Disentangled Variational Autoencoder based Multi-Label Classification with Covariance-Aware Multivariate Probit Model
Authors: Junwen Bai, Shufeng Kong, Carla Gomes
IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We thoroughly test MPVAE with multiple public datasets on a variety of metrics. MPVAE outperforms (or is comparable to) other state-of-the-art multi-label prediction models. MPVAE is validated on 9 real-world datasets from a variety of fields including ecology, biology, images, texts, etc. Table 1: Top: example-F1 scores, Middle: micro-F1 scores, and Bottom: macro-F1 scores for different methods on all the datasets. The best scores are in bold. Each score is the average after 3 runs. |
| Researcher Affiliation | Academia | Junwen Bai , Shufeng Kong and Carla Gomes Department of Computer Science, Cornell University {jb2467, sk2299}@cornell.edu, gomes@cs.cornell.edu |
| Pseudocode | Yes | Algorithm 1 Training MPVAE |
| Open Source Code | Yes | Our code is available on https://github.com/Junwen Bai/MPVAE |
| Open Datasets | Yes | MPVAE is validated on 9 real-world datasets from a variety of fields including ecology, biology, images, texts, etc. The datasets are e Bird [Munson et al., 2011], North American fish [Morley et al., 2018], mirflickr [Huiskes and Lew, 2008], NUS-WIDE2 [Chua et al., 2009], yeast [Nakai and Kanehisa, 1992], scene [Boutell et al., 2004], sider [Kuhn et al., 2016], bibtex [Katakis et al., 2008], and delicious [Tsoumakas et al., 2008]. Most datasets are available on a public website3. http://mulan.sourceforge.net/datasets-mlc.html |
| Dataset Splits | Yes | If a dataset has been split a priori, we follow those divisions. Otherwise, we separate the dataset into training (80%), validation (10%) and testing (10%). |
| Hardware Specification | No | No specific hardware details (like GPU models, CPU types, or memory) used for running experiments were provided. |
| Software Dependencies | No | No specific software versions (e.g., library or framework versions like PyTorch 1.9 or Python 3.8) were provided. |
| Experiment Setup | Yes | The encoders and decoder of MPVAE are parameterized by 3-layer fully connected neural networks with latent dimensionalities 512 and 256. The activation function in the neural networks is set to Re LU. Σr is a shared learnable parameter of size L L. By default, we set β = 1.1, λ1 = λ3 = 0.5, λ2 = 10.0. These default hyperparameter values are inherited from the existing well-trained DMVP [Chen et al., 2018] and β-VAE [Higgins et al., 2017] models. We achieve the best performance for our own model in the neighborhood of this default set of parameters via grid search. We also use grid search to find the best learning rate, learning rate decay ratio and dropout ratio hyperparameters. |