Correlative Channel-Aware Fusion for Multi-View Time Series Classification
Authors: Yue Bai, Lichen Wang, Zhiqiang Tao, Sheng Li, Yun Fu6714-6722
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on three real-world datasets demonstrate the superiority of our C2AF over the state-of-the-art methods. A detailed ablation study is also provided to illustrate the indispensability of each model component. |
| Researcher Affiliation | Academia | 1 Department of Electrical and Computer Engineering, Northeastern University, Boston, USA 2 Department of Computer Science and Engineering, Santa Clara University, Santa Clara, USA 3 Department of Computer Science, University of Georgia, Athens, USA |
| Pseudocode | Yes | Algorithm 1 The procedure of training C2AF algorithm. |
| Open Source Code | No | Code will be released at https://github.com/yueb17/C2AF |
| Open Datasets | Yes | EV-Action (Wang et al. 2019) is a multi-view human action dataset... NTU RGB+D (Shahroudy et al. 2016) is a large-scale dataset for multi-view action recognition... UCI Daily and Sports Activities (Asuncion and Newman 2007) is a multivariate time series dataset... |
| Dataset Splits | Yes | We choose the first 40 subjects for training and the rest 13 subjects for test. (EV-Action)... We use the cross-subject benchmark provided by the original dataset paper, which contains 40320 samples for training and 16560 samples for test. (NTU RGB+D) |
| Hardware Specification | No | Our model is implemented using Tensorflow with GPU acceleration. (No specific GPU model or other hardware details are provided.) |
| Software Dependencies | No | Our model is implemented using Tensorflow with GPU acceleration. (Tensorflow is mentioned, but no specific version number or other software dependencies with versions are listed.) |
| Experiment Setup | Yes | We set 128 as batch size. The Adam optimizer (Kingma and Ba 2014) is utilized for optimization and the learning rates are set as 0.0001 for all the view-specific and final classifiers synchronously. |