Yuri Bykov

A time-predefined local search approach to exam timetabling problems

In recent years the processing speed of computers has increased dramatically. This in turn has allowed search algorithms to execute more iterations in a given amount of real-time. Does this necessarily always lead to an improvement in the quality of final solutions? This paper is devoted to the investigation of that question. We present two variants of local search where the search time can be set as an input parameter. These two approaches are: a time-predefined variant of simulated annealing and an adaptation of the “great deluge” method. We present a comprehensive series of experiments which show that these approaches significantly outperform the previous best results (in terms of solution quality) on a range of benchmark exam timetabling problems. Of course, there is a price to pay for such better results: increased execution time. We discuss the impact of this trade-off between quality and execution time. In particular we discuss issues involving the proper estimation of the algorithms execution time and the assessment of its importance.

Examination timetabling using late acceptance hyper-heuristics

A hyperheuristic is a high level problem solving methodology that performs a search over the space generated by a set of low level heuristics. One of the hyperheuristic frameworks is based on a single point search containing two main stages: heuristic selection and move acceptance. Most of the existing move acceptance methods compare a new solution, generated after applying a heuristic, against a current solution in order to decide whether to reject it or replace the current one. Late Acceptance Strategy is presented as a promising local search methodology based on a novel move acceptance mechanism. This method performs a comparison between the new candidate solution and a previous solution that is generated L steps earlier. In this study, the performance of a set of hyper-heuristics utilising different heuristic selection methods combined with the Late Acceptance Strategy are investigated over an examination timetabling problem. The results illustrate the potential of this approach as a hyperheuristic component. The hyper-heuristic formed by combining a random heuristic selection with Late Acceptance Strategy improves on the best results obtained in a previous study.

A Multicriteria Approach to Examination Timetabling

The main aim of this paper is to consider university examination timetabling problems as multicriteria decision problems. A new multicriteria approach to solving such problems is presented. A number of criteria will be defined with respect to a number of exam timetabling constraints. The criteria considered in this research concern room capacities, the proximity of the exams for the students, the order and locations of events, etc. Of course, the criteria have different levels of importance in different situations and for different institutions. The approach that we adopt is divided into two phases. The goal of the first phase is to find high-quality timetables with respect to each criterion separately. In the second phase, trade-offs between criteria values are carried out in order to find a compromised solution with respect to all the criteria simultaneously. This approach involves considering an ideal point in the criteria space which optimises all criteria at once. It is, of course, generally the case that a solution that corresponds to such a point does not exist. The heuristic search of the criteria space starts from the timetables obtained in the first phase with the aim of finding a solution that is as close as possible to this ideal point with respect to a certain defined distance measure. The developed methodology is validated, tested and discussed using real world examination data from various universities.

The Late Acceptance Hill-Climbing Heuristic

This paper introduces a new and very simple search methodology called Late Acceptance Hill-Climbing (LAHC). It is a local search algorithm, which accepts non-improving moves when a candidate cost function is better than it was a number of iterations before. This number appears as a single algorithmic input parameter which determines the total processing time of the search procedure. The properties of this method are experimentally studied in this paper with a range of Travelling Salesman and Exam Timetabling benchmark problems. Also, we present a fair comparison of LAHC with well-known search techniques, which employ a cooling schedule: Simulated Annealing (SA), Threshold Accepting (TA) and the Great Deluge Algorithm (GDA). In addition, we discuss the success of the method in winning an international competition to automatically solve the Magic Square problem. Our experiments have shown that the LAHC approach is simple, easy to implement and yet is an effective search procedure. For most of the studied instances (especially for the large sized ones), its average performance is better than competitor methods. Moreover, the LAHC approach has an additional advantage (in contrast to the above cooling schedule based methods) in its scale independence. We present an example where the rescaling of a cost function in the Travelling Salesman Problem dramatically deteriorates the performance of three cooling schedule based techniques, but has absolutely no influence upon the performance of LAHC.

A multiobjective optimisation technique for exam timetabling based on trajectories

The most common approach to multiobjective examination timetabling is the weighted sum aggregation of all criteria into one cost function and application of some single-objective metaheuristic. However, the translation of user preferences into the weights of criteria is a sophisticated task, which requires experience on the part of the user, especially for problems with a high number of criteria. Moreover, the results produced by this technique are usually substantially scattered. Thus, the outcome of weighted sum algorithms is often far from user expectation.

Search Strategies in Structural Bioinformatics

Optimisation problems pervade structural bioinformatics. In this review, we describe recent work addressing a selection of bioinformatics challenges. We begin with a discussion of research into protein structure comparison, and highlight the utility of Kolmogorov complexity as a measure of structural similarity. We then turn to research into de novo protein structure prediction, in which structures are generated from first principles. In this endeavour, there is a compromise between the detail of the model and the extent to which the conformational space of the protein can be sampled. We discuss some developments in this area, including off-lattice structure prediction using the great deluge algorithm. One strategy to reduce the size of the search space is to restrict the protein chain to sites on a regular lattice. In this context, we highlight the use of memetic algorithms, which combine genetic algorithms with local optimisation, to the study of simple protein models on the two-dimensional square lattice and the face-centred cubic lattice.

A Step Counting Hill Climbing Algorithm applied to University Examination Timetabling

This paper presents a new single-parameter local search heuristic named step counting hill climbing algorithm (SCHC). It is a very simple method in which the current cost serves as an acceptance bound for a number of consecutive steps. This is the only parameter in the method that should be set up by the user. Furthermore, the counting of steps can be organised in different ways; therefore, the proposed method can generate a large number of variants and also extensions. In this paper, we investigate the behaviour of the three basic variants of SCHC on the university exam timetabling problem. Our experiments demonstrate that the proposed method shares the main properties with the late acceptance hill climbing method, namely its convergence time is proportional to the value of its parameter and a non-linear rescaling of a problem does not affect its search performance. However, our new method has two additional advantages: a more flexible acceptance condition and better overall performance. In this study, we compare the new method with late acceptance hill climbing, simulated annealing and great deluge algorithm. The SCHC has shown the strongest performance on the most of our benchmark problems used.

An Adaptive Flex-Deluge Approach to University Exam Timetabling

This paper presents a new methodology for university exam timetabling problems, which draws upon earlier work on the Great Deluge metaheuristic. The new method introduces a “flexible” acceptance condition. Even a simple variant of this technique (with fixed flexibility) outperforms the original Great Deluge algorithm. Moreover, it enables a run-time adaptation of an acceptance condition for each particular move. We investigate the adaptive mechanism where the algorithm accepts the movement of exams in a way that is dependent upon the difficulty of assigning that exam. The overall motivation is to encourage the exploration of a wider region of the search space. We present an analysis of the results of our tests of this technique on two international collections of benchmark exam timetabling problems. We show that 9 of 16 solutions in the first collection and 11 of 12 solutions in the second collection produced by our technique have a higher level of quality than previously published methodologies.