The Feature Importance Ranking Measure

By Alexander Zien, Nicole Krämer, Sören Sonnenburg, Gunnar Rätsch. 2009

169 citations recorded by the source corpus.

Most accurate predictions are typically obtained by learning machines with complex feature spaces (as e.g. induced by kernels). Unfortunately, such decision rules are hardly accessible to humans and cannot easily be used to gain insights about the application domain. Therefore, one often resorts to linear models in combination with variable selection, thereby sacrificing some predictive power for presumptive interpretability. Here, we introduce the Feature Importance Ranking Measure (FIRM), which by retrospective analysis of arbitrary learning machines allows to achieve both excellent predictive performance and superior interpretation. In contrast to standard raw feature weighting, FIRM takes the underlying correlation structure of the features into account. Thereby, it is able to discover the most relevant features, even if their appearance in the training data is entirely prevented by noise. The desirable properties of FIRM are investigated analytically and illustrated in simulations.

Subfields: Artificial Intelligence

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

Citations in the corpus

A Survey of Methods for Explaining Black Box Models (2018, methodology): “In Reference [142] is introduced FIRM (Feature Importance Ranking Measure) for implementing function f as an extension of the POIMs [111].”
On the interpretation of weight vectors of linear models in multivariate neuroimaging (2014, methodology): “FIRM is now de ﬁ ned as the standard deviation of q f ( t ) FIRM : ¼ ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ Var q f t ð Þ (cid:6) (cid:7) r : ð D : 2 Þ In other words FIRM measures how the output of a source estimation algorithm or a stimulus estimator changes when a given feature changes.”
Explanation and Reliability of Individual Predictions (2013, background): “Most general explanation methods are based on marginalization of features [17, 35].”
Counterfactual Explanations for Machine Learning: A Review (2020, methodology): “Some approaches are model-agnostic, e.g. Golden Eye, PALM [59, 71, 116].”
Feature Importance Measure for Non-linear Learning Algorithms (2016, background): “Building up on POIMs and FIRM, we proposed MFI, which is a general measure of feature importance that is applicable to arbitrary learning machines.”
A General Method for Visualizing and Explaining Black-Box Regression Models (2011, methodology): “A recently published paper introduces FIRM, a method for computing the importance of features for a given model [9].”
TOWARDS HIERARCHICAL IMPORTANCE ATTRIBU- (2019, background): “On the other hand, global explanation algorithms (Guidotti et al., 2018) have also been studied for identifying generally important features, such as Feature Importance Ranking Measure (Zien et al., 2009), Accumulated Local Effects (Apley, 2016).”
Explaining the unique nature of individual gait patterns with deep learning (2018, background): “Also, interpreting linear and non-linear models have helped to gain interesting insights in neuroscience [24, 64], bioinformatics [80, 61, 71] and physics [56].”