Machine Learning: Between Accuracy and Interpretability

By Ivan Bratko. 1997

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Scalable Bayesian Rule Lists (2016, background): “Interpretability has long since been a fundamental topic in artiﬁcial intelligence (see R¨uping, 2006; Bratko, 1997; Dawes, 1979; Vellido, Mart´ın-Guerrero, & Lisboa, 2012; Giraud-Carrier, 1998; Holte, 1993; Shmueli, 2010; Huysmans, Dejaeger, Mues, Vanthienen, & Baesens, 2011; Freitas, 2014).”
Supersparse linear integer models for optimized medical scoring systems (2015, background): “Many medical scoring systems that are currently in use were hand-crafted by physicians, whereby a panel of experts simply agreed on a model (see e.g., the CHADS2 score of Gage et al., 2001).”
Methods and Models for Interpretable Linear Classification (2014, methodology): “easures, such as an L 1-norm), or the way they train their models (i.e. by using heuristic procedures, such as rounding). Approximations result in a poor trade-o between accuracy and interpretability [17, 34, 93]. Linear classiers that minimize surrogate loss functions, for instance, are not robust ”
An Interpretable Stroke Prediction Model using Rules and Bayesian Analysis (2013, unknown): “Call the rule˜r i .”
Explainable Artificial Intelligence: a Systematic Review (2020, methodology): “Interactivity The capacity of an explanation system to reason about previous utterances both to interpret and answer users’ follow-up questions [80, 81] Interestingness The capacity of a method for explainability to facilitate the discovery of novel knowledge and to engage user’s attention [64, 82, ”
Supersparse Linear Integer Models for Interpretable Classification (2013, background): “Even so, Bratko[10] points out that these shorter trees are often unnatural and unintuitive, even when their measured accuracy is higher than domain experts’ accuracy.”
Asking 'Why' in AI: Explainability of intelligent systems - perspectives and challenges (2018, background): “Early 1990s work on knowledge discovery also points to the role in background domain knowledge as an input to a ML sys-temtosupportexplanationgeneration(Bratko,1997).”
Machine Learning for 5G/B5G Mobile and Wireless Communications: Potential, Limitations, and Future Directions (2019, background): “level of abstraction to understand its functionality [180].”
Interpretability of machine learning‐based prediction models in healthcare (2020, unknown): “Traditionally, the focus in ML research was on global interpretability in an effort to help users understand the relationship between all possible inputs to an ML model and the space of all predictions made by the model (Bratko, 1997; Martens, Huysmans, Setiono, Vanthienen, & Baesens, 2008).”
Algorithmic Accountability and Public Reason (2017, background): “Considering the wide range of machine learning methods available, there are trade-offs to be made between interpretability and accuracy (Bratko, 1997; Ribeiro, Singh, & Guestrin, 2016).”
Learning interpretable models (2006, unknown): “On the other hand, it has also been reported that in the presence of prior knowledge experts said that shorter trees are “unnatural” and do not “tell enough”, despite the fact that the measured accuracy of the trees was significantly higher than the experts accuracy [Bratko, 1996].”