An Information Theoretic Framework for Multi-view Learning

By Karthik Sridharan, Sham M. Kakade. 2008

161 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning, Information Theory

How researchers see this paper

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

In Representation & Self-Supervised Learning

In Information Theory

Citations in the corpus

Co-regularized Alignment for Unsupervised Domain Adaptation (2018, methodology): “2Sridharan and Kakade [38] show that the bias introduced by co-regularization is small when each view carries sufficient information about Y on its own (i.e., mutual information I(Y ;X1|X2) and I(Y ;X2|X1) are small), and the generalization bounds comparing with the Bayes optimal predictor are also ”
What Makes Multimodal Learning Better than Single (Provably) (2021, background): “Besides, previous multi-view analysis [37, 1, 45, 13] typically assumes that each view alone is sufficient to predict the target accurately, which may not hold in our multi-modal setting.”
Lossy Compression for Lossless Prediction (2021, background): “This is related to the multi-view literature, where one only needs to retain information which is invariant across views [30, 31, 118, 119].”
Multitask Bregman clustering (2010, background): “Clustering ensemble [24,25] aims to combine multiple given partitions on an identical set of samples to reach a consensus.”
Improvability Through Semi-Supervised Learning: A Survey of Theoretical Results (2019, methodology): “Many existing methods can be cast in this framework; transductive support vector machines (Joachims, 1999; Boyd and Vandenberghe, 2004), Multi-View assumptions (Blum and Mitchell, 1998; Leskes, 2005; Sridharan and Kakade, 2008) and transductive graph-based methods (Blum and Chawla, 2001).”
Multi-View Intact Space Learning (2015, background): “" info with some small " info > 0 [46] states that both Z 1 and Z 2 are redundant with regards to their information about X .”
Self-supervised Learning from a Multi-view Perspective (2020, methodology): “Second, we adopt a common assumption in multi-view learning: either view alone is (approximately) sufﬁcient for the downstream tasks (see Assumption 1 in prior work (Sridharan & Kakade, 2008)).”
Dual Contrastive Prediction for Incomplete Multi-View Representation Learning (2022, methodology): “In contrast, our DCP unifies the consistency learning and data recovery into the framework of information theory [31].”
Multi-view learning overview: Recent progress and new challenges (2017, unknown): “It has been applied in co-training [36, 37].”
From Canonical Correlation Analysis to Self-supervised Graph Neural Networks (2021, methodology): “We formulate this hypothesis as a building block for analysis on downstream tasks [36, 9].”