Background Check: A General Technique to Build More Reliable and Versatile Classifiers

Nieto, Miquel Perelló; Filho, Telmo M. Silva; Kull, Meelis; Flach, Peter A.

doi:10.1109/icdm.2016.0150

Cited by 12 publications

(8 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After all, if a prediction is of poor quality, it would be unreasonable to expect the explanation to be sensible. We suggest that an explanation be accompanied by a list of uncertainty sources, one of which may be the confidence of the predictive model for the instance being explained [39]. For example, if a method relies on synthetic data (as opposed to real data) this should be clearly stated as a source of variability, hence uncertainty.…”

Section: S4 Explanation Alitymentioning

confidence: 99%

Explainability fact sheets

Sokol

Flach

2020

Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency

Self Cite

171

View full text Add to dashboard Cite

Explanations in Machine Learning come in many forms, but a consensus regarding their desired properties is yet to emerge. In this paper we introduce a taxonomy and a set of descriptors that can be used to characterise and systematically assess explainable systems along five key dimensions: functional, operational, usability, safety and validation. In order to design a comprehensive and representative taxonomy and associated descriptors we surveyed the eXplainable Artificial Intelligence literature, extracting the criteria and desiderata that other authors have proposed or implicitly used in their research. The survey includes papers introducing new explainability algorithms to see what criteria are used to guide their development and how these algorithms are evaluated, as well as papers proposing such criteria from both computer science and social science perspectives. This novel framework allows to systematically compare and contrast explainability approaches, not just to better understand their capabilities but also to identify discrepancies between their theoretical qualities and properties of their implementations. We developed an operationalisation of the framework in the form of Explainability Fact Sheets, which enable researchers and practitioners alike to quickly grasp capabilities and limitations of a particular explainable method. When used as a Work Sheet, our taxonomy can guide the development of new explainability approaches by aiding in their critical evaluation along the five proposed dimensions.

show abstract

Section: S4 Explanation Alitymentioning

confidence: 99%

Explainability fact sheets

Sokol

Flach

2020

Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency

Self Cite

171

View full text Add to dashboard Cite

show abstract

“…As a third method, we use the Background Check technique proposed by Perello-Nieto e.a. [22]. This is an integrated methodology based on an explicit background class b, that they use to update the posterior distribution of the classifier as well as to detect ambiguous and novel cases.…”

Section: Background Check Methodsmentioning

confidence: 99%

“…Perhaps most related to our setting is Background Check [22] and we will therefore use it in our experiments. It concerns a technique that does not depend on a specific type of base classifier.…”

Section: Related Workmentioning

confidence: 99%

Probability of default estimation, with a reject option

Coenen

Abdullah

Guns

2020

2020 IEEE 7th International Conference on Data Science and Advanced Analytics (DSAA)

View full text Add to dashboard Cite

Many companies, such as credit granting companies, have to decide on granting or denying customer or invoice loans on a daily basis. Increasingly, machine learning is used to learn probability-of-default models from previously granted cases and, thus, whether the outcome was positive or negative for the company, i.e. whether the client paid back or defaulted. However, as the outcome can only be observed for the granted cases, the data inherently has sample selection bias and caution should be taken when applying the probability-of-default model to the full through-the-door population. In reject inference, this problem is studied with respect to whether using the unlabeled rejected instances can help improve a classifier that is only trained on granted instances, e.g. using semi-supervised learning. In contrast, we investigate under what circumstances a model trained on the granted instances, with known outcome, can be used on all possible instances. For this, we believe a model should indicate when it cannot reliably predict the outcome. That is, it should refrain from making predictions on instances unlike those on which it was trained. If not, the credit granting company would expose itself to great risk, and experts could lose their trust in the predictions. We discuss similarities and differences of this problem compared to novelty detection, classification with a reject option and reject inference. We compare a number of methods that combine novelty detection with classification, with decent results even for two-stage methods and especially when using data of existing instances with unknown outcome.

show abstract

“…Hence, the standard approach is to calibrate the transformation such that γ percent of the examples have a probability > 0.5. Beyond the logistic calibration approach, there is a long literature of approaches [5,7,8,10,11,17] for ensuring that this property is obtained and we now briefly describe some prominent approaches. Isotonic Calibration [20] is a non-parametric form of regression in which the transformation function is chosen from the class of all non-decreasing functions.…”

Section: From Anomaly Scores To Outlier Probabilitiesmentioning

confidence: 99%

Quantifying the Confidence of Anomaly Detectors in Their Example-Wise Predictions

Perini

Vercruyssen

Davis

2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Anomaly detection focuses on identifying examples in the data that somehow deviate from what is expected or typical. Algorithms for this task usually assign a score to each example that represents how anomalous the example is. Then, a threshold on the scores turns them into concrete predictions. However, each algorithm uses a different approach to assign the scores, which makes them difficult to interpret and can quickly erode a user's trust in the predictions. This paper introduces an approach for assessing the reliability of any anomaly detector's example-wise predictions. To do so, we propose a Bayesian approach for converting anomaly scores to probability estimates. This enables the anomaly detector to assign a confidence score to each prediction which captures its uncertainty in that prediction. We theoretically analyze the convergence behaviour of our confidence estimate. Empirically, we demonstrate the effectiveness of the framework in quantifying a detector's confidence in its predictions on a large benchmark of datasets.

show abstract

Background Check: A General Technique to Build More Reliable and Versatile Classifiers

Cited by 12 publications

References 16 publications

Explainability fact sheets

Explainability fact sheets

Probability of default estimation, with a reject option

Quantifying the Confidence of Anomaly Detectors in Their Example-Wise Predictions

Contact Info

Product

Resources

About