Optimal Planning and Shortcut Learning: An Unfulfilled Promise
نویسندگان
چکیده
An existential optimal landmark is a set of actions, one of which must be used in some optimal plan. Recently, Karpas and Domshlak (2012) introduced a technique for deriving such existential optimal landmarks, which is based on using shortcut rules — rules which take a path, and attempt to find a cheaper path that achieves some of the propositions that the original path achieved. The shortcut rules that were originally used were of a limited form, and only attempted to remove parts of the given path. One would expect that using more sophisticated shortcut rules would result in a more informative heuristic, although possibly at the cost of increased computation time. We show that, somewhat surprisingly, more sophisticated shortcut rules, which are learned online, during search, result in a very small increase in informativeness on IPC benchmarks. Together with the increased computational cost, this leads to a decrease in the number of problems solved, and leaves finding efficient, informative shortcut rules as a standing challenge.
منابع مشابه
Latent Learning in Agents
Various cognitive models have been proposed to determine optimal paths in spatial navigation tasks, some of which demonstrate latent learning in Agents. We view and present the model as a Reinforcement Learning problem by using QLearning and Model-Based Learning in a deterministic environment. This paper uses a QLearning algorithm and compares its performance with the Dyna (Model based) algorit...
متن کاملNear-Minimum-Time Motion Planning of Manipulators along Specified Path
The large amount of computation necessary for obtaining time optimal solution for moving a manipulator on specified path has made it impossible to introduce an on line time optimal control algorithm. Most of this computational burden is due to calculation of switching points. In this paper a learning algorithm is proposed for finding the switching points. The method, which can be used for both ...
متن کاملIS Research Relevance Revisited: Subtle Accomplishment, Unfulfilled Promise, or Serial Hypocrisy?
The topic of “IS research relevance” is receiving increasing attention from the IS research community. In this article, based on a panel discussion at the 2001 International Conference on Information Systems, three arguments are explored in connection with IS research relevance. The first argument, called “subtle accomplishment”, suggests that IS research is relevant, but in a subtle way. The s...
متن کاملResidual Codean Autoencoder for Facial Attribute Analysis
Facial attributes can provide rich ancillary information which can be utilized for different applications such as targeted marketing, human computer interaction, and law enforcement. This research focuses on facial attribute prediction using a novel deep learning formulation, termed as R-Codean autoencoder. The paper first presents Cosine similarity based loss function in an autoencoder which i...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2013