Download presentation
Presentation is loading. Please wait.
Published byBrian Elliott Modified over 9 years ago
1
Binary Exponential Back Off for Tabu Tenure in Hyperheuristics Speaker: 郭益銘 C. Cotta and P. Cowling (Eds.): EvoCOP 2009, LNCS 5482, pp. 109–120, 2009. © Springer-Verlag Berlin Heidelberg 2009
2
Binary Exponential Back off (BEBO) 二元指數後退演算法 隨機延遲時間 (Random Delay Time) – 其延遲時間的長短由一個亂數決定,以便儘可 能使所有參與衝撞的工作站的延遲時間不相同, 以避免因為延遲時間相同而造成的再次衝撞 – CSMA/CD 通訊協定
3
When a collision occurs, the computer will increase its backoff value by 1, and wait a random amount of time between 0 and 2 backoff -1 before trying to retransmit. If the transmission is successful, the back off value is reset to 0, otherwise the back-off value is increased by 1 again and the process is repeated.
4
Workforce Scheduling Problem f = SP - 4SC - 2TT – SP is the sum of the priority of scheduled tasks – SC is the sum of the time window costs in the schedule (both resource and task) – TT is the total amount of travel time. This objective is to maximise the total priority of tasks scheduled while minimising travel time and cost. The weights in this expression are sensible in a large class of the problems encountered in practice.
5
Related Work PapersMethods [13][14] 2001 2003Performance [15] 2003Random [16] 2002(GA) Evolve good sequences of LLH [4] 2005Step by Step Reduction (SSR) [4] 2005Warming up (WU) [5] [19] [20] 2003 2005 2004(Tabu) Fixed Tabu Tenure [5] [18] [20] 2003 1996 2004(Tabu) Random Tabu Tenure [11] 2007Reduced Variable Neighbourhood Search (rVNS) [5] [20] : results provide no clear advantage of using adaptive tabu tenures over fixed ones. [4] : SSR > WU [11] [14] [18] : not a hyperheuristic
6
Hyperheuristic Approaches 1.Task Selector – Selecting a task to be scheduled 2.Resource Allocator – Allocating potential resources (including time) for that task LLHs = TSs * RAs
7
Hyperheuristic Approaches HyperRandom selects at random a LLH (i.e. a (task order, resource selector) pair) to use at each iteration and applies it if the application will result in a positive improvement. This continues until no improvement has been found for a certain number of iterations. HyperGreedy evaluates all the LLH at each iteration and applies the best if it makes an improvement. This continues until no improvement is found. CPU-intensive, Good quality, Inefficient ¼ LLHs were never used ; ½ LLHs were only applied once
8
Pseudocode
10
Deemed Bad Two methods to decide which of the low level heuristics, which were tried, to back off (those “deemed bad”): 1)“Best x” – only the best x improving low level heuristics are not backed off. 2) “Prop x” – all non improving low level heuristics and those improving low level heuristics not in the top x% of the range of the fitness are backed off.
14
Conclusions The results have shown the potential of the new method to generate good solutions much more quickly than exhaustive (greedy) approaches and standard tabu approaches. In principle our exponential back-off methods could be used in any tabu implementation with large neighbourhoods, which provides a promising possible direction for further research.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.