Decision Policy Design as Pareto-Minimization of Infeasible Lower Bounds
نویسندگان
چکیده
A decision policy chooses an outcome dependent on given input parameters. Policies can adequately be represented by production rules, which are at the heart of modern business rule management systems. Classic ways of policy design are rule authoring by experts or learning from data. In this paper, we show that policies can also be derived from a model consisting of constraints and preferences. We can design a policy that respects the given preferences by solving a particular combinatorial Pareto-optimization problem. We consider a combined parameter and decision space and introduce a rule for each Pareto-minimal infeasible lower bound in this space. The approach gives interesting insights in the relationships between combinatorial optimization under preferences and rule-based decision making.
منابع مشابه
Classification Methods with Reject Option Based on Convex Risk Minimization
In this paper, we investigate the problem of binary classification with a reject option in which one can withhold the decision of classifying an observation at a cost lower than that of misclassification. Since the natural loss function is non-convex so that empirical risk minimization easily becomes infeasible, the paper proposes minimizing convex risks based on surrogate convex loss functions...
متن کاملLower Bound Sifting for MDDs
Decision Diagrams (DDs) are a data structure for the representation and manipulation of discrete logic functions often applied in VLSI CAD. Common DDs to represent Boolean functions are Binary Decision Diagrams (BDDs). Multiple-valued logic functions can be represented by Multiple-valued Decision Diagrams (MDDs). The efficiency of a DD representation strongly depends on the variable ordering; t...
متن کاملAssessing Patient Participation in Health Policy Decision-Making in Cyprus
Although the importance of patient participation in the design and evaluation of health programs and services is well-documented, there is scarcity of research with regard to patient association (PA) participation in health policy decision-making processes. To this end, the present study aimed to validate further a previously developed instrument as well as to investigate the degree of PA parti...
متن کاملEfficient Policy Learning
We consider the problem of using observational data to learn treatment assignment policies that satisfy certain constraints specified by a practitioner, such as budget, fairness, or functional form constraints. This problem has previously been studied in economics, statistics, and computer science, and several regret-consistent methods have been proposed. However, several key analytical compone...
متن کاملSelf-Optimizing and Pareto-Optimal Policies in General Environments Based on Bayes-Mixtures
The problem of making sequential decisions in unknown probabilistic environments is studied. In cycle t action yt results in perception xt and reward rt, where all quantities in general may depend on the complete history. The perception xt and reward rt are sampled from the (reactive) environmental probability distribution . This very general setting includes, but is not limited to, (partial ob...
متن کامل