Explaining Preference Learning Alyssa Glass CS229 Final Project Computer Science Department, Stanford University  Augment PLIANT to gather additional meta-information about the SVM itself:  Support vectors identified by SVM  Support vectors nearest to the query point  Margin to the query point  Average margin over all data points  Non-support vectors nearest to the query point  Kernel transform used, if any  Represent SVM learning and meta-information as justification in Proof Markup Language (PML), adding SVM rules as needed.  Design abstraction strategies for presenting justification to user as a similarity-based explanation. (Work on PML representation and abstraction strategies is on-going; details will be in final report.) Active Preference Learning in PLIANT (Yorke-Smith et al. 2007) Motivation Studies of users interacting with systems that learn preferences show that, when the system behaves incorrectly, users quickly lose patience and trust in the system. Even when the system is correct, users view such outcomes as “magical” in some way, but are unable to understand why a particular suggestion is correct, or whether the system is likely to be helpful in the future. We describe the augmentation of a preference learner to provide meaningful feedback to the user through explanations. This work extends the PLIANT (Preference Learning through Interactive Advisable Nonintrusive Training) SVM-based preference learner, part of the PTIME personalized scheduling assistant in the CALO project. Acknowledgements We thank Melinda Gervasio, Pauline Berry, Neil Yorke-Smith, and Bart Peintner for access to the PLIANT and PTIME systems, the above architecture picture, and for helpful collaborations, partnerships, and feedback on this work. We also thank Deborah McGuinness, Michael Wolverton, and Paulo Pinheiro da Silva for the IW and PML systems, and for related discussions and previous work that helped to lay the foundation for this effort. We thank Mark Gondek for access to the CALO CLP data, and Karen Myers for related discussions, support, and ideas. Usability and Active Learning Providing Transparency into Preference Learning Select References  PLIANT: Yorke-Smith, N., Peintner, B., Gervasio, M., and Berry, P. M. Balancing the Needs of Personalization and Reasoning in a User- Centric Scheduling Assistant. Conference on Intelligent User Interfaces 2007 (IUI-07) (to appear).  PTIME: Berry, P., Gervasio, M., Uribe, T., Pollack, M., and Moffitt, M. A Personalized Time Management Assistant. AAAI Spring Symposium Series, Stanford, CA, March  Partial preference updates: Joachims, T. Optimizing Search Engines using Clickthrough Data. Proceedings of the ACM Conference on Knowledge Discovery and Data Mining (KDD), ACM,  User study on explaining statistical machine learning methods: Stumpf, S., Rajaram, V., Li, L., Burnett, M., Dietterich, T., Sullivan, E., Drummond, R., and Herlocker, J. Towards Harnessing User Feedback for Machine Learning. Conference on Intelligent User Interfaces 2007 (IUI-07) (to appear). System Workflow 1. Elicit initial preferences from user ( A vector from above) 2. User specifies new meeting parameters 3. Constraint solver generates candidate schedules ( Z’s ) 4. Candidate schedules ranked using evaluation function, F`(Z) 5. Candidate schedules presented to user in (roughly) the calculated preference order, with explanations for each one 6. User can ask questions, then chooses a schedule ( Z ) 7. Preferences ( a i and a ij weights) are updated based on choice Features: 1. Scheduling windows for requested meeting 2. Duration of meeting 3. Overlaps and conflicts 4. Location of meeting 5. Participants in meeting 6. Preferences of other meeting participants Model of preferences: aggregation function, a 2-order Choquet integral over partial utility functions based on the above features  learning 21 coefficient weights: F(z 1, …, z n ) =  i a i z i +  i,j a ij (z i  z j ) where each z i = u i (x i ), the utility for criterion i based on value x i Evaluation function: combine learned weights with initial elicited preferences: F`(Z) =   A  Z + (1-  )  W  Z Each schedule chosen by the user provides information about a partial preference ordering, as in (Joachims 2002). Architecture Several user studies show that transparency is key to trusting learning systems:  Our trust study  Lack of understanding of update gives appearance that preferences are ignored  seems untrustworthy  Typical user reaction: “I trust [the system’s] accuracy, but not its judgment.”  PTIME user study (Yorke-Smith et al. 2007) “The preference model must be explainable to the user … in terms of familiar, domain-relevant concepts.”  Explaining statistical ML methods (Stumpf et al. 2007)  Looked at explaining naïve Bayes learner and rule- learning system (classification), not SVMs  Rule-based explanations most easily understood, but similarity-based explanations found to be more natural and easily trusted Our approach: extend similarity-based explanations to SVM learning explanation SVM meta- information selected schedule solution set Calendar Manager Constraint Reasoner preference profile PLIANT scheduling request presentation set SVM Explainer current profile 4 5