https://www.ics.uci.edu/~dechter/publications/r76A.pdf

Bucket elimination: a unifying framework for reasoning

1. Programs. 2. Functions and Categories. 3. Applications. 4. Relationships and Allegories. 5. Datatypes in Allegories. 6. Optimisation Problems. 7. Thinning Algorithms. 8. Dynamic Programming. 9. Greedy Algorithms. Appendices.

Algebra of programming

Contraction mappings in the theory underlying dynamic programming

Reinforcement learning (RL) is a powerful paradigm for sequential decision-making under uncertainties, and most RL algorithms aim to maximize some numerical value which represents only one long-term objective. However, multiple long-term objectives are exhibited in many real-world decision and control systems, so recently there has been growing interest in solving multiobjective reinforcement learning (MORL) problems where there are multiple conflicting objectives. The aim of this paper is to present a comprehensive overview of MORL. The basic architecture, research topics, and naive solutions of MORL are introduced at first. Then, several representative MORL approaches and some important directions of recent research are comprehensively reviewed. The relationships between MORL and other related research are also discussed, which include multiobjective optimization, hierarchical RL, and multiagent RL. Moreover, research challenges and open problems of MORL techniques are suggested.

Multiobjective Reinforcement Learning: A Comprehensive Overview

Ve cOllf:iider multi- criteria f:iequent,ial decision making problems where the vcctor-"valucd evaluations arc compared by a given, fixed total ordering. Condit.ions for the opt.irnality of statiOIl<-l,r}' p()lichs ;-weI the Bellman opti­ malit,y equatio n a re given for a. speci al, but. important class of problems ''v hell the eval­ uation of policies can be computed for the criteria, independently of each other. The anal)'sis requires special cafC as t.he t.opol­ ag,Y int.roduced by polnL\visc convergence a.ncl the or<1cr- topology introduced by the prefer­ ence order arc in general incompatible. Re­ inf orcement. learning algorithms are proposed and analY7,ed.

/pdf/multi-criteria-reinforcement-learning-5fjmdz8350.pdf

Multi-criteria Reinforcement Learning

Any multistage process may be constructed by a sequence of serial and/or parallel compositions. With certain restrictions on the form of the return composition function, optimal processes may be constructed in an efficient manner from optimal subprocesses. A number of simple sufficient conditions on the return functions are given and several approximation procedures are outlined.

Composition Principles for Synthesis of Optimal Multistage Processes

This paper shows how the methods of dynamic programming may be applied to the problem of optimal design and operation of countercurrent flow processes. The physical relations involved are summarized by a set of transformations describing the technology of the process. The dynamic programming recursive maximum return functions are then used to generate a set of optimal technologies. The stages above and below the feed stage are treated as independent dynamic programming problems that are then coupled through the feed stage technology. The optimal design of a complex of chemical plants is used as an illustrative example.

L. G. Mitten

Papers

Composition Principles for Synthesis of Optimal Multistage Processes

Application of Dynamic Programming to Countercurrent Flow Processes