Explainable AI in Credit Risk Management

Bussmann, Niklas; Giudici, Paolo; Marinelli, Dimitri; Papenbrock, Jochen

doi:10.2139/ssrn.3506274

Cited by 10 publications

(4 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This mini review provides a simplified, yet substantive, discussion of key definitions and considerations for using ML within the United States lending context. While questions remain as to which methods will be most useful for ensuring compliance with regulatory requirements, variants of constrained models, Shapley values, and counterfactual explanations appear to be gaining some momentum in the broader lending community (Bracke et al, 2019;Bussman et al, 2019). From the fair lending perspective, there are well-established discrimination testing and mitigation methodologies that have been used for decades.…”

Section: Discussionmentioning

confidence: 99%

A United States Fair Lending Perspective on Machine Learning

Hall

Cox

Dickerson³

et al. 2021

Front. Artif. Intell.

View full text Add to dashboard Cite

The use of machine learning (ML) has become more widespread in many areas of consumer financial services, including credit underwriting and pricing of loans. ML’s ability to automatically learn nonlinearities and interactions in training data is perceived to facilitate faster and more accurate credit decisions, and ML is now a viable challenger to traditional credit modeling methodologies. In this mini review, we further the discussion of ML in consumer finance by proposing uniform definitions of key ML and legal concepts related to discrimination and interpretability. We use the United States legal and regulatory environment as a foundation to add critical context to the broader discussion of relevant, substantial, and novel ML methodologies in credit underwriting, and we review numerous strategies to mitigate the many potential adverse implications of ML in consumer finance.

show abstract

Section: Discussionmentioning

confidence: 99%

A United States Fair Lending Perspective on Machine Learning

Hall

Cox

Dickerson³

et al. 2021

Front. Artif. Intell.

View full text Add to dashboard Cite

show abstract

“…Taking specific model examples, GAMs are relatively easy and transparent to understand and are used for risk assessments in financial applications [43][44][45]. The authors in [46] use traditional XGboost and Logistic Regression (LR), with LR principally used for comparison purposes. After training the model, the Shapley values [33] from the testing set of the companies are calculated.…”

Section: Xai In Imagesmentioning

confidence: 99%

Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation

et al. 2021

View full text Add to dashboard Cite

Trust and credibility in machine learning models are bolstered by the ability of a model to explain its decisions. While explainability of deep learning models is a well-known challenge, a further challenge is clarity of the explanation itself for relevant stakeholders of the model. Layer-wise Relevance Propagation (LRP), an established explainability technique developed for deep models in computer vision, provides intuitive human-readable heat maps of input images. We present the novel application of LRP with tabular datasets containing mixed data (categorical and numerical) using a deep neural network (1D-CNN), for Credit Card Fraud detection and Telecom Customer Churn prediction use cases. We show how LRP is more effective than traditional explainability concepts of Local Interpretable Model-agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP) for explainability. This effectiveness is both local to a sample level and holistic over the whole testing set. We also discuss the significant computational time advantage of LRP (1–2 s) over LIME (22 s) and SHAP (108 s) on the same laptop, and thus its potential for real time application scenarios. In addition, our validation of LRP has highlighted features for enhancing model performance, thus opening up a new area of research of using XAI as an approach for feature subset selection.

show abstract

“…Figure 17: The predictor Figure 18: The decoder Before we proceed to the models' design, we have to apply some feature engineering in the input data. We firstly discarded some features, like the third operational setting and some sensors [1,5,6,10,16,18,19,22,23,24,25,26], because they did not contain any information (all values were NaN, or they had the exact same value). Then, inspired by a recent work [63], we removed the two additional operating settings.…”

Section: Turbofan Engine Degradation Simulation Dataset -Time-seriesmentioning

confidence: 99%

“…It is a promising field that aims to address several key socio-economic and ethical issues that machine learning systems may create [1]. For example, interpretable machine learning systems can assist underwriters in the insurance and banking sectors [5], and provide explanations as to why someone's right to free speech has been trampled by an automated process of a social network [6]. In addition, interpretable machine learning is the key to transform efficient machine learning procedures, such as predictive maintenance [7], into more descriptive and reliable ones, like prescriptive maintenance [8].…”

Section: Introductionmentioning

confidence: 99%

LioNets: A Neural-Specific Local Interpretation Technique Exploiting Penultimate Layer Information

Mollas,

Bassiliades,

Tsoumakas

2021

Preprint

View full text Add to dashboard Cite

Artificial Intelligence (AI) has a tremendous impact on the unexpected growth of technology in almost every aspect. AI-powered systems are monitoring and deciding about sensitive economic and societal issues. The future is towards automation, and it must not be prevented. However, this is a conflicting viewpoint for a lot of people, due to the fear of uncontrollable AI systems. This concern could be reasonable if it was originating from considerations associated with social issues, like gender-biased, or obscure decision-making systems. Explainable AI (XAI) is recently treated as a huge step towards reliable systems, enhancing the trust of people to AI. Interpretable machine learning (IML), a subfield of XAI, is also an urgent topic of research. This paper presents a small but significant contribution to the IML community, focusing on a local-based, neural-specific interpretation process applied to textual and time-series data. The proposed methodology introduces new approaches to the presentation of feature importance based interpretations, as well as the production of counterfactual words on textual datasets. Eventually, an improved evaluation metric is introduced for the assessment of interpretation techniques, which supports an extensive set of qualitative and quantitative experiments.

show abstract

Explainable AI in Credit Risk Management

Cited by 10 publications

References 9 publications

A United States Fair Lending Perspective on Machine Learning

A United States Fair Lending Perspective on Machine Learning

Explaining Deep Learning Models for Tabular Data Using Layer-Wise Relevance Propagation

LioNets: A Neural-Specific Local Interpretation Technique Exploiting Penultimate Layer Information

Contact Info

Product

Resources

About