Fused Feature Representation Discovery for High-Dimensional and Sparse Data
Authors: Jun Suzuki, Masaaki Nagata
AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the effectiveness of our method in experiments consisting of two well-studied natural language processing tasks. Experiments We conducted experiments on the data sets of two well-studied natural language processing (NLP) tasks, namely named entity recognition (NER) and dependency parsing (DEPAR). |
| Researcher Affiliation | Industry | Jun Suzuki and Masaaki Nagata NTT Communication Science Laboratories, NTT Corp. 2-4 Hikaridai, Seika-cho, Soraku-gun, Kyoto, 619-0237 Japan {suzuki.jun, nagata.masaaki}@lab.ntt.co.jp |
| Pseudocode | Yes | Figure 2: Entire optimization framework of our method based on ADMM (Boyd et al. 2011) |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described in this paper. |
| Open Datasets | Yes | We gathered 150 million data points (sentences) from the LDC corpus, that is, |D|=150M. |
| Dataset Splits | No | The paper mentions using supervised data (DL) and unsupervised data (D), but does not provide specific train/validation/test dataset splits (percentages, sample counts, or explicit partitioning methodology). |
| Hardware Specification | No | The paper states that approximately 14 hours were taken for calculating the gradient step with 512 nodes of our Map Reduce system, but it does not specify any exact hardware details like GPU/CPU models or memory amounts. |
| Software Dependencies | No | The paper mentions using CRF and the online structured output learning version of the Passive Aggressive algorithm (ost PA) but does not provide specific version numbers for these or any other software dependencies. |
| Experiment Setup | No | The paper states, 'We simply followed the experimental settings described in previous studies providing state-of-the-art results, i.e., (Suzuki and Isozaki 2008) for NER, and (Koo, Carreras, and Collins 2008) for DEPAR.' While it mentions varying 'K' and controlling 'λ1', it does not provide specific hyperparameter values like learning rates, batch sizes, or epoch counts within the main text. |