Causal Interpretations of Black-Box Models
نویسندگان
چکیده
منابع مشابه
Causal Interpretations of Black-box Models
Abstract. Starting from the observation that Friedman’s partial dependence plot has exactly the same formula as Pearl’s back-door adjustment, we explore the possibility of extracting causal information from black-box models trained by machine learning algorithms. There are three requirements to make causal interpretations: a model with good predictive performance, some domain knowledge in the f...
متن کاملBlack-box Fault Models
Four black-box fault models are introduced in the paper. The test generation task for the black-box model is more complicated, because possible realizations of the design must be taken into account. However, the time required to generate tests is not very critical factor, because the test generation can be done in parallel with the circuit synthesis process without a prolongation of Time-to-Mar...
متن کاملCausal Interpretations of Probability
scientific theories or hypotheses. Since the range of alternatives is not known in these cases, it seems implausible to construct a collective and relatedly the measure remains undetermined. If one requires probabilities to be predictive, the range of hypotheses to which probabilities should be ascribed is thus rather restricted. 44 We are therefore in the position to assess the plausibility of...
متن کاملA causal framework for explaining the predictions of black-box sequence-to-sequence models
We interpret the predictions of any blackbox structured input-structured output model around a specific input-output pair. Our method returns an “explanation” consisting of groups of input-output tokens that are causally related. These dependencies are inferred by querying the black-box model with perturbed inputs, generating a graph over tokens from the responses, and solving a partitioning pr...
متن کاملApplications of Abduction #3: \black-box" to \gray-box" Models
Military Operations Research makes extensive use of large complex simulation models. These simulation models are often black-boxes; i.e. they are opaque and incomprehensible. A considerable effort is involved in commissioning a new model into service. We characterise this activity as the construction of gray-box approximate causal models of the functional behaviour of black-box simulation softw...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Journal of Business & Economic Statistics
سال: 2019
ISSN: 0735-0015,1537-2707
DOI: 10.1080/07350015.2019.1624293