Enhancing Multi-View Classification Reliability with Adaptive Rejection

Authors: Wei Liu, Yufei Chen, Xiaodong Yue

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The effectiveness of our method is demonstrated through comprehensive theoretical analysis and empirical experiments on various multi-view datasets, establishing its superiority in enhancing the reliability of multi-view classification.
Researcher Affiliation Academia Wei Liu1, Yufei Chen1 *, Xiaodong Yue2 1 School of Computer Science and Technology, Tongji University, Shanghai, China 2 Artificial Intelligence Institute of Shanghai University, Shanghai University, Shanghai, China EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes mathematical formulations and propositions but does not include any explicitly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements about releasing source code or provide links to a code repository.
Open Datasets Yes Datasets: We conducted experiments on six real-world multi-view datasets as follows: ANIMAL (Lampert, Nickisch, and Harmeling 2013) [...] HAND (van Breukelen et al. 1998) [...] CUB (Wah et al. 2011) [...] SCENE (Fei-Fei and Perona 2005) [...] MRNet (Bien et al. 2018) [...] CAL (Fei-Fei, Fergus, and Perona 2004)
Dataset Splits Yes For all multi-view datasets, the data was split into training (70%), testing (20%), and calibration (10%) sets.
Hardware Specification Yes The model was implemented in Py Torch and run on a Ge Force RTX 4090 GPU with 24GB memory.
Software Dependencies No The Adam optimizer (Kingma and Ba 2014) was used for network training, with l2-norm regularization set to 1e 5. [...] The model was implemented in Py Torch and run on a Ge Force RTX 4090 GPU with 24GB memory. The paper mentions Py Torch and Adam optimizer but does not specify their version numbers.
Experiment Setup Yes The Adam optimizer (Kingma and Ba 2014) was used for network training, with l2-norm regularization set to 1e 5. A 5-fold cross-validation was employed to select the learning rate from 1e 5, 3e 4, 1e 3, 3e 3 . For all multi-view datasets, the data was split into training (70%), testing (20%), and calibration (10%) sets.