Knowledge Amalgamation for Multi-Label Classification via Label Dependency Transfer
Authors: Jidapa Thadajarassiri, Thomas Hartvigsen, Walter Gerych, Xiangnan Kong, Elke Rundensteiner
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show that ANT succeeds in unifying label dependencies among teachers, outperforming five state-of-the-art methods on eight real-world datasets.Our comprehensive experimental study on eight real-world datasets demonstrates that ANT significantly outperforms state-of-the-art alternatives by achieving the best averaged rank across numerous standard multi-label metrics for all datasets. |
| Researcher Affiliation | Academia | 1Worcester Polytechnic Institute, 2Massachusetts Institute of Technology {jthadajarassiri, wgerych, xkong, rundenst}@wpi.edu, tomh@mit.edu |
| Pseudocode | No | The paper describes the proposed method step-by-step using text and equations, but it does not include a formally labeled 'Pseudocode' or 'Algorithm' block. |
| Open Source Code | Yes | All code, datasets, and experimental details are made publicly available at https://github.com/jida-thada/ANT. |
| Open Datasets | Yes | We conduct experiments on eight established benchmark datasets for evaluating multi-label classifiers including EMOTIONS (Trohidis et al. 2008), SCENE (Boutell et al. 2004), YELP (Sajnani et al. 2012), YEAST (Elisseeff and Weston 2001), BIRD (Briggs et al. 2013), TMC (Srivastava and Zane-Ulman 2005), GENBASE (Diplaris et al. 2005), and MEDICAL (Pestian et al. 2007). |
| Dataset Splits | No | The paper states 'using 75% of the 30% left-out data to train the student models, and use the remaining 25% for testing,' which describes train/test split percentages for the student, but does not explicitly mention a separate validation split or its percentage. |
| Hardware Specification | No | The paper does not provide any specific details regarding the hardware used for running the experiments, such as GPU models, CPU types, or memory. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers, such as programming language versions or library versions (e.g., Python 3.x, PyTorch 1.x). |
| Experiment Setup | No | The paper describes the general training process for the student model (e.g., using RNN with LSTMs, minimizing binary cross entropy) but does not provide specific experimental setup details such as hyperparameter values (learning rate, batch size, epochs), optimizer settings, or model initialization details. |