Maximizing Subset Accuracy with Recurrent Neural Networks in Multi-label Classification
Authors: Jinseok Nam, Eneldo Loza Mencía, Hyunwoo J. Kim, Johannes Fürnkranz
NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In order to see whether solving MLC problems using RNNs can be a good alternative to classifier chain (CC)-based approaches, we will compare traditional multi-label learning algorithms such as BR and PCCs with the RNN architectures (Fig. 1) on multi-label text classification datasets. |
| Researcher Affiliation | Academia | Jinseok Nam1, Eneldo Loza Mencía1, Hyunwoo J. Kim2, and Johannes Fürnkranz1 1Knowledge Engineering Group, TU Darmstadt 2Department of Computer Sciences, University of Wisconsin-Madison |
| Pseudocode | No | The paper describes its models and processes using textual descriptions and figures (e.g., Figure 1 illustrating RNN architectures), but it does not include any clearly labeled 'Pseudocode' or 'Algorithm' blocks. |
| Open Source Code | No | The paper does not provide an explicit statement about releasing its own source code or a link to a code repository for the methodology described. |
| Open Datasets | Yes | We use three multi-label text classification datasets for which we had access to the full text as it is required for our approach Enc Dec, namely Reuters-21578,2 RCV1-v2 [15] and Bio ASQ,3 each of which has different properties. |
| Dataset Splits | Yes | For tuning hyperparameters, we set aside 10% of the training instances as the validation set for both Reuters-21578 and RCV1-v2, but chose randomly 50 000 documents for Bio ASQ. |
| Hardware Specification | Yes | We used the NVIDIA Titan X to train NN models including RNNs and base learners. For Fast XML, a machine with 64 cores and 1024GB memory was used. |
| Software Dependencies | No | The paper mentions optimization algorithms like Adam and techniques like dropout, and the use of word2vec for word embeddings, but does not provide specific version numbers for software dependencies such as deep learning frameworks or programming languages. |
| Experiment Setup | Yes | The dimensionality of hidden states of all the NN baselines as well as the RNNs is set to 1024. The size of label embedding vectors is set to 256. |