Pattern classification with missing data: a review

By Pedro J. García-Laencina, José-Luis Sancho-Gómez, Aníbal R. Figueiras-Vidal. Neural computing & applications (Print), 2010

799 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/3f631c73b429fd2e59ab0b2a80a12a42d6fd00a8

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

1. Mihaela van der Schaar
2. Randall C. Wetzel
3. David C. Kale
4. Zachary Chase Lipton
5. Jinsung Yoon
6. James Jordon
7. William R. Zame
8. Zhiyong Cui

In Computer Vision and Pattern Recognition

1. David C. Kale
2. Randall C. Wetzel
3. Yinhai Wang
4. Ruimin Ke
5. Zhiyong Cui
6. Dana Hughes
7. John P. Dickerson
8. Mihaela van der Schaar

Citations in the corpus

• Directly Modeling Missing Data in Sequences with RNNs: Improved Classification of Clinical Time Series (2016, background): “While a long and rich literature addresses pattern recognition with missing data (Cohen and Cohen, 1975; Allison, 2001), most of this literature addresses fixed-length feature vectors, as indicated by reviews of the literature from Garc´ıa-Laencina et al. (2010) and Pigott (2001).”
• Counterfactual Explanations and Algorithmic Recourses for Machine Learning: A Review (2020, neutral): “Counterfactual explanation algorithms should also be able to handle missing feature values, which often happens in the real world [125].”
• Estimating Missing Data in Temporal Data Streams Using Multi-Directional Recurrent Neural Networks (2017, methodology): “…that our method yields large and statistically significant improvements in estimation accuracy over previous methods, including interpolation methods such as [1, 2], imputation methods such as [3, 4, 5, 6], RNN-based imputation methods such as [23, 24, 25] and matrix completion methods such as [7].”
• Flexible Imputation of Missing Data (2012, methodology): “This is, of course, an absurd imputation method, but one that is actually the best method according to Garćıa-Laencina et al. (2010), as it provides us with the highest classification accuracy.”
• GAIN: Missing Data Imputation using Generative Adversarial Nets (2018, methodology): “For instance, the approach for data imputation based on (Garc´ıa-Laencina et al., 2010) makes assumptions about the underlying distribution and fails to generalize well when datasets contain mixed categorical and continuous variables.”
• Modeling Missing Data in Clinical Time Series with RNNs (2016, background): “While a long and rich literature addresses pattern recognition with missing data (Cohen and Cohen, 1975; Allison, 2001), most of this literature addresses fixed-length feature vectors (Garćıa-Laencina et al., 2010; Pigott, 2001).”
• Materials that couple sensing, actuation, computation, and communication (2015, methodology): “Several techniques already exist for handling missing data [37].”
• Stacked Bidirectional and Unidirectional LSTM Recurrent Neural Network for Forecasting Network-wide Traffic State with Missing Values (2020, methodology): “Various data imputation methods for time series, including regression Chen et al. (2003), spectral analysis Mondal and Percival (2010), EM algorithm Garćıa-Laencina et al. (2010), and matrix factorization Koren et al. (2009) Chen et al. (2019b) have been developed and applied to estimate missing…”
• Counterfactual Explanations for Machine Learning: A Review (2020, background): “planation algorithms should also be able to handle missing feature values, which often happens in the real world [52].”