Varieties of Justification in Machine Learning

Corfield, David

doi:10.1007/s11023-010-9191-1

Cited by 10 publications

(12 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the law of large numbers, we know that the expected standard deviation of this randomly drawn subset is proportional to 1∕ √ . Definition (15) amounts to the well-known student's t test 10 and it can be converted into a p value 11 . Even though Definition ( 15) is rather intuitive and has good statistical properties, it is possible to opt for different versions of strengh in Algocate (e.g.…”

Section: Strength Relationmentioning

confidence: 99%

“…Another distinction is made by Klass and Finin [16] based on the intention, which should be to "produce knowledge in the hearer" for explanations and "to affect the beliefs of the hearer" for justifications. From a different perspective, [10], introduces a classification of justifications in machine learning related to the performances (accuracy) of the systems. A series of works [7,24,28] refer to justifications as ways of ensuring that a decision is good (in contrast to understanding a decision), which is in line with the approach followed in this paper and definitions of explanations and justifications in philosophy [3].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

A framework to contest and justify algorithmic decisions

Henin

Métayer

2021

AI Ethics

View full text Add to dashboard Cite

In this paper, we argue that the possibility of contesting the results of Algorithmic Decision Systems (ADS) is a key requirement for ADS used to make decisions with a high impact on individuals. We discuss the limitations of explanations and motivate the need for better facilities to contest or justify the results of an ADS. While the goal of an explanation is to make it possible for a human being to understand, the goal of a justification is to convince that the decision is good or appropriate. To claim that a result is good, it is necessary (1) to refer to an independent definition of what a good result is (the norm) and(2) to provide evidence that the norm applies to the case. Based on these definitions, we present a challenge and justification framework including three types of norms, a proof-of-concept implementation of this framework and its application to a credit decision system.

show abstract

Section: Strength Relationmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

A framework to contest and justify algorithmic decisions

Henin

Métayer

2021

AI Ethics

View full text Add to dashboard Cite

show abstract

“…A different approach aims at justifying the output of a DLN by supplementing it with 31 Justification in this sense is not to be equated with mathematical criteria or measures of the performance of learning machines. Different varieties of such measures are discussed in an interesting paper by Corfield (2010). additional information intended to strengthen the user's confidence.…”

Section: Mckeown Arguementioning

confidence: 99%

Judging machines: philosophical aspects of deep learning

Schubbach

2019

Synthese

View full text Add to dashboard Cite

“…There are only few works attempting to provide an epistemological treatment of statistical learning theory [cf. Harman and Kulkarni 2007;Corfield et al 2009;Corfield 2010;von Luxburg and Schölkopf 2011;Spelda 2018], which, moreover, captures only a part of the story. Second, the recent history of ML has been dominated by an empiricist practice that derives estimates of performance from the a posteriori evaluation.…”

Section: Introductionmentioning

confidence: 99%

Human Induction in Machine Learning

Spelda

Střítecký

2021

ACM Comput. Surv.

View full text Add to dashboard Cite

As our epistemic ambitions grow, the common and scientific endeavours are becoming increasingly dependent on Machine Learning (ML). The field rests on a single experimental paradigm, which consists of splitting the available data into a training and testing set and using the latter to measure how well the trained ML model generalises to unseen samples. If the model reaches acceptable accuracy, then an a posteriori contract comes into effect between humans and the model, supposedly allowing its deployment to target environments. Yet the latter part of the contract depends on human inductive predictions or generalisations, which infer a uniformity between the trained ML model and the targets. The article asks how we justify the contract between human and machine learning. It is argued that the justification becomes a pressing issue when we use ML to reach “elsewhere” in space and time or deploy ML models in non-benign environments. The article argues that the only viable version of the contract can be based on optimality (instead of on reliability, which cannot be justified without circularity) and aligns this position with Schurz's optimality justification. It is shown that when dealing with inaccessible/unstable ground-truths (“elsewhere” and non-benign targets), the optimality justification undergoes a slight change, which should reflect critically on our epistemic ambitions. Therefore, the study of ML robustness should involve not only heuristics that lead to acceptable accuracies on testing sets. The justification of human inductive predictions or generalisations about the uniformity between ML models and targets should be included as well. Without it, the assumptions about inductive risk minimisation in ML are not addressed in full.

show abstract

Varieties of Justification in Machine Learning

Cited by 10 publications

References 5 publications

A framework to contest and justify algorithmic decisions

A framework to contest and justify algorithmic decisions

Judging machines: philosophical aspects of deep learning

Human Induction in Machine Learning

Contact Info

Product

Resources

About