Disambiguated Attention Embedding for Multi-Instance Partial-Label Learning
Authors: Wei Tang, Weijia Zhang, Min-Ling Zhang
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on benchmark and real-world datasets validate the superiority of DEMIPL against the compared MIPL and partial-label learning approaches. |
| Researcher Affiliation | Academia | School of Computer Science and Engineering, Southeast University, Nanjing 210096, China; Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, China; School of Information and Physical Sciences, The University of Newcastle, Callaghan, NSW 2308, Australia |
| Pseudocode | Yes | Algorithm 1 summarizes the complete procedure of DEMIPL. |
| Open Source Code | Yes | Additionally, the code of DEMIPL, the benchmark datasets, and the real-world dataset are publicly available at http://palm.seu.edu.cn/zhangml/. |
| Open Datasets | Yes | We utilize four benchmark MIPL datasets stemming from MIPLGP literature [36], i.e., MNIST-MIPL, FMNIST-MIPL, Birdsong-MIPL, and SIVAL-MIPL from domains of image and biology [48 51]. Additionally, the code of DEMIPL, the benchmark datasets, and the real-world dataset are publicly available at http://palm.seu.edu.cn/zhangml/. |
| Dataset Splits | No | The paper states 'We conduct ten runs of random train/test splits with a ratio of 7 : 3 for all datasets' but does not explicitly mention a separate validation split or its proportion. |
| Hardware Specification | Yes | DEMIPL is implemented using Py Torch [59] on a single Nvidia Tesla V100 GPU. |
| Software Dependencies | No | The paper mentions 'Py Torch [59]' but does not provide a specific version number for PyTorch or other software dependencies with their versions. |
| Experiment Setup | Yes | We employ the stochastic gradient descent (SGD) optimizer with a momentum of 0.9 and weight decay of 0.0001. The initial learning rate is chosen from a set of {0.01, 0.05} and is decayed using a cosine annealing method [60]. The number of epochs is set to 200 for the SIVAL-MIPL and CRC-MIPL datasets, and 100 for the remaining three datasets. The value of λa is selected from a set of {0.0001, 0.001}. |