Spatially Regularized Streaming Sensor Selection
Authors: Changsheng Li, Fan Wei, Weishan Dong, Xiangfeng Wang, Junchi Yan, Xiaobin Zhu, Qingshan Liu, Xin Zhang
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on both synthetic and real datasets validate the effectiveness of the proposed method. |
| Researcher Affiliation | Collaboration | IBM Research-China, China, {lcsheng,dongweis,zxin}@cn.ibm.com Department of Mathematics, Stanford University, USA, fanwei@stanford.edu East China Normal University, China, {xfwang, jcyan}@sei.ecnu.edu.cn Beijing Technology and Business University, China, brucezhucas@gmail.com Nanjing University of Information Science and Technology, China, qsliu@nuist.edu.cn |
| Pseudocode | Yes | Algorithm 1 The SRSSS Algorithm |
| Open Source Code | No | The paper does not provide an explicit statement or link for the source code of the described methodology. |
| Open Datasets | Yes | We further perform the proposed algorithm on a real-world dataset derived from the Intel research laboratory at Berkeley. This dataset is popular for testing sensor selection algorithms (Aggarwal, Xie, and Yu 2011)3. It has 54 sensors and 5 days of temperature readings. The readings are sampled every 30 seconds, and so the dataset contains in total 14400 readings from the rest 52 sensors. The dataset is downloaded from http://www.ulb.ac.be/di/labo/datasets.html. |
| Dataset Splits | No | The paper mentions 'regularization parameters for all the methods are tuned by cross-validation on a mini batch dataset. In the experiments, we use the first 100 observation values to tune parameters.' This implies a tuning process but does not specify distinct training, validation, and test splits with percentages or sample counts. |
| Hardware Specification | Yes | We conduct the experiments on a laptop with 2.8 GHz Intel i-5 CPU and 12GB memory by a single thread, and implement the algorithms using MATLAB R2014b 64bit edition. |
| Software Dependencies | No | The paper states 'implement the algorithms using MATLAB R2014b 64bit edition.' While this names a specific software version, it does not list additional ancillary software dependencies (e.g., libraries, packages, or solvers) with their specific version numbers. |
| Experiment Setup | Yes | The parameter maxlap is set to 5 for PMRL. For SRSSS, the forgetting factor μ is set to 0.9 throughout the experiments. For all the methods, the power constraint P is set to 10, and the prediction interval L is set to 5 throughout the experiments, unless otherwise stated. The regularization parameters for all the methods are tuned by cross-validation on a mini batch dataset. In the experiments, we use the first 100 observation values to tune parameters. |