Lars Henry Berge Olsen

In Explainable AI (XAI), Shapley values are a popular model-agnostic framework for explaining predictions made by complex machine learning models. The computation of Shapley values requires estimating non-trivial contribution functions representing predictions with only a subset of the features present. As the number of these terms grows exponentially with the number of features, computational costs escalate rapidly, creating a pressing need for efficient and accurate approximation methods. For tabular data, the framework is considered the state-of-the-art model-agnostic approximation framework. approximates the Shapley values using a weighted sample of the contribution functions for different feature subsets. We propose a novel modification of which replaces the stochastic weights with deterministic ones to reduce the variance of the resulting Shapley value approximations. This may also be combined with our simple, yet effective modification to the variant implemented in the popular Python library. Additionally, we provide an overview of established methods. Numerical experiments demonstrate that our methods can reduce the required number of contribution function evaluations by 5% to 50 % while preserving the same accuracy of the approximated Shapley values – essentially reducing the running time by up to 50%. These computational advancements push the boundaries of the feature dimensionality and number of predictions that can be accurately explained with Shapley values within a feasible runtime.

Vi snakker med Jon Nordby ("Head of Data Science" i Soundsensing) og Lars Henry Berge Olsen (forsker på NR): Hvordan utvikler Soundsensing prediktiv feildeteksjon for næringseiendommer? Og hva er egentlig forskjellen på avviksdeteksjon og feildeteksjon? Dette jobber vi med i "EarOnEdge" – som står for "On-Edge Anomaly Detection in Machinery Using Sound as a Data Source". EarOnEdge er et såkalt Innovasjonsprosjekt i Næringslivet (IPN) og finansiert av Norges forskningsråd. Prosjektet utvikles i direkte samarbeid med brukergruppen med Malling & Co AS som kompetansepartner for drift og forvaltning av bygg. I prosjektet jobber NRs forskere med kompetanse på maskinlæring og statistisk modellering sammen med Soundsensing for å utvikle bedre algoritmer for feildeteksjon for næringseiendommer. EarOnEdge kan bidra til at Soundsensing får et stort internasjonalt marked om noen år. Sannsynligvis VIKTIG er en serie om KI og digitalisering av Norsk Regnesentral, Anders Løland er programleder, gjesteopptreden av Sigurd Rønning. Produsent: Elin Ruhlin Gjuvsland.

Lars Henry Berge Olsen er til vanlig PhD-student ved BigInsight og Universitetet i Oslo, og akkurat nå er han ved Alan Turing-instituttet i London. Vi snakker om hva Alan Turing-instituttet er og om Lars' egen forskning på forklarbar kunstig intelligens, som kan ligne litt på en diskusjon om hvordan en bør dele taxi-regninga. Med Anders Løland i studio, produsent er Elin Ruhlin Gjuvsland. En podkastserie av Norsk Regnesentral.

Shapley values originated in cooperative game theory but are extensively used today as a model-agnostic explanation framework to explain predictions made by complex machine learning models in the industry and academia. There are several algorithmic approaches for computing different versions of Shapley value explanations. Here, we consider Shapley values incorporating feature dependencies, referred to as conditional Shapley values, for predictive models fitted to tabular data. Estimating precise conditional Shapley values is difficult as they require the estimation of non-trivial conditional expectations. In this article, we develop new methods, extend earlier proposed approaches, and systematize the new refined and existing methods into different method classes for comparison and evaluation. The method classes use either Monte Carlo integration or regression to model the conditional expectations. We conduct extensive simulation studies to evaluate how precisely the different method classes estimate the conditional expectations, and thereby the conditional Shapley values, for different setups. We also apply the methods to several real-world data experiments and provide recommendations for when to use the different method classes and approaches. Roughly speaking, we recommend using parametric methods when we can specify the data distribution almost correctly, as they generally produce the most accurate Shapley value explanations. When the distribution is unknown, both generative methods and regression models with a similar form as the underlying predictive model are good and stable options. Regression-based methods are often slow to train but quickly produce the Shapley value explanations once trained. The vice versa is true for Monte Carlo-based methods, making the different methods appropriate in different practical situations.

Shapley values are today extensively used as a model-agnostic explanation framework to explain complex predictive machine learning models. Shapley values have desirable theoretical properties and a sound mathematical foundation in the field of cooperative game theory. Precise Shapley value estimates for dependent data rely on accurate modeling of the dependencies between all feature combinations. In this paper, we use a variational autoencoder with arbitrary conditioning (VAEAC) to model all feature dependencies simultaneously. We demonstrate through comprehensive simulation studies that our VAEAC approach to Shapley value estimation outperforms the state-of-the-art methods for a wide range of settings for both continuous and mixed dependent features. For high-dimensional settings, our VAEAC approach with a non-uniform masking scheme significantly outperforms competing methods. Finally, we apply our VAEAC approach to estimate Shapley value explanations for the Abalone data set from the UCI Machine Learning Repository.