Strongly-Typed Agents are Guaranteed to Interact Safely

As artificial agents proliferate, it is becoming increasingly important to ensure that their interactions with one another are well-behaved. In this paper, we formalize a common-sense notion of when algorithms are well-behaved: an algorithm is safe if it does no harm. Motivated by recent progress in deep learning, we focus on the specific case where agents update their actions according to gradient descent. The paper shows that that gradient descent converges to a Nash equilibrium in safe games. The main contribution is to define strongly-typed agents and show they are guaranteed to interact safely, thereby providing sufficient conditions to guarantee safe interactions. A series of examples show that strong-typing generalizes certain key features of convexity, is closely related to blind source separation, and introduces a new perspective on classical multilinear games based on tensor decomposition.

[1]  Sergey Levine,et al.  Learning deep control policies for autonomous aerial vehicles with MPC-guided policy search , 2015, 2016 IEEE International Conference on Robotics and Automation (ICRA).

[2]  Yousef Saad,et al.  On the Tensor SVD and the Optimal Low Rank Orthogonal Approximation of Tensors , 2008, SIAM J. Matrix Anal. Appl..

[3]  Joachim M. Buhmann,et al.  Correlated random features for fast semi-supervised learning , 2013, NIPS.

[4]  Seungjin Choi,et al.  Independent Component Analysis , 2009, Handbook of Natural Computing.

[5]  Karthik Sridharan,et al.  Optimization, Learning, and Games with Predictable Sequences , 2013, NIPS.

[6]  Joachim M. Buhmann,et al.  Kickback Cuts Backprop's Red-Tape: Biologically Plausible Credit Assignment in Neural Networks , 2014, AAAI.

[7]  Paul W. Goldberg,et al.  The complexity of computing a Nash equilibrium , 2006, STOC '06.

[8]  David D. Cox,et al.  Tensor Switching Networks , 2016, NIPS.

[9]  Guigang Zhang,et al.  Deep Learning , 2016, Int. J. Semantic Comput..

[10]  Gene H. Golub,et al.  Rank-One Approximation to High Order Tensors , 2001, SIAM J. Matrix Anal. Appl..

[11]  Demis Hassabis,et al.  Mastering the game of Go with deep neural networks and tree search , 2016, Nature.

[12]  Chrisantha Fernando,et al.  PathNet: Evolution Channels Gradient Descent in Super Neural Networks , 2017, ArXiv.

[13]  E. Rowland.  Theory of Games and Economic Behavior , 1946, Nature.

[14]  L. Shapley,et al.  Potential Games , 1994 .

[15]  Muhammad Ghifary,et al.  Strongly-Typed Recurrent Neural Networks , 2016, ICML.

[16]  David Balduzzi,et al.  Grammars for Games: A Gradient-Based, Game-Theoretic Framework for Optimization in Deep Learning , 2016, Front. Robot. AI.

[17]  Surya Ganguli,et al.  Identifying and attacking the saddle point problem in high-dimensional non-convex optimization , 2014, NIPS.

[18]  Tara N. Sainath,et al.  FUNDAMENTAL TECHNOLOGIES IN MODERN SPEECH RECOGNITION Digital Object Identifier 10.1109/MSP.2012.2205597 , 2012 .

[19]  Shun-ichi Amari,et al.  Information Geometry and Its Applications: Convex Function and Dually Flat Manifold , 2009, ETVC.

[20]  Shane Legg,et al.  Human-level control through deep reinforcement learning , 2015, Nature.

[21]  Francis Crick,et al.  The recent excitement about neural networks , 1989, Nature.

[22]  Ian J. Goodfellow,et al.  NIPS 2016 Tutorial: Generative Adversarial Networks , 2016, ArXiv.

[23]  Joos Vandewalle,et al.  A Multilinear Singular Value Decomposition , 2000, SIAM J. Matrix Anal. Appl..

[24]  Haipeng Luo,et al.  Fast Convergence of Regularized Learning in Games , 2015, NIPS.

[25]  Huda Khayrallah,et al.  Deep Generalized Canonical Correlation Analysis , 2017, RepL4NLP@ACL.

[26]  Sébastien Bubeck,et al.  Convex Optimization: Algorithms and Complexity , 2014, Found. Trends Mach. Learn..

[27]  Andreas Krause,et al.  Safe Exploration in Finite Markov Decision Processes with Gaussian Processes , 2016, NIPS.

[28]  Sham M. Kakade,et al.  Multi-view Regression Via Canonical Correlation Analysis , 2007, COLT.

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

[30]  Luca Cardelli.  Type systems , 1996, CSUR.

[31]  Andreas Krause,et al.  Safe learning of regions of attraction for uncertain, nonlinear systems with Gaussian processes , 2016, 2016 IEEE 55th Conference on Decision and Control (CDC).

[32]  Konrad P. Kording,et al.  Towards an integration of deep learning and neuroscience , 2016, bioRxiv.

[33]  Yoshua Bengio,et al.  Deep Learning of Representations: Looking Forward , 2013, SLSP.

[34]  J. Nash.  Equilibrium Points in N-Person Games. , 1950, Proceedings of the National Academy of Sciences of the United States of America.

[35]  John Schulman,et al.  Concrete Problems in AI Safety , 2016, ArXiv.

[36]  Tim Roughgarden,et al.  Algorithmic Game Theory , 2007 .

[37]  Sayan Mukherjee,et al.  The Information Geometry of Mirror Descent , 2013, IEEE Transactions on Information Theory.

[38]  Shun-ichi Amari,et al.  Natural Gradient Works Efficiently in Learning , 1998, Neural Computation.