Principles of Explanatory Debugging to Personalize Interactive Machine Learning

By Todd Kulesza, Margaret M. Burnett, Weng-Keen Wong, Simone Stumpf. 2015

692 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Safety Research, Software, Safety, Robustness & Interpretability, Human-Computer Interaction

How researchers see this paper

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

In Artificial Intelligence

In Safety Research

In Software

In Safety, Robustness & Interpretability

In Human-Computer Interaction

Citations in the corpus

Refining Language Models with Compositional Explanations (2021, methodology): “A more interpretable direction is to associate features with specific labels as a source of supervision [32], or let users suggest adjustments of low-level features by examining model explanations [24, 43, 26].”
Explanation in Human-AI Systems: A Literature Meta-Review, Synopsis of Key Ideas and Publications, and Bibliography for Explainable AI (2019, methodology): “…of researchers have focused more specifically on explanations that might be presented to non-developer users of the systems (Hendricks et al., 2016; Kulesza et al, 2012; 2015; Lomas et al., 2012; Park et al., 2016; Ribiero et al., 2016; Rosenthal et al., 2016) This third generation is typified by…”
Explainable AI: The New 42? (2018, background): “First steps have been taken towards making these systems understandable to their users, by providing textual and visual explanations [13,22] (see Figs.”
What do we need to build explainable AI systems for the medical domain? (2017, background): “…emphasize that successful explainable-AI systems need eﬀective user interfaces, fostering new strategies for presenting human understandable explanations, e.g. explanatory debugging (Kulesza et al., 2015), aﬀective computing (Picard, 1997), sentiment analysis (Maas et al., 2011; Petz et al., 2015).”
Explanation in Artificial Intelligence: Insights from the Social Sciences (2017, background): “Already, projects of this type are underway, with impressive results; for example, see [91, 89, 156].”
Making deep neural networks right for the right scientific reasons by interacting with their explanations (2020, background): “This correction step is crucial for more directly affecting the learner’s beliefs and is integral to modulating trust [6, 12].”
Improving Fairness in Machine Learning Systems: What Do Industry Practitioners Need? (2018, background): “The design, prototyping, and maintenance of machine learning systems raises many unique challenges [27, 63, 78, 87] not commonly faced with other kinds of intelligent systems or computing systems more broadly [34, 66, 67].”
The challenge of crafting intelligible intelligence (2018, positive): “Here, we expect to build on tools for interactive machine learning [1] and explanatory debugging [20, 21], which have explored interactions for adding new training examples, correcting erroneous labels in existing data, specifying new features, and modifying shape functions.”
Explaining Explanations in AI (2018, background): “…to comply with relevant legislation (Doshi-Velez et al., 2017), verify and improve the functionality of a system (i.e. as a type of ‘debugging’; (Kulesza et al., 2015)) and help developers and humans working with a system learn from it (Samek et al., 2017), and enhance the trust between…”
Evaluating XAI: A comparison of rule-based and example-based explanations (2021, background): “Other purposes include support in system debugging [16,12], verification [13] and justification [17].”
Explanatory Interactive Machine Learning (2019, background): “This correction step is crucial for more directly affecting the learner’s beliefs and is integral to modulating trust [14, 18].”
Counterfactual Explanations of Machine Learning Predictions: Opportunities and Challenges for AI Safety (2019, methodology): “Given their concise and easily understandable format they are a useful tool to identify bugs and errors in the underlying predictive model (Kulesza et al. 2015).”
Explanation-Based Human Debugging of NLP Models: A Survey (2021, background): “, 2009), topic classification (Kulesza et al., 2015; Teso and Kersting, 2019), spam classification (Koh and Liang, 2017), and auto-coding of transcripts (Kulesza et al.”
Intelligible Artificial Intelligence (2018, methodology): “Here we expect to use affordances from interactive machine learning [4] and explanatory debugging [21].”
Explainability fact sheets: a framework for systematic assessment of explainable approaches (2019, background): “Finally, explanations should be selective and succinct enough to avoid overwhelming the explainee with unnecessary information, i.e., ﬁll in the most gaps with the fewest arguments [26, 33–35 , 54 ].”
Software Engineering for AI-Based Systems: A Survey (2021, background): “Similarly, two studies are concerned with providing specialized approaches or tool support for the debugging of ML software by focusing on explanatory debugging in interactive ML [113], and proposing their Tensorflow debugger based on dataflow graphs [34].”
Questioning the AI: Informing Design Practices for Explainable AI User Experiences (2020, background): “[4] shows, before this wave of work on ML systems, the HCI community have studied explainable systems in various contexts, most notably context-aware applications [12, 55, 56], recommender systems [41], debugging tools [52] and algorithmic transparency [27, 79].”