An accurate comparison of methods for quantifying variable importance in artificial neural networks using simulated data

By Julian D. Olden, M. Joy, R. Death. 2004

1,033 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Ecological Modeling, Nature and Landscape Conservation

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

In Ecological Modeling

In Nature and Landscape Conservation

Citations in the corpus

On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation (2015, methodology): “Quantifying the importance of input variables using a neural network model has also been studied in specific areas such as ecological modeling, where [29, 30] surveyed a large ensemble of possible analyses, including, computing partial derivatives, perturbation analysis, weights analysis, and studyi”
All Models are Wrong, but Many are Useful: Learning a Variable’s Importance by Studying an Entire Class of Prediction Models Simultaneously (2018, background): “We focus on the empirical versions of our importance metrics (M̂R, M̂CR− and M̂CR+), and compare them against AR, which is typically interpreted as an in-sample measure (Breiman, 2001), or as an intermediate step to estimate an alternate importance measure in terms of variable rankings (Gevrey et al”
A Comparative Study of PSO-ANN, GA-ANN, ICA-ANN, and ABC-ANN in Estimating the Heating Load of Buildings’ Energy Efficiency for Smart City Planning (2019, methodology): “, ANN 8-24-18-1) was investigated using the Olden method [78] to analyze the importance of the input variables.”
The use of artificial neural networks to analyze and predict alongshore sediment transport (2010, methodology): “Recently, some of the techniques available have also been reviewed (Gevrey et al., 2003; Olden et al., 2004) and the partial derivatives (PaD) method has been shown to have the best explanatory power.”
All Models are Wrong but many are Useful: Variable Importance for Black-Box, Proprietary, or Misspecified Prediction Models, using Model Class Reliance (2018, background): “…of our importance metrics (M̂R, M̂CR− and M̂CR+), and compare them against AR, which is typically interpreted as an in-sample measure (Breiman, 2001), or as an intermediate step to estimate an alternate importance measure in terms of variable rankings (Gevrey et al., 2003; Olden et al., 2004).”
Definitions, methods, and applications in interpretable machine learning (2019, background): “Methods have been developed to score individual features in many models including neural networks (52), random forests, (53, 54), and generic classifiers (55).”
History and trends in solar irradiance and PV power forecasting: A preliminary assessment and review using text mining (2018, methodology): “This situation is not irremediable since there are several tools to explore and understand the mechanics of ANNs (e.g., neural interpretation diagram, Garson’s algorithm and sensitivity analysis; Olden et al., 2004).”
Machine Learning Methods Without Tears: A Primer for Ecologists (2008, methodology): “Recent studies in the biological sciences have provided a variety of methods for quantifying and interpreting the contributions of the independent variables in neural networks (see Olden and Jackson 2002b; Gevrey et al. 2003; Olden et al. 2004).”
Multiscale effects of flow regime and habitat and their interaction on fish assemblage structure in eastern Australia (2007, methodology): “This approach is deemed the most appropriate as it has been shown to outperform other techniques for quantifying variable contributions in neural networks (Olden et al. 2004).”
VisualBackProp: visualizing CNNs for autonomous driving (2016, methodology): “Finally, understanding CNNs in the context relevant to this paper was also done in the past by characterizing their intriguing and often counter-intuitive properties [25, 26], quantifying the importance of input variables using a neural network model [27, 28], extracting the rules learned by the dec”
Hydrological disturbance benefits a native fish at the expense of an exotic fish. (2005, methodology): “To do this, we used the connection weight procedure ( Olden, Joy & Death 2004 ), running each model 100 times (Joy & Death 2004), and displaying the relationships graphically by means of ‘Lowess’ smoothing plots (Trexler & Travis 1993) for the most strongly contributing variables (i.e.”
Nonparametric variable importance assessment using machine learning techniques (2020, background): “The former category includes, for example, variable importance measures for random forests (Breiman, 2001; Ishwaran, 2007; Strobl et al. , 2007; Grömping, 2009) and neural networks (see, eg, Olden et al. , 2004), and ANOVA in linear models.”
Explainable artificial intelligence (XAI) in deep learning-based medical image analysis (2021, background): “, how much do features contribute to the output across the entire dataset (Olden et al. (2004)). As an illustration, one might observe from a model that – or even how much – high blood pressure increases the risk of a cardiac event. Another example of global explanation could be visualization of lea”