What Makes Planners Predictable?
نویسندگان
چکیده
In recent work we showed that models constructed from planner performance data over a large suite of benchmark problems are surprisingly accurate; 91-99% accuracy for success and 3-496 seconds RMSE for runtime. In this paper, we examine the underlying causes of these accurate models. We deconstruct the learned models to assess how the features, the planners, the search space topology and the amount of training data facilitate predicting planner performance. We find that the models can be learned from relatively little training data (e.g., performance on 10% of the problems in some cases). Generally, having more features improves accuracy. However, the effect is often planner-dependent: in some cases, adding features degrades performance. We identify that the most prominent features in the models are domain features, though we find that the runtime models still have a need for better features. In the last part of the paper, we examine explanatory models to refine the planner dependencies and to identify linkages between problem structure and specific planners’ performance.
منابع مشابه
تردید در وجود تلقی جهانشمول از عقلانیت در رویههای شهرسازی
There is a deep-seated belief among planners that the historical association between urban planning and rationality is an unquestioned presumption which they follow their daily practices based on. It seems that truth of urban planning claims has been shown predominantly by applying the concept of rationality. As a result, reflection on rationality can be taken into consideration as an importan...
متن کاملOptimization of Cycle Time and Utilization in Semiconductor Test Manufacturing Using Simulation Based, On-line, Near-real-time Scheduling System
A discrete event simulation based “on-line near-real-time” dynamic scheduling and optimization system has been conceptualized, designed, and developed to optimize cycle time and asset utilization in the complex manufacturing environment of semiconductor test manufacturing. Our approach includes the application of rules and optimization algorithm, using multiple variables as an integral part of ...
متن کاملOn the performance comparison of multi-objective evolutionary UAV path planners
The big number of evolutionary planners for Unmanned Aerial Vehicles (UAV) that have been developed demonstrates the good acceptance that the evolutionary techniques enjoy within the UAV community. However, the minor or nonexistent statistical characterization of the results obtained by the majority of the planners not only makes it difficult to assess their actual performance but also to justi...
متن کاملChanging Roles and Skills of Urban Planners in the Turkish Planning System
This research aims to find an answer to the question of which knowledge and skills do the Turkish urban planners need in their business practice. Understanding change in cities, making a prediction, making an urban decision and putting it into practice, working together with actors from different organizations from various academic disciplines, persuading people to accept something and developi...
متن کامل