Game-Theoretic Approach for Non-Cooperative Planning
Authors: Jaume Jordán, Eva Onaindia
AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We perform some experiments and discuss the solutions obtained with our game-theoretical approach, analyzing how the conflicts between the plans determine the strategic behavior of the agents. |
| Researcher Affiliation | Academia | Jaume Jord an and Eva Onaindia Universitat Polit ecnica de Val encia Departamento de Sistemas Inform aticos y Computaci on Camino de Vera s/n. 46022 Valencia, Spain {jjordan,onaindia}@dsic.upv.es |
| Pseudocode | No | The paper describes algorithms and game mechanics but does not include structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper states 'We implemented a program to generation the extensive-form tree and apply the backward induction algorithm' and mentions using 'the Gambit tool (Mc Kelvey, Mc Lennan, and Turocy 2014)' but does not provide concrete access to their own source code. |
| Open Datasets | Yes | For the experiments we used problems of the well-known Zeno-Travel domain from the International Planning Competition (IPC-3)4. ... 4http://ipc.icaps-conference.org/ |
| Dataset Splits | No | The paper describes using problems from the Zeno-Travel domain but does not provide specific train/validation/test dataset splits or cross-validation details. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU/GPU models, memory, or cloud instances) used for running the experiments. |
| Software Dependencies | Yes | The NE in the normal-form game is computed with the tool Gambit (Mc Kelvey, Mc Lennan, and Turocy 2014). ... Gambit: Software tools for game theory, version 13.1.2. |
| Experiment Setup | Yes | Table 1 shows the problems used in these experiments: the set of initial plans of each agent, the number of actions of each plan and its utility. Given a plan π, which earliest plan execution is denoted by ψ0, βi(π) is calculated as follows: βi(π) = n Goals(π) 10 makespan(ψ0), where n Goals(π) represents the number of goals solved by π and makespan(ψ0) represents the minimum duration schedule for π. The utility of a particular schedule ψ Ψπ is a function of βi(π) and the number of time units that the actions of π are delayed in ψ with respect to the earliest plan execution ψ0; in other words, the difference in the makespan of ψ and ψ0. Thus, µi(ψ) = βi(π) delay(ψ), where delay(ψ) is the delay in the makespan of ψ with respect to the makespan of ψ0. |