Dynamic Resource Routing using Real-Time Dynamic Programming
Authors: Sebastian Schmoll, Matthias Schubert
IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our evaluation was executed on the following system: Linux 4.4.0-109-generic x86 64 GNU/Linux, 8x Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz, 32GB DIMM DDR3 Synchron 1600 MHz. In order to evaluate the quality of the MEWT upper bound, we compare it to the DS-MPI upper bound from [Mc Mahan et al., 2005]. The experimental setting uses a sub-graph of Melbourne with 2148 nodes and 4057 edges and five randomly chosen resources. Figure 2 plots the percentage of states that have at most the respective value on the x-axis. The utility values for U , MEWT and DS-MPI are visualized. |
| Researcher Affiliation | Academia | Sebastian Schmoll, Matthias Schubert Ludwig-Maximilians-Universit at, Munich, Germany {schmoll, schubert}@dbs.ifi.lmu.de |
| Pseudocode | Yes | Algorithm 1 Bounded Real-time Dynamic Programming; Algorithm 2 Input: ϵ, probability list r0 of free resources, output: a list of resource state arrays with a probability > 1 ϵ |
| Open Source Code | No | The paper does not contain any explicit statements or links indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | In our evaluation, we demonstrate that our new method outperforms state-of-the-art solutions for solving general MDPs on the road network of Melbourne where public data on the parking spot vacancy is available. The paper mentions using "public data" from the "road network of Melbourne" but does not provide concrete access information such as a specific link, DOI, repository name, or a formal citation with authors and year for this dataset. |
| Dataset Splits | No | The paper does not provide specific details on dataset splits (e.g., percentages, sample counts, or references to predefined splits) for training, validation, or testing. |
| Hardware Specification | Yes | Our evaluation was executed on the following system: Linux 4.4.0-109-generic x86 64 GNU/Linux, 8x Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz, 32GB DIMM DDR3 Synchron 1600 MHz. |
| Software Dependencies | No | The paper mentions the operating system as "Linux 4.4.0-109-generic x86 64 GNU/Linux", but does not provide specific version numbers for any other key software components, libraries, or frameworks used in the experiments. |
| Experiment Setup | Yes | In our experiments we set this value [penalty] to 30s if the agent turns and 0s in all other cases. [...] we set the discount factor γ to one for the rest of this work. [...] The τ hyperparameter can be set to anything greater than 1 and has little effect on the runtime [Mc Mahan et al., 2005]. [...] In other words, we accept a maximum error of ϵ by considering the most likely transitions only while pruning a large amount of very unlikely transitions. (Figures 4, 5, 6 also show evaluation with varying epsilon values, e.g., ϵ = 0.005) |