methodology): “EM was successfully applied to document classification [17] and learning of object categories [8].”
PEBL: Web page classification without negative examples (2004, methodology): “Kamal Nigam inserted two parameters into EM (to relax the generative assumptions): one for controlling the contributions of labeled data and unlabeled data and the other for controlling the quantity of mixture components corresponding to one class [16].”
Introduction to the special topic section on mining Web resources for enhancing information retrieval (2007, methodology): “Techniques such as PEBL learning framework (Yu et al., 2004), semisupervised learning ( Nigam, McCallum, Thrun, & Mitchell, 2000 ), singleclass learning (Denis, 1998), and one class support vector machine (OSVM) (Manevitz & Yousef, 2002) have been adopted to solve the problem.”
Hyperparameter Learning for Graph Based Semi-supervised Learning Algorithms (2006, background): “(Nigam, 2001; Nigam et al., 1999) proposed introducing hierarchical class relationship: sub-topics under one relatively broad topic (i.”