CSWA: Aggregation-Free Spatial-Temporal Community Sensing
Authors: Jiang Bian, Haoyi Xiong, Yanjie Fu, Sajal Das
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Simulation experiments based on real-world datasets demonstrate that CSWA exhibits low approximation error (i.e., less than 0.2 C in city-wide temperature sensing task and 10 units of PM2.5 index in urban air pollution sensing) and performs comparably to (sometimes better than) state-of-the-art algorithms based on the data aggregation and centralized computation. In order to evaluate the CSWA algorithm, we use the Temperature (TEMP) and PM 2.5 air quality (PM25) dataset, where the Experimental Setup section will cover all the settings and assumptions. |
| Researcher Affiliation | Academia | Jiang Bian, Haoyi Xiong, Yanjie Fu, Sajal K. Das Missouri University of Science and Technology, United States |
| Pseudocode | Yes | Algorithm 1: Initializing Batch and Matrix Factors ( ˆP, ˆQ) on Organizer. Algorithm 2: Parallelized Optimization on the jth Participant. Algorithm 3: Mobile Sensing Recovery on the Organizer. |
| Open Source Code | No | The paper does not provide any explicit statements about making its source code open, nor does it provide a link to a code repository for the described methodology. |
| Open Datasets | Yes | For TEMP (Ingelrest et al. 2010) and PM25 (Zheng, Liu, and Hsieh 2013) datasets, the sensing value of temperature ( C)/PM2.5 (air quality index) are located on each participant s mobile sensor in varying time slots (sensing cycle) and at different subareas. |
| Dataset Splits | No | The paper mentions using "real-world datasets" for evaluation and discusses various experimental settings and comparisons. However, it does not specify explicit training, validation, or test dataset splits (e.g., percentages, sample counts, or cross-validation setup) that would allow for reproduction of data partitioning. |
| Hardware Specification | No | The paper does not specify any hardware details (e.g., GPU/CPU models, memory, or specific computing environments like cloud instances) used for running the experiments. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers (e.g., programming languages, libraries, or frameworks like Python, PyTorch, or TensorFlow with their versions). |
| Experiment Setup | Yes | Experimental Setup For TEMP (Ingelrest et al. 2010) and PM25 (Zheng, Liu, and Hsieh 2013) datasets, the sensing value of temperature ( C)/PM2.5 (air quality index) are located on each participant s mobile sensor in varying time slots (sensing cycle) and at different subareas. In details, the TEMP dataset contains the temperature readings in 57 cells (Subareas) and each sensing cycle lasts for 30 minutes. The PM25 dataset includes the PM2.5 air quality values on 36 stations (Subareas) with the same sensing cycle. In order to simulate the settings of the centralized computing patterns, we aggregate the collected sensing data from each participant. In details, we follow the aforementioned three phases to set the appropriate value of four key factors: the Number of Participants (m), the Number of Subareas that each participant covers in each sensing cycle, the Size of Windows (w) and the Size of Latent Space (l). Note that each participant can sense the temperature/PM2.5 at a subset of subarea. Specifically, we use the maximum number of subareas s (1 s |S|) in the experiments, assuming the participant can cover no more than s subareas. To simulate the scenario that each participant can cover various number of subareas, the actual number of subareas covered by the participant will follow the discrete uniform distribution U{1, s}. |