Contrastive Learning Using Spectral Methods

By James Y. Zou, Daniel J. Hsu, D. Parkes, Ryan P. Adams. 2013

100 citations recorded by the source corpus.

Subfields: Representation & Self-Supervised Learning

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

Citations in the corpus

ALICE: Active Learning with Contrastive Natural Language Explanations (2020, background): “Contrastive learning has previously been shown to substantially improve unsupervised learning (Abid et al., 2018), feature learning (Zou et al., 2015), and learning probabilistic models (Zou et al., 2013).”
Escaping From Saddle Points - Online Stochastic Gradient for Tensor Decomposition (2015, methodology): “Algorithm 2 Projected Noisy Stochastic Gradient Require: Stochastic gradient oracleSG(w), initial pointw0, desired accuracyκ.”
Bayesian group factor analysis with structured sparsity (2016, methodology): “, 2014), we used observationspecific sparse factors to a construct a gene co-expression network that is uniquely found in the samples from that exposure while explicitly controlling for shared covariance across exposures (Zou et al., 2013).”
Tensor decompositions for learning latent variable models (2012, positive): “Using the fact that M3 is an empirical third-order moment tensor, these steps can be computed with O(dk + N) operations, where N is the number of non-zero entries in the term-document matrix (Zou et al., 2013).”
Weakly-supervised Discovery of Visual Pattern Configurations (2014, methodology): “An alternative discriminative approach, but less scalable than covering, uses spectral methods [32].”
A Spectral Algorithm for Latent Dirichlet Allocation (2012, methodology): “We note that it is also possible to handle the case where ci,2 < 0 and ci,p < 0 for some i ∈ [k] (using techniques from [46]), though we avoid here this case for simplicity.”
Mind the Gap: A Generative Approach to Interpretable Feature Selection and Extraction (2015, background): “Contrastive learning [31] focuses on finding the directions that are most distinguish background data from foreground data.”
Spectral Methods Meet EM: A Provably Optimal Algorithm for Crowdsourcing (2014, methodology): “Our method for initializing the EM algorithm in crowdsourcing is inspired by recent work using spectral methods to estimate latent variable models [3, 1, 4, 2, 5, 27, 12, 13].”
Rich Component Analysis (2015, background): “A different notion of contrastive learning was introduced in (Zou et al., 2013).”