Metric Learning for Value Alignment

Preference are central to decision making by both machines and humans. Representing, learning, and reasoning with preferences is an important area of study both within computer science and across the social sciences. When we give our preferences to an AI system we expect the system to make decisions or recommendations that are consistent with our preferences but the decisions should also adhere to certain norms, guidelines, and ethical principles. Hence, when working with preferences it is necessary to understand and compute a metric (distance) between preferences – especially if we encode both the user preferences and ethical systems in the same formalism. In this paper we investigate the use of CP-nets as a formalism for representing orderings over actions for AI systems. We leverage a recently proposed metric for CP-nets and a neural network architecture, CPMETRIC, for computing this metric. Using these two tools we look at the how one can build a fast and flexible value alignment system.

[1]  Eyke Hüllermeier,et al.  Preferences in AI: An overview , 2011, Artif. Intell..

[2]  Francesca Rossi,et al.  Reasoning with PCP-nets in a Multi-Agent Context , 2015, AAMAS.

[3]  Nicholas Mattei,et al.  Uniform Random Generation and Dominance Testing for CP-Nets , 2017, J. Artif. Intell. Res..

[4]  Michael I. Jordan,et al.  Distance Metric Learning with Application to Clustering with Side-Information , 2002, NIPS.

[5]  M. Kendall.  A NEW MEASURE OF RANK CORRELATION , 1938 .

[6]  Athman Bouguettaya,et al.  A CP-Net Based Qualitative Composition Approach for an IaaS Provider , 2018, WISE.

[7]  Nicholas Mattei,et al.  A behavioral perspective on social choice , 2012, Annals of Mathematics and Artificial Intelligence.

[8]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[9]  Stuart J. Russell,et al.  Research Priorities for Robust and Beneficial Artificial Intelligence , 2015, AI Mag..

[10]  Anna Papst,et al.  Fast and Slow , 2008, Science.

[11]  Henri Prade,et al.  Graphical Models for Preference Representation: An Overview , 2016, SUM.

[12]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[13]  Craig Boutilier,et al.  CP-nets: a tool for represent-ing and reasoning with conditional ceteris paribus state-ments , 2004 .

[14]  Andrea Loreggia,et al.  Value Alignment via Tractable Preference Distance , 2018, Artificial Intelligence Safety and Security.

[15]  Mark D. McDonnell,et al.  Understanding Data Augmentation for Classification: When to Warp? , 2016, 2016 International Conference on Digital Image Computing: Techniques and Applications (DICTA).

[16]  Athman Bouguettaya,et al.  Web Service Selection with Incomplete or Inconsistent User Preferences , 2009, ICSOC/ServiceWave.

[17]  Li Chen,et al.  Usability Guidelines for Product Recommenders Based on Example Critiquing Research , 2011, Recommender Systems Handbook.

[18]  Francesca Rossi,et al.  Beyond Theory and Data in Preference Modeling: Bringing Humans into the Loop , 2015, ADT.

[19]  Francesca Rossi,et al.  Preferences and Ethical Priorities: Thinking Fast and Slow in AI , 2019, AAMAS.

[20]  Eyke Hüllermeier,et al.  Preference Learning , 2005, Künstliche Intell..

[21]  Francesca Rossi,et al.  CPDist: Deep Siamese Networks for Learning Distances Between Structured Preferences , 2018, ArXiv.

[22]  Ronald Fagin,et al.  Comparing Partial Rankings , 2006, SIAM J. Discret. Math..

[23]  Subbarao Kambhampati.  Synthesizing Explainable Behavior for Human-AI Collaboration , 2019, AAMAS.

[24]  Francesca Rossi,et al.  Updates and Uncertainty in CP-Nets , 2013, Australasian Conference on Artificial Intelligence.

[25]  T. Walsh,et al.  PREFLIB: A Library for Preferences , 2013 .

[26]  Judy Goldsmith,et al.  Preference Handling for Artificial Intelligence , 2008, AI Mag..

[27]  Francesca Rossi,et al.  On the Distance Between CP-nets , 2018, AAMAS.

[28]  Kihyuk Sohn,et al.  Improved Deep Metric Learning with Multi-class N-pair Loss Objective , 2016, NIPS.

[29]  Toby Walsh,et al.  A Short Introduction to Preferences: Between Artificial Intelligence and Social Choice , 2011, A Short Introduction to Preferences.

[30]  Francesca Rossi,et al.  Preferences and Ethical Principles in Decision Making , 2018, AAAI Spring Symposia.

[31]  Miroslaw Truszczynski,et al.  The computational complexity of dominance and consistency in CP-nets , 2005, IJCAI.

[32]  Mike Fitzpatrick.  Choice , 2004, The Lancet.