Collective Multiagent Sequential Decision Making Under Uncertainty
Authors: Duc Thien Nguyen, Akshat Kumar, Hoong Chuin Lau
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comparisons with previous best approaches on synthetic instances and a real world taxi dataset modeling supply-demand matching show that our approach significantly outperforms them w.r.t.solution quality. |
| Researcher Affiliation | Academia | Duc Thien Nguyen, Akshat Kumar, Hoong Chuin Lau School of Information Systems Singapore Management University |
| Pseudocode | Yes | Algorithm 1: FEM: Collective Sampling based Fictitious EM |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | Yes | We next test on the large scale real-world taxi problem described in section 2, introduced previously in (Varakantham et al. 2012). The dataset contains the actual movement traces of 8000 taxis roaming in Singapore divided into 81 zones as shown in figure 2 for one year. |
| Dataset Splits | No | The paper does not explicitly provide training/validation/test splits, only mentioning the use of a real-world dataset and synthetic instances without specifying division percentages or counts for each split. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used to run the experiments. |
| Software Dependencies | No | The paper does not list any specific software dependencies with version numbers, such as programming languages or libraries. |
| Experiment Setup | Yes | For EM, we compute both the closed loop and open loop policies. As previous approaches (FP-SAP, SMFU, MIP) are based on average flow approximation, they cannot compute closed loop policies. Each data point is an average of 10 instances. As our policy evaluation is based on sampling, we also report 95%-confidence intervals over 200 samples. For each approach, iteration limit was 500, with convergence occurring much earlier within a time limit of 0.5 hour; MIP had 2 hour limit. For closed loop EM, we use a piecewise policy with 5 pieces. |