Price of Transparency in Strategic Machine Learning
By Emrah Akyol, Cédric Langbort, Tamer Basar. 2016
23 citations recorded by the source corpus.
Based on the observation that the transparency of an algorithm comes with a cost for the algorithm designer when the users (data providers) are strategic, this paper studies the impact of strategic intent of the users on the design and performance of transparent ML algorithms. We quantitatively study the {\bf price of transparency} in the context of strategic classification algorithms, by modeling the problem as a nonzero-sum game between the users and the algorithm designer. The cost of having a transparent algorithm is measured by a quantity, named here as price of transparency which is the ratio of the designer cost at the Stackelberg equilibrium, when the algorithm is transparent (which allows users to be strategic) to that of the setting where the algorithm is not transparent.
Subfields: Game & Decision Theory
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Game & Decision Theory
Citations in the corpus
- Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI) (2018, background): “[179] proposed a first attempt to quantitatively analyze the cost of transparency (PoT) in ML algorithms.”
- Interpretable to Whom? A Role-based Model for Analyzing Interpretable Machine Learning Systems (2018, background): “In scenario 3, the lender may be concerned that an applicant could game the system if it is highly interpretable to them — Akyol et al. formally analyzed this kind of strategic behaviour (Akyol et al., 2016).”
- Interpreting Black-Box Models: A Review on Explainable Artificial Intelligence (2023, unknown): “Akyol et al. [183]. initially attempted a quantitative analysis of the cost of transparency (PoT) in ML algorithms.”
- The Disparate Effects of Strategic Manipulation (2018, methodology): “As in previous work in strategic classication, we cast the problem as a Stackelberg game in which the learner moves rst and publishes her classier before candidates best-respond and manipulate their features [1, 5, 7, 14].”