We describe multi-objective influence diagrams, based on a set of p objectives, where utility val... more We describe multi-objective influence diagrams, based on a set of p objectives, where utility values are vectors in Rp, and are typically only partially ordered. These can still be solved by a variable elimination algorithm, leading to a set of maximal values of expected utility. If the Pareto ordering is used this set can often be prohibitively large. We consider approximate representations of the Pareto set based on e-coverings, allowing much larger problems to be solved. In addition, we define a method for incorporating user tradeoffs, which also greatly improves the efficiency.
Proceedings of the 15th International Conference on Principles and Practice of Constraint Programming, 2009
Multi-objective optimization is concerned with problems involving multiple measures of performanc... more Multi-objective optimization is concerned with problems involving multiple measures of performance which should be optimized simultaneously. In this paper, we extend AND/OR Branch-and-Bound (AOBB), a well known search algorithm, from mono-objective to multi-objective optimization. The new algorithm MO-AOBB exploits efficiently the problem structure by traversing an AND/OR search tree and uses static and dynamic mini-bucket heuristics to guide the search. We show that MO-AOBB improves dramatically over the traditional OR search approach, on various benchmarks for multi-objective optimization.
Multi-objective optimization is concerned with problems involving multiple measures of performanc... more Multi-objective optimization is concerned with problems involving multiple measures of performance which should be optimized simultaneously. In this paper, we extend AND/OR Branch-and-Bound (AOBB), a well known search algorithm, from mono-objective to multi-objective optimization. The new algorithm MO-AOBB exploits efficiently the problem structure by traversing an AND/OR search tree and uses static and dynamic mini-bucket heuristics to guide the search. We show that MO-AOBB improves dramatically over the traditional OR search approach, on various benchmarks for multi-objective optimization.
In this paper, we propose new depth-first heuristic search algorithms to approximate the set of P... more In this paper, we propose new depth-first heuristic search algorithms to approximate the set of Pareto optimal solutions in multi-objective constraint optimization. Our approach builds upon recent advances in multi-objective heuristic search over weighted AND/OR search spaces and uses an ǫ-dominance relation between cost vectors to significantly reduce the set of non-dominated solutions. Our empirical evaluation on various benchmarks demonstrates the power of our scheme which improves the resolution times dramatically over recent stateof-the-art competitive approaches.
Research and Development in Intelligent Systems XXVI, 2009
ABSTRACT Influence diagrams are a widely used framework for decision making under uncertainty. Th... more ABSTRACT Influence diagrams are a widely used framework for decision making under uncertainty. The paper presents a new algorithm for maximizing the expected utility over a set of policies by traversing an AND/OR search space associated with an influence diagram. AND/OR search spaces accommodate advanced algorithmic schemes for graphical models which can exploit the structure of the problem. The algorithm also exploits the deterministic information encoded by the influence diagram and avoids redundant computations for infeasible decision choices. We demonstrate empirically the effectiveness of the AND/OR search approach on various benchmarks for influence diagrams.
This paper presents an axiomatic framework for influence di-agram computation, which allows reaso... more This paper presents an axiomatic framework for influence di-agram computation, which allows reasoning with partially or-dered values of utility. We show how an algorithm based on sequential variable elimination can be used to compute the set of maximal values of expected utility (up to an equivalence relation). Formalisms subsumed by the framework include decision making under uncertainty based on multi-objective utility, or on interval-valued utilities, as well as a more quali-tative decision theory based on order-of-magnitude probabil-ities and utilities.
ABSTRACT AND/OR search spaces have recently been introduced as a unifying paradigm for advanced a... more ABSTRACT AND/OR search spaces have recently been introduced as a unifying paradigm for advanced algorithmic schemes for graphical models. The main virtue of this representation is its sensitivity to the structure of the model, which can translate into exponential time savings for search algorithms. In this paper we extend the recently introduced AND/OR Branch-and-Bound algorithm [1] for solving 0/1 Mixed Integer Linear Programming problems. We propose a static version based on pseudo-trees, as well as a dynamic one based on hypergraph separators. Preliminary evaluation on problem instances from MIPLIB2003 shows promise that the new schemes are likely to improve over the traditional methods.
AND/OR search spaces are a unifying paradigm for advanced algorithmic schemes for graphical model... more AND/OR search spaces are a unifying paradigm for advanced algorithmic schemes for graphical models. The main virtue of this representation is its sensitivity to the structure of the model, which can translate into exponential time savings for search algorithms. In this paper we introduce an AND/OR search algorithm that explores a context-minimal AND/OR search graph in a best-first manner for solving 0/1 Integer Linear Programs (0/1 ILP). We also extend to the 0/1 ILP domain the depth-first AND/OR Branch-and-Bound search with caching algorithm which was recently proposed by [1] for solving optimization tasks in graphical models. The effectiveness of the best-first AND/OR search approach compared to the depth-first AND/OR Branch-and-Bound search is demonstrated on a variety of benchmarks for 0/1 ILPs, including instances from the MIPLIB library, real-world combinatorial auctions, random uncapacitated warehouse location problems and MAX-SAT instances. several extensions of AOBB t that incorporate dynamic variable ordering heuristics and explore dynamic AND/OR search trees. Two such extensions, AND/OR Branch-and-Bound with Partial Variable Ordering (AOBB t +PVO) and AND/OR Branch-and-Bound with Full Dynamic Variable Ordering (AOBB t +DVO) were shown to outperform significantly the static AOBB t algorithm as well as state-of-the-art classic OR Branch-and-Bound algorithms on various domains, including 0/1 ILPs.
AND/OR search spaces have recently been introduced as a unifying paradigm for advanced algorithmi... more AND/OR search spaces have recently been introduced as a unifying paradigm for advanced algorithmic schemes for graphical models. The main virtue of this representation is its sensitivity to the structure of the model, which can translate into exponential time savings for search algorithms. In this paper we extend the recently introduced AND/OR Branch-and-Bound algorithm (AOBB) [1] for solving pure 0/1 Integer Linear Programs . Since the variable selection can have a dramatic impact on search performance, we introduce a new dynamic AND/OR Branch-and-Bound algorithm able to accommodate variable ordering heuristics. The effectiveness of the dynamic AND/OR approach is demonstrated on a variety of benchmarks for pure 0/1 integer programming, including instances from the MIPLIB library, real-world combinatorial auctions and random uncapacitated warehouse location problems.
The paper presents the results of a feasibility study in need to determine if viable, effective a... more The paper presents the results of a feasibility study in need to determine if viable, effective and reliable web based early warning service in the domain of drinking water utilities can be developed. The early warning service is intended to be used by many actors in the Republic of Ireland including water utilities managers, the Irish Environmental Protection Agency personnel and the Irish Health and Safety Executive personnel. The results of this feasibility study include an assessment of whether or not a statement of requirements for a novel early warning service can be realised using a set of existing technologies. Based on these results it is concluded that it is feasible to develop a reliable and effective web-based early warning service for drinking water utilities by adapting technologies derived from multiple disciplines including safety engineering, artificial intelligence, software and web engineering.
The notion of refactoring —transforming the source- code of an object-oriented program without ch... more The notion of refactoring —transforming the source- code of an object-oriented program without changing its external behaviour — has been studied intensively within the last decade. This diversity has created a plethora of toy-examples, cases and code snippets, which make it hard to assess the current state-of-the-art. Moreover, due to this diversity, there is currently no accepted way of teaching
Uploads
Papers by Radu Marinescu