The IBaCoP Planning System: Instance-Based Configured Portfolios
Authors: Isabel Cenamor, Tomás de la Rosa, Fernando Fernández
JAIR 2016 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The experimental evaluation shows that the resulting portfolios provide an improvement when compared with non-informed strategies. One of the proposed portfolios was the winner of the Sequential Satisficing Track of the International Planning Competition held in 2014. In this section, we describe the settings of the experimental evaluation and present the results of the planners on the benchmarks used in the IPC-2014, specifically, in the Sequential Satisficing track. |
| Researcher Affiliation | Academia | Isabel Cenamor EMAIL Tomás de la Rosa EMAIL Fernando Fernández EMAIL Departamento de Informática, Universidad Carlos III de Madrid Avda. de la Universidad, 30. Leganés (Madrid). Spain |
| Pseudocode | Yes | Algorithm 1 shows how to use these components to configure the portfolio for a given planning task. Algorithm 1: Algorithm for configuring the portfolio for a particular planning task. |
| Open Source Code | No | The paper discusses the IBACOP Planning System and its components, but there is no explicit statement about making the source code for IBACOP publicly available, nor is a specific repository link provided for the methodology described. References to IPC-14 benchmark and errata links are related to the competition, not the authors' implementation. |
| Open Datasets | Yes | We have included the planning problems available from IPC-2006 onwards. If we do not mention the test set explicitly, it will always refer to the satisficing tracks of the competitions. The included domain and problems are: IPC-2006: openstacks, pathways, rovers, storage, tpp and trucks. IPC-2008: cybersec, elevators, openstacks, pegsol, pipesworld, scanalyzer, sokoban, transport and woodworking. IPC-2011: barman, elevators, floortile, nomystery, visitall, tidybot, openstacks, parcprinter, parking, pegsol, sokoban, scanalyzer, transport and woodworking. Learning track IPC-2008: gold-miner, matching-bw, n-puzzle, parking, thoughful and sokoban. Learning track IPC-2011: barman, blockworld, depots, gripper, parking, rovers satellite, spanner and tpp. |
| Dataset Splits | Yes | The performance of the predictive models was evaluated with a 10-fold cross-validation on a uniform random permutation of all training data. |
| Hardware Specification | Yes | Experiments were run on a cluster with Intel XEON 2.93 Ghz nodes, each with 8 GB of RAM, using Linux Ubuntu 12.04 LTS. |
| Software Dependencies | No | The paper mentions using Weka as an off-the-shelf data-mining tool, and various algorithms (e.g., J48, Rotation Forest, Decision Table) implemented within it, but does not provide specific version numbers for Weka or any other key software libraries or frameworks directly used for implementing the IBACOP methodology. Linux Ubuntu 12.04 LTS is mentioned for the operating system, but not as an ancillary software dependency for the methodology itself. |
| Experiment Setup | Yes | All planners had a cutoff of 1,800 seconds and 4 GB of RAM. For IBACOP configurations requiring feature extraction, this process was limited to 4 GB of RAM (following IPC competition rules) and 300 seconds. IBACOP2 and IBACOP2-B5E used a Best N confidence strategy with N = 5. The performance of the predictive models was evaluated with a 10-fold cross-validation on a uniform random permutation of all training data. |