Conditional Likelihood Maximisation: A Unifying Framework for Information Theoretic Feature Selection

By Gavin Brown, Adam Craig Pocock, Ming-Jie Zhao, Mikel Luján. 2012

1,252 citations recorded by the source corpus.

Subfields: Statistics and Probability, 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 Statistics and Probability

In Information Theory

Citations in the corpus

Is Feature Selection Secure against Training Data Poisoning? (2015, background): “For instance, although in different forms, wrapper and embedded methods aim to minimize classiﬁcation error, while information theoretic ﬁlters optimize different estimates of the information gain (Brown et al., 2012).”
Adversarial Feature Selection Against Evasion Attacks (2016, methodology): “improve classification performance by tackling the course of dimensionality [38], [44].”
Single Subject Prediction of Brain Disorders in Neuroimaging: Promises and Pitfalls (2016, background): “MRMR tries to select features with maximum mutual information with class labels while minimizing the mutual information among those features (Brown et al., 2012a).”
Exploring the cloud of variable importance for the set of all good models (2020, background): “As a result, equation (10) implies the following.”
Minimum uncertainty latent variable models for robot recognition of sequential human activities (2017, methodology): “Then, a total number of 300 words are selected via a feature selection method [40] to reduce the complexity.”
Fizzy: feature subset selection for metagenomics (2015, methodology): “The current implementation of Fizzy has nine subset selection objective functions, which are all based on information theory (see Brown et al. for the mathematical details about the objective functions [17]).”
AccelPrint: Imperfections of Accelerometers Make Smartphones Trackable (2014, methodology): “To select features, we ranked features using the FEAST toolbox [7] and utilized the joint mutual information criterion for ranking (known to be effective for small training data [14]).”
Structural Agnostic Modeling: Adversarial Learning of Causal Graphs (2018, methodology): “…by minimizing the conditional log-likelihood of the data, given by: Snj (aj , θj , D) = − 1 n n∑ `=1 log qj(x (`) j |x (`) Pa(j;Ĝ) , θj) (4) Following (Brown et al., 2012), each conditional log-likelihood term is decomposed into three terms as follows, where p is the data distribution: log…”
Bootstrap rank‐ordered conditional mutual information (broCMI): A nonlinear input variable selection method for water resources modeling (2016, methodology): “BOOTSTRAP RANK-ORDERED INPUT VARIABLE SELECTION 2299 selection’’ [Brown et al., 2012].”
Data Analysis Strategies in Medical Imaging (2018, methodology): “Alternatively, supervised methods can be used to select features that are relevant to the particular target outcome(43,44).”
On the Stability of Feature Selection Algorithms (2017, methodology): “…methods can often control the model fit, but if the task is to identify meaningful subsets of features, there is a jungle of heuristics and domain-specific feature selection methods from which to pick, surveyed in several previous works, e.g. Guyon and Elisseeff (2003); Brown et al. (2012).”
Machine Learning methods for Quantitative Radiomic Biomarkers (2015, methodology): “Furthermore, publicly available implementations were readily available for these methods 27,28 , which increases their reusability.”
Recent advances and emerging challenges of feature selection in the context of big data (2015, background): “[55] presented a unifying framework for information theoretic feature selection, bringing almost two decades of research into heuristic filter criteria under a single theoretical umbrella.”
Mining on Someone Else's Dime: Mitigating Covert Mining Operations in Clouds and Enterprises (2017, methodology): “To obtain MI and JMI for each counter, we used FEAST, an open source toolbox for feature selection algorithms [31].”
Fifty years of pulsar candidate selection: from simple filters to a new principled real-time classification approach (2016, methodology): “It is one of the most common ﬁlter methods (Ko-havi & John 1997; Guyon & Elisseeff 2003; Brown et al. 2012) used for feature selection (Brown 2009).”