Abstract:In this paper we suggest NICE: a new algorithm to generate counterfactual explanations for heterogeneous tabular data. The design of our algorithm specifically takes into account algorithmic requirements that often emerge in real-life deployments: the ability to provide an explanation for all predictions, being efficient in run-time, and being able to handle any classification model (also nondifferentiable ones). More specifically, our approach exploits information from a nearest instance to speed up the searc… Show more
“…Case-based Reasoning (CBR) [7,8] and optimisation techniques [12,13] have been the pillars of discovering counterfactuals. Recent work in CBR has shown how counterfactual case generation can be conveniently supported through the case adaptation stage, where query-retrieval pairs of successful counterfactual explanation experiences are used to create an explanation case-base [7].…”
Section: Related Workmentioning
confidence: 99%
“…DiCE is a generative algorithm that discovers counterfactuals by optimising a randomly initialised input to maximise diversity and minimise sparsity and proximity [12]. NICE is a NUN-based counterfactual discovery algorithm that uses a reward function to minimise sparsity, proximity and to preserve plausibility [8].…”
Section: A Comparison Of Counterfactual Discovery Algorithmsmentioning
confidence: 99%
“…The prevailing CBR approach to counterfactual discovery harnesses similarities to Nearest-unlike Neighbours (NUN), i.e. similar cases with different class labels (see Figure 1) [7,8]. A NUN represents potential changes to the current problem, with feature attribution prioritising the changes that, when actioned, can lead to a different outcome [8,9].…”
Section: Introductionmentioning
confidence: 99%
“…similar cases with different class labels (see Figure 1) [7,8]. A NUN represents potential changes to the current problem, with feature attribution prioritising the changes that, when actioned, can lead to a different outcome [8,9]. Focusing on a small number of key "actionable" features is more desirable from a practical standpoint, and has the benefit of reducing the recipient's cognitive burden for understanding the counterfactual.…”
Counterfactual explanations describe how an outcome can be changed to a more desirable one. In XAI, counterfactuals are "actionable" explanations that help users to understand how model decisions can be changed by adapting features of an input. A case-based approach to counterfactual discovery harnesses Nearest-unlike Neighbours as the basis to identify the minimal adaptations needed for outcome change. This paper presents the DisCERN algorithm which uses the query, its NUN and substitution-based adaptation operations to create a counterfactual explanation case. DisCERN uses Integrated Gradients (IntG) feature attribution as adaptation knowledge to order substitution operations and to bring about the desired outcome with as few changes as possible. We present our novel approach with IntG where the NUN is used as the baseline against which the feature attributions are calculated. DisCERN also uses feature attributions to identify a NUN closer to the query, and thereby minimise the total change needed, but results suggest that the number of feature changes can increase. Overall, DisCERN outperforms other counterfactual algorithms such as DiCE and NICE in generating valid counterfactuals with fewer adaptations.
“…Case-based Reasoning (CBR) [7,8] and optimisation techniques [12,13] have been the pillars of discovering counterfactuals. Recent work in CBR has shown how counterfactual case generation can be conveniently supported through the case adaptation stage, where query-retrieval pairs of successful counterfactual explanation experiences are used to create an explanation case-base [7].…”
Section: Related Workmentioning
confidence: 99%
“…DiCE is a generative algorithm that discovers counterfactuals by optimising a randomly initialised input to maximise diversity and minimise sparsity and proximity [12]. NICE is a NUN-based counterfactual discovery algorithm that uses a reward function to minimise sparsity, proximity and to preserve plausibility [8].…”
Section: A Comparison Of Counterfactual Discovery Algorithmsmentioning
confidence: 99%
“…The prevailing CBR approach to counterfactual discovery harnesses similarities to Nearest-unlike Neighbours (NUN), i.e. similar cases with different class labels (see Figure 1) [7,8]. A NUN represents potential changes to the current problem, with feature attribution prioritising the changes that, when actioned, can lead to a different outcome [8,9].…”
Section: Introductionmentioning
confidence: 99%
“…similar cases with different class labels (see Figure 1) [7,8]. A NUN represents potential changes to the current problem, with feature attribution prioritising the changes that, when actioned, can lead to a different outcome [8,9]. Focusing on a small number of key "actionable" features is more desirable from a practical standpoint, and has the benefit of reducing the recipient's cognitive burden for understanding the counterfactual.…”
Counterfactual explanations describe how an outcome can be changed to a more desirable one. In XAI, counterfactuals are "actionable" explanations that help users to understand how model decisions can be changed by adapting features of an input. A case-based approach to counterfactual discovery harnesses Nearest-unlike Neighbours as the basis to identify the minimal adaptations needed for outcome change. This paper presents the DisCERN algorithm which uses the query, its NUN and substitution-based adaptation operations to create a counterfactual explanation case. DisCERN uses Integrated Gradients (IntG) feature attribution as adaptation knowledge to order substitution operations and to bring about the desired outcome with as few changes as possible. We present our novel approach with IntG where the NUN is used as the baseline against which the feature attributions are calculated. DisCERN also uses feature attributions to identify a NUN closer to the query, and thereby minimise the total change needed, but results suggest that the number of feature changes can increase. Overall, DisCERN outperforms other counterfactual algorithms such as DiCE and NICE in generating valid counterfactuals with fewer adaptations.
“…Note that in spite of this reasoning, we did also compare the results found with our counterfactual explanation method with the results when using NICE (Brughmans & Martens, 2021) as counterfactual explanation method. We see that in general the same patterns are found, i.e.…”
This paper studies how counterfactual explanations can be used to assess the fairness of a model. Using machine learning for high-stakes decisions is a threat to fairness as these models can amplify bias present in the dataset, and there is no consensus on a universal metric to detect this. The appropriate metric and method to tackle the bias in a dataset will be case-dependent, and it requires insight into the nature of the bias first. We aim to provide this insight by integrating explainable AI (XAI) research with the fairness domain. More specifically, apart from being able to use (Predictive) Counterfactual Explanations to detect
explicit bias
when the model is directly using the sensitive attribute, we show that it can also be used to detect
implicit bias
when the model does not use the sensitive attribute directly but does use other correlated attributes leading to a substantial disadvantage for a protected group. We call this metric
PreCoF
, or Predictive Counterfactual Fairness. Our experimental results show that our metric succeeds in detecting occurrences of
implicit bias
in the model by assessing which attributes are more present in the explanations of the protected group compared to the unprotected group. These results could help policymakers decide on whether this discrimination is
justified
or not.
Machine learning plays a role in many deployed decision systems, often in ways that are difficult or impossible to understand by human stakeholders. Explaining, in a human-understandable way, the relationship between the input and output of machine learning models is essential to the development of trustworthy machine learning based systems. A burgeoning body of research seeks to define the goals and methods of
explainability
in machine learning. In this paper, we seek to review and categorize research on
counterfactual explanations
, a specific class of explanation that provides a link between what could have happened had input to a model been changed in a particular way. Modern approaches to counterfactual explainability in machine learning draw connections to the established legal doctrine in many countries, making them appealing to fielded systems in high-impact areas such as finance and healthcare. Thus, we design a rubric with desirable properties of counterfactual explanation algorithms and comprehensively evaluate all currently proposed algorithms against that rubric. Our rubric provides easy comparison and comprehension of the advantages and disadvantages of different approaches and serves as an introduction to major research themes in this field. We also identify gaps and discuss promising research directions in the space of counterfactual explainability.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.