Partially Supervised Classification of Text Documents
By B. Liu, Wee Sun Lee, Philip S. Yu, Xiaoli Li. International Conference on Machine Learning, 2002
660 citations recorded by the source corpus.
View Paper: https://www.semanticscholar.org/paper/006b577653e0e91e1cfd37c32c6d1cfb198a730e
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
Citations in the corpus
- Learning from positive and unlabeled data: a survey (2018, methodology): “Under this assumption, the optimal classifier can be found by looking for the classifier that classifies all labeled examples as positive and as few as possible examples as negative [63,5].”
- Learning with Positive and Unlabeled Examples Using Weighted Logistic Regression (2003, methodology): “Sample complexity for the case where the positive and unlabeled examples can be sampled is given in (Liu et al., 2002), where it was shown that maximizing the number of examples classi-fied as negative while constraining the function to correctly classify positive examples will give good performance…”
- A bagging SVM to learn from positive and unlabeled examples (2010, methodology): “Many previous works, including Liu et al. (2003) and Yu et al.”
- Transferring Naive Bayes Classifiers for Text Classification (2007, methodology): “Liu et al. (2002) proposed a heuristic approach Spy-EM that can learn how to handle training and test data with nonoverlapping class labels.”
- Positive-Unlabeled Learning with Non-Negative Risk Estimator (2017, methodology): “The first category (e.g., Liu et al., 2002; Li & Liu, 2003) naively identifies U data that are likely to be negative (N), and then runs PN learning (i.e., ordinary supervised learning).”
- Sentiment Analysis and Opinion Mining (2012, methodology): “Learning from positive and unlabeled examples (PU learning) using the S-EM algorithm (Liu et al., 2002) was considerably better.”
- Introduction to Semi-Supervised Learning (2009, background): “• learning from positive and unlabeled data,when there is no negative labeled data [61,114,109];”
- Federated Learning with Only Positive Labels (2020, background): “Different from PU learning (Liu et al., 2002; Elkan & Noto, 2008; Plessis et al., 2015; Hsieh et al., 2015), in the federated learning setting, the clients do not have access to unlabeled data for both positive and negative classes.”
- Building text classifiers using positive and unlabeled examples (2003, methodology): “Using NB in each iteration, EM employs the same equations as those used in building a NB classifier (equation (3) for the Expectation step, and equations (1) and (2) for the Maximization step) [26] [20].”
- One-class classification: taxonomy of study and review of techniques (2013, unknown): “It is a common practice to build text classifiers using positive and unlabeled examples(6) (in semi-supervised setting(7)) as collecting unlabeled samples is relatively easy and fast in many text or Web page domains (Nigam et al., 2000; Liu et al., 2002).”
- A survey on opinion mining and sentiment analysis: Tasks, approaches and applications (2015, methodology): “Fusilier [232] developed a modified PU-learning (Positive and Unlabeled samples based learning) [233] in order to detect positive and negative deceptive opinions.”
- PEBL: Web page classification without negative examples (2004, background): “Many works attempt rule-based or probability-based learning from positive or positive and unlabeled data [18], [19], [10], [9].”
- Semi-Supervised Novelty Detection (2010, methodology): “Most notably, Liu et al. (2002) provide sample complexity bounds for VC classes for the learning rule that minimizes the number of false negatives while controlling the proportion of false positiveat a certain level.”
- A Survey of Recent Trends in One Class Classification (2009, methodology): “[29] propose a method (called Spy EM) to solve this problem in the text domain.”