In this paper, we introduce pilco, a practical, data-efficient model-based policy search method. Pilco reduces model bias, one of the key problems of model-based reinforcement learning, in a principled way. By learning a probabilistic dynamics model and explicitly incorporating model uncertainty into long-term planning, pilco can cope with very little data and facilitates learning from scratch ...

We show that a modified base-stock policy is optimal for multiproduct, infinite-horizon production-inventory systems, where demand for the products is random and the products share a finite resource every period. We characterize the optimal policy for the case of homogeneous products. Because of the difficulty in computing the optimal base-stock levels for the heterogeneous case, we propose a h...

Effective enterprise information security policy management requires review and assessment activities to ensure information security policies are aligned with business goals and objectives. As security policy management involves the elements of policy development process and the security policy as output, the context for security policy assessment requires goal-based metrics for these two eleme...

We consider the problem of off-policy policy selection in reinforcement learning: using historical data generated from running one policy to compare two or more policies. We show that approaches based on importance sampling can be unfair—they can select the worse of two policies more often than not. We give two examples where the unfairness of importance sampling could be practically concerning...

This chapter reviews multi-stage financial planning models, with a focus on practical approaches for optimizing investors’ performance over time. We discuss two major frameworks for constructing financial planning models: 1) policy rule simulation and optimization; and 2) multi-stage stochastic programming. We advocate an integrated approach, in which a stylized stochastic program helps the inv...

Formal foundations for access control policies with both authority delegation and policy composition operators are partial and limited. Correctness guarantees cannot therefore be formally stated and verified for decentralized composite access control systems, such as those based on XACML 3. To address this problem we develop a formal policy language BelLog that can express both delegation and c...

Previous research on attitude accessibility has found that accessible attitudes influence both visual and cognitive attention. Smith, Fazio, and Cejka (1996), for instance, found that categories towards which we have accessible attitudes are more likely to be brought to mind when assessing a related object. Research in our lab (Young & Fazio, in prep) has found evidence that not only are catego...

Several recent operating systems provide system calls that allow an application to explicitly manage the privileges of modules with which the application interacts. Such privilege-aware operating systems allow a programmer to a write a program that satisfies a strong security policy, even when it interacts with untrusted modules. However, it is often non-trivial to rewrite a program to correctl...