Co-clustering based classification for out-of-domain documents

By Wenyuan Dai, Gui-Rong Xue, Qiang Yang, Yong Yu. 2007

387 citations recorded by the source corpus.

In many real world applications, labeled data are in short supply. It often happens that obtaining labeled data in a new domain is expensive and time consuming, while there may be plenty of labeled data from a related but different domain. Traditional machine learning is not able to cope well with learning across different domains. In this paper, we address this problem for a text-mining task, where the labeled data are under one distribution in one domain known as in-domain data, while the unlabeled data are under a related but different domain known as out-of-domain data. Our general goal is to learn from the in-domain and apply the learned knowledge to out-of-domain. We propose a co-clustering based classification (CoCC) algorithm to tackle this problem. Co-clustering is used as a bridge to propagate the class structure and knowledge from the in-domain to the out-of-domain. We present theoretical and empirical analysis to show that our algorithm is able to produce high quality classification results, even when the distributions between the two data are different. The experimental results show that our algorithm greatly improves the classification performance over the traditional learning algorithms.

Subfields: Artificial Intelligence

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

Pre-Trained Models: Past, Present and Future (2021, methodology): “fer methods pre-train effective feature representations to pre-encode knowledge across domains and tasks (Johnson and Zhang, 2005; Evgeniou and Pontil, 2007; Dai et al., 2007; Raina et al., 2007).”
A Comprehensive Survey on Transfer Learning (2019, methodology): “In addition, the top two well-performed algorithms in terms of People versus Places are CoCC and TrAdaBoost.”
Transfer Learning with Graph Co-Regularization (2012, methodology): “, statistical property [8], [9], [10], [29], [11], [3], [12], and geometric structure [13], [14], [15], [16], [17].”
Generalized Transfer Subspace Learning Through Low-Rank Constraint (2014, background): “Liu et al. (2010) note that a data set is rarely well described by a single subspace, instead data are more likely to lie in several subspaces.”
Cross-Domain Learning from Multiple Sources: A Consensus Regularization Perspective (2010, methodology): “The data preparation method is similar to the one in [3].”
Multi-Fidelity Physics-Constrained Neural Network and Its Application in Materials Modeling (2019, background): “Transfer learning [3] is an approach to reduce the required amount of training data ,byreusingthesourcedata (instancetransfer)[4,5],previouslytrainedfeaturesorbasisfunctions(featurerepresentationtransfer)[6,7],orpreviouslytrainedparame-ters(parametertransfer)[8,9].”
Cross-domain sentiment classification via spectral feature alignment (2010, methodology): “Most existing domain adaptation methods can be classified into two categories: feature-representation adaptation [11, 8, 33, 2, 7, 12] and instance-weight adaptation [20].”
Knowledge transfer via multiple model local structure mapping (2008, background): “The “clustering” assumption in our work is exploited in some transfer learning and semi-supervised learning works [9, 28], where clustering structure is utilized in smoothing predictions among neighbors.”
Transfer across Completely Different Feature Spaces via Spectral Embedding (2013, background): “One line of research is to find a new feature space in which the training and test data have strong similarities (e.g., [9], [11], [14], [13], [24]).”
Domain Adaptation Problems: A DASVM Classification Technique and a Circular Validation Strategy (2010, neutral): “Other interesting approaches for domain adaptation have been presented by Dai et al. [34] and [35].”
Bilingual Co-Training for Sentiment Classification of Chinese Product Reviews (2011, methodology): “Dai et al. (2007b) use co-clustering as a bridge to propagate the class structure and knowledge from the in-domain to the out-of-domain.”
Feature-based transfer learning with real-world applications (2010, methodology): “In this experiment, we follow the preprocessing strategy in [47, 48] to create six data sets from this collection.”