Principles of Explanatory Debugging to Personalize Interactive Machine Learning

How can end users efficiently influence the predictions that machine learning systems make on their behalf? This paper presents Explanatory Debugging, an approach in which the system explains to users how it made each of its predictions, and the user then explains any necessary corrections back to the learning system. We present the principles underlying this approach and a prototype instantiating it. An empirical evaluation shows that Explanatory Debugging increased participants' understanding of the learning system by 52% and allowed participants to correct its mistakes up to twice as efficiently as participants using a traditional learning system.

[1]  Anind K. Dey,et al.  Assessing demand for intelligibility in context-aware applications , 2009, UbiComp.

[2]  M. Rowe,et al.  Teaching science as continuous inquiry , 1973 .

[3]  Jerry Alan Fails,et al.  Interactive machine learning , 2003, IUI '03.

[4]  S. Hart,et al.  Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research , 1988 .

[5]  Geoff Holmes,et al.  Multinomial Naive Bayes for Text Categorization Revisited , 2004, Australian Conference on Artificial Intelligence.

[6]  Brad A. Myers,et al.  Debugging reinvented , 2008, 2008 ACM/IEEE 30th International Conference on Software Engineering.

[7]  Desney S. Tan,et al.  Interactive optimization for steering machine classification , 2010, CHI.

[8]  Tobias Höllerer,et al.  TasteWeights: a visual interactive hybrid recommender system , 2012, RecSys.

[9]  Desney S. Tan,et al.  EnsembleMatrix: interactive visualization to support machine learning with multiple classifiers , 2009, CHI.

[10]  Evie McCrum-Gardner,et al.  Which is the correct statistical test to use? , 2008, The British journal of oral & maxillofacial surgery.

[11]  Anind K. Dey,et al.  Improving Understanding and Trust with Intelligibility in Context-Aware Applications , 2012 .

[12]  Ashok K. Goel,et al.  GAIA: A CAD Environment for Model-Based Adaptation of Game-Playing Software Agents , 2013, CSER.

[13]  Nava Tintarev,et al.  Evaluating the effectiveness of explanations for recommender systems , 2012, User Modeling and User-Adapted Interaction.

[14]  Perry R. Cook,et al.  Human model evaluation in interactive supervised learning , 2011, CHI.

[15]  Anind K. Dey,et al.  Why and why not explanations improve the intelligibility of context-aware intelligent systems , 2009, CHI.

[16]  Eugene Creswick,et al.  Strategies and behaviors of end-user programmers with interactive fault localization , 2003, IEEE Symposium on Human Centric Computing Languages and Environments, 2003. Proceedings. 2003.

[17]  Weng-Keen Wong,et al.  Explanatory Debugging: Supporting End-User Debugging of Machine-Learned Programs , 2010, VL/HCC.

[18]  Alex Groce,et al.  You Are the Only Possible Oracle: Effective Test Selection for End Users of Interactive Machine Learning Systems , 2014, IEEE Transactions on Software Engineering.

[19]  Burr Settles,et al.  Active Learning Literature Survey , 2009 .

[20]  John Riedl,et al.  Navigating the tag genome , 2011, IUI '11.

[21]  Mark W. Newman,et al.  Learning from a learning thermostat: lessons for intelligent systems for the home , 2013, UbiComp.

[22]  Lora Aroyo,et al.  The effects of transparency on trust in and acceptance of a content-based art recommender , 2008, User Modeling and User-Adapted Interaction.

[23]  Weng-Keen Wong,et al.  Why-oriented end-user debugging of naive Bayes text classification , 2011, ACM Trans. Interact. Intell. Syst..

[24]  Maya Cakmak,et al.  Designing Interactions for Robot Active Learners , 2010, IEEE Transactions on Autonomous Mental Development.

[25]  Weng-Keen Wong,et al.  Too much, too little, or just right? Ways explanations impact end users' mental models , 2013, 2013 IEEE Symposium on Visual Languages and Human Centric Computing.

[26]  Peter Stone,et al.  Reinforcement learning from human reward: Discounting in episodic tasks , 2012, 2012 IEEE RO-MAN: The 21st IEEE International Symposium on Robot and Human Interactive Communication.

[27]  Colin Potts,et al.  Design of Everyday Things , 1988 .

[28]  Andrea Bunt,et al.  Are explanations always important?: a study of deployed, low-cost intelligent interactive systems , 2012, IUI '12.

[29]  Donald A. Norman,et al.  Some observations on mental models , 1987 .

[30]  Maya Cakmak,et al.  Power to the People: The Role of Humans in Interactive Machine Learning , 2014, AI Mag..

[31]  Regina A. Pomranky,et al.  The role of trust in automation reliance , 2003, Int. J. Hum. Comput. Stud..

[32]  C. Breazeal,et al.  Transparency and Socially Guided Machine Learning , 2006 .

[33]  Desney S. Tan,et al.  Examining multiple potential models in end-user interactive concept learning , 2010, CHI.

[34]  John Riedl,et al.  Explaining collaborative filtering recommendations , 2000, CSCW '00.

[35]  Jude W. Shavlik,et al.  Using neural networks for data mining , 1997, Future Gener. Comput. Syst..

[36]  D. Norman.  The Design of Everyday Things: Revised and Expanded Edition , 2013 .

[37]  Joe Tullio,et al.  How it works: a field study of non-technical users interacting with an intelligent system , 2007, CHI.

[38]  Carmen Lacave,et al.  A review of explanation methods for Bayesian networks , 2002, The Knowledge Engineering Review.

[39]  P. Johnson-Laird,et al.  Mental Models: Towards a Cognitive Science of Language, Inference, and Consciousness , 1985 .

[40]  Gautham J. Mysore,et al.  ISSE: an interactive source separation editor , 2014, CHI.

[41]  Willett Kempton,et al.  Two Theories of Home Heat Control , 1986, Cogn. Sci..

[42]  John M. Carroll,et al.  Principles and Heuristics for Designing Minimalist Instruction , 1998 .

[43]  Yiming Yang,et al.  A Comparative Study on Feature Selection in Text Categorization , 1997, ICML.

[44]  Alan F. Blackwell,et al.  First steps in programming: a rationale for attention investment models , 2002, Proceedings IEEE 2002 Symposia on Human Centric Computing Languages and Environments.

[45]  Thomas G. Dietterich,et al.  Interacting meaningfully with machine learning systems: Three experiments , 2009, Int. J. Hum. Comput. Stud..

[46]  Todd Kulesza,et al.  Tell me more?: the effects of mental model soundness on personalizing an intelligent agent , 2012, CHI.

[47]  Desney S. Tan,et al.  CueFlik: interactive concept learning in image search , 2008, CHI.

[48]  Zhiyong Lu,et al.  Explaining Naive Bayes Classifications , 2003 .