Semi-supervised learning for structured regression on partially observed attributed graphs

@article{Stojanovic2015SemisupervisedLF,
  title={Semi-supervised learning for structured regression on partially observed attributed graphs},
  author={Jelena Stojanovic and M. Jovanovi{\'c} and Djordje Gligorijevic and Zoran Obradovic},
  journal={ArXiv},
  year={2015},
  volume={abs/1803.10705}
}
Conditional probabilistic graphical models provide a powerful framework for structured regression in spatio-temporal datasets with complex correlation patterns. However, in real-life applications a large fraction of observations is often missing, which can severely limit the representational power of these models. In this paper we propose a Marginalized Gaussian Conditional Random Fields (m-GCRF) structured regression model for dealing with missing labels in partially observed temporal… 
Structured regression on partially observed evolving graphs with uncertainty propagation
TLDR
An uncertainty propagation framework for Gaussian structured models that has shown great improvements over non-uncertainty propagation models as well as different state-of-the art uncertainty propagation regression models is developed.
Residual Correlation in Graph Neural Network Regression
TLDR
This work model the joint distribution of residuals on vertices with a parameterized multivariate Gaussian, and estimate the parameters by maximizing the marginal likelihood of the observed labels and develops linear time algorithms for low-variance, unbiased model parameter estimates, allowing us to scale to large networks.
Adaptive Skip-Train Structured Regression for Temporal Networks
TLDR
A novel model, Adaptive Skip-Train Structured Ensemble (AST-SE), which is a sampling-based structured regression ensemble for prediction on top of temporal networks, which takes advantage of the scheme of ensemble methods to allow multiple GCRFs to learn from several subnetworks.
Gaussian Conditional Random Fields for Classification
TLDR
A Gaussian conditional random fields model for structured binary classification (GCRFBC) is proposed, applicable to classification problems with undirected graphs, intractable for standard classification CRFs and achieves better prediction performance than unstructured predictors.
Functional norm regularization for margin-based ranking on temporal data
TLDR
Hypothesis of this dissertation is that use of the functional norm regularization can help alleviating mentioned challenges, by improving generalization abilities and/or learning efficiency of predictive models, in this case specifically of the approaches based on the ranking SVM framework.
Distance Based Modeling of Interactions in Structured Regression
TLDR
A Gaussian Conditional Random Field model is utilized, where it is extended its originally proposed interaction potential to include a distance term and an increase in predictive accuracy was observed on both synthetic examples and real-world applications, including challenging tasks from climate and healthcare domains.
Structured Regression on Multiscale Networks
TLDR
The authors walk through the intuitiveness of using multiple scales of networks on a real-life health informatics application and the time and memory demands from using the GCRF approach are logarithmic compared to naive implementations.
Uncertainty Propagation in Long-Term Structured Regression on Evolving Networks
TLDR
An effective novel iterative method is developed for Gaussian structured learning models in this study for propagating uncertainty in temporal graphs by modeling noisy inputs and provides evidence that the new approach allows better uncertainty propagation as compared to published alternatives.
Decomposition Based Reparametrization for Efficient Estimation of Sparse Gaussian Conditional Random Fields
TLDR
This work studied how reparametrization of the original problem may lead to more efficient estimation procedures, and used its Cholesky factor, which attractive properties allowed inexpensive coordinate descent based optimization algorithm, that is highly parallelizable.
Modeling Customer Engagement from Partial Observations
TLDR
This work proposes a robust framework for structured regression on deficient data in evolving networks with a supervised representation learning based on neural features embedding for predicting customer behavior on user networks generated from customer databases of two companies from different industries.
...
...

References

SHOWING 1-10 OF 28 REFERENCES
Aspects of Semi-supervised and Active Learning in Conditional Random Fields
TLDR
This work considers an extrapolation of a recently proposed semi-supervised criterion to the model of conditional random fields, and shows its drawbacks, and discusses alternative usage of the marginal probability and proposes a pool-based active learning approach based on quota sampling.
Continuous Conditional Random Fields for Regression in Remote Sensing
TLDR
Experimental results on the remote sensing problem of estimating Aerosol Optical Depth provide strong evidence that the proposed CRF model successfully exploits the inherent spatio-temporal properties of AOD data.
Scalable Tensor Factorizations for Incomplete Data
Simple, robust, scalable semi-supervised learning via expectation regularization
TLDR
Expectation regularization is presented, a semi-supervised learning method for exponential family parametric models that augments the traditional conditional label-likelihood objective function with an additional term that encourages model predictions on unlabeled data to match certain expectations---such as label priors.
Imputation of missing links and attributes in longitudinal social surveys
TLDR
The proposed method, based on an exponential random graph model, is evaluated experimentally for five scenarios of missing data models utilizing synthetic and real life datasets and suggests that it can be used as a viable imputation tool in longitudinal studies.
Semi-Supervised Learning Using Gaussian Fields and Harmonic Functions
TLDR
An approach to semi-supervised learning is proposed that is based on a Gaussian random field model, and methods to incorporate class priors and the predictions of classifiers obtained by supervised learning are discussed.
Graphical Models for Inference with Missing Data
TLDR
This work employs a formal representation called 'Missingness Graphs' to explicitly portray the causal mechanisms responsible for missingness and to encode dependencies between these mechanisms and the variables being measured.
Semi-Supervised Conditional Random Fields for Improved Sequence Segmentation and Labeling
We present a new semi-supervised training procedure for conditional random fields (CRFs) that can be used to train sequence segmentors and labelers from a combination of labeled and unlabeled
Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data
TLDR
This work presents iterative parameter estimation algorithms for conditional random fields and compares the performance of the resulting models to HMMs and MEMMs on synthetic and natural-language data.
Gaussian fields for semi-supervised regression and correspondence learning
...
...