Learning from Multiple Partially Observed Views - an Application to Multilingual Text Categorization

By Massih-Reza Amini, Nicolas Usunier, Cyril Goutte. Neural Information Processing Systems, 2009

383 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/82402bf63b039073e2027246d1d4332781b15002

How researchers see this paper

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

1. Qiang Yang
2. Gabriella Csurka
3. Mingsheng Long
4. Wen Li
5. Yu Zhang
6. Dacheng Tao
7. Dong Xu
8. Trevor Darrell

Citations in the corpus

• Multimodal Manifold Analysis by Simultaneous Diagonalization of Laplacians (2015, unknown): “Method Circles Text Caltech [24], [28] NUS [24], [45] Digits [46], [47] Reuters [47], [48]”
• Co-regularized Multi-view Spectral Clustering (2011, background): “• Reuters Multilingual data: The test collection contains feature characteristics of do cuments originally written in five different languages (English, Fr ench, German, Spanish and Italian), and their translations, over a common set of 6 categories [13].”
• A Co-training Approach for Multi-view Spectral Clustering (2011, background): “7 • Reuters Multilingual data: The test collection contains feature characteristics of documents originally written in five different languages (English, French, German, Spanish and Italian), and their translations, over a common set of 6 categories (Amini et al., 2009).”
• Learning with Feature and Distribution Evolvable Streams (2020, background): “• Reuters multilingual dataset (Amini et al., 2009) contains about 11K articles from 6 classes in 5 languages so that we can simulate the evolving stream by various languages.”
• Multi-view low-rank sparse subspace clustering (2017, background): “For example, same documents may be available in multiple languages [1] or different descriptors can be constructed from the same images [2].”
• Inductive multi-task learning with multiple view data (2012, background): “The underlying assumption of multi-view algorithm is that each view is conditionally independent from other views [6] and is sufficient for constructing a predictive model [2, 14].”
• Deep multi-view learning methods: A review (2021, background): “1https://www.cs.columbia.edu/CAVE/software/softlib/coil-20.php 2http://lear.inrialpes.fr/people/vandeweijer/data 3http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html Dataset Type Views Categories Size Year Highlights 20NGs 4 Text 3 5 500 - Selected from the popular 20Newsgroups dataset”
• Learning with Augmented Features for Heterogeneous Domain Adaptation (2012, methodology): “We use the Reuters multilingual dataset 2 (Amini et al., 2009), which is collected by sampling parts of the Reuters RCV1 and RCV2 collections.”
• Data-Dependent Convergence for Consensus Stochastic Optimization (2016, methodology): “Covertype is the forest covertype dataset [24] used in [4] obtained from the UC Irvine Machine Learning Repository [25], and rcv1 is from the Reuters collection [26] obtained from libsvm collection [27].”
• Generalized Latent Multi-View Subspace Clustering (2020, unknown): “Taking the datasets Reuters and Politicsie for example, lLMSC outperforms all the traditional methods; (2) by exploring the general correlation with a neural network, gLMSC significantly improves lLMSC on 6 out of 8 datasets.”
• Large-Scale Heterogeneous Feature Embedding (2019, methodology): “Reuters (Amini, Usunier, and Goutte 2009): 18,758 documents from Reuters are used as instances.”
• On constrained spectral clustering and its applications (2012, methodology): “We used the Reuters Multilingual dataset, first introduced by Amini et al (2009).”
• Bilingual Co-Training for Sentiment Classification of Chinese Product Reviews (2011, background): “The most recent related work includes multilingual text categorization based on multi-view learning (Amini, Usunier, and Goutte 2009; Amini and Goutte 2010).”
• Multi-View Clustering via Joint Nonnegative Matrix Factorization (2013, background): “• Reuters Multilingual dataset(5): This test collection contains feature characteristics of documents originally written in five different languages , and their translations, over a common set of 6 categories [2].”
• Transfer Hashing with Privileged Information (2016, methodology): “To verify the effectiveness of our proposed approaches, ITQ+ and LapITQ+, we conduct a series of experiments on three benchmark datasets: BBC Collection [Greene and Cunningham, 2006], multilingual Reuters [Amini et al., 2009], and NUS-WIDE [Chua et al., 2009].”