Tri-training: exploiting unlabeled data using three classifiers

By Zhi-Hua Zhou, Ming Li. 2005

1,189 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Evaluation & Benchmarks

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 Evaluation & Benchmarks

Citations in the corpus

Semi-Supervised Sequence Modeling with Cross-View Training (2018, background): “It is also possible to use multiple models or prediction modules for the teacher, such as in tri-training (Zhou and Li, 2005; Ruder and Plank, 2018).”
Break-It-Fix-It: Unsupervised Learning for Program Repair (2021, background): “Similarly, co-training (Blum & Mitchell, 1998) and tri-training (Zhou & Li, 2005) train multiple models and add predicted targets on which these models agree.”
Addressing Semantic Drift in Question Generation for Semi-Supervised Question Answering (2019, neutral): “models agree on (Blum and Mitchell, 1998; Zhou and Li, 2005).”
Revisiting Self-Training for Neural Sequence Generation (2019, background): “In the same spirit, democratic co-training (Zhou & Goldman, 2004) or tri-training (Zhou & Li, 2005) trains multiple models with different conﬁgurations on the same data feature set, and a subset of the models act as teachers for others.”
A Survey on Data Collection for Machine Learning: A Big Data - AI Integration Perspective (2018, background): “However, among the ensemble learning algorithms bagging, boosting, and random subspace method, bagging (also used by Tri-training) has the best empirical performance according to the survey [99].”
A Survey on Multi-view Learning (2013, methodology): “Zhou and Li (2005b) proposed another co-training style semi-supervised algorithm called tri-training, which does not require that the instance space be described with sufficient and redundant views, nor does it put any constraints on the supervised algorithm, as do Goldman and Zhou (2000).”
Deep Co-Training for Semi-Supervised Image Recognition (2018, background): “But dual-view is only a special case of multi-view learning, and multiview co-training has also been studied for other problems [30, 31].”
The Cascade Transformer: an Application for Efficient Answer Sentence Selection (2020, background): “…variants of ST have 2 Also known as auxiliary or diagnostic classiﬁers. been proposed, including treating teacher predictions as soft labels (Hinton et al., 2015), masking part of the label (Clark et al., 2018), or use multiple modules for the teacher (Zhou and Li, 2005; Ruder and Plank, 2018).”
Self-labeled techniques for semi-supervised learning: taxonomy, software and empirical study (2015, background): “The goal of this approach, which has been shown to perform well in several domains, is simplicity [20].”
Data collection and quality challenges in deep learning: a data-centric AI perspective (2021, methodology): “This approach assumes that we can trust the high confidence, but there are other techniques including Tritraining [186], Co-learning [185], and Co-training [20] that do not rely on this assumption.”
Towards Making Unlabeled Data Never Hurt (2011, methodology): “The third category is disagreement-based methods, e.g., [8], [50].”
Collaborative Unsupervised Domain Adaptation for Medical Image Diagnosis (2019, background): “Moreover, tritraining [47] can be regarded as an extension of co-training.”
UDALM: Unsupervised Domain Adaptation through Language Modeling (2021, methodology): “UDA approaches in the literature can be grouped in three major categories, namely pseudo-labeling techniques (e.g. Yarowsky, 1995; Zhou and Li, 2005), domain adversarial training (e.g. Ganin et al., 2016) and pivot-based approaches (e.g. Blitzer et al., 2006; Pan et al., 2010).”
Co-regularized Alignment for Unsupervised Domain Adaptation (2018, background): “Tri-training [45] can be regarded as an extension of co-training [3] and uses the output of three different classifiers to assign pseudo-labels to unlabeled examples.”
Stochastic Classifiers for Unsupervised Domain Adaptation (2020, methodology): “In this paper, we follow the adversarial training pipeline, but it is worth noting that our method can also be applied to non-adversarial based methods such as [43, 61] as long as they require multiple classifiers.”
A survey on semi-supervised learning (2019, methodology): “Zhou and Li (2005b) proposed tri-training, where three classifiers are alternately trained.”
A Survey on Deep Semi-Supervised Learning (2021, methodology): “Tri-net [220], a deep learning-based method inspired by the tri-training [221].”
Deep Contextualized Self-training for Low Resource Dependency Parsing (2019, neutral): “Recently, Ruder and Plank (2018) evaluated the impact of self-training and the closely related tri-training method (Zhou and Li, 2005; Søgaard, 2010) on DNNs for partof-speech (POS) tagging and sentiment analysis.”