MDFL: Multi-Domain Diffusion-Driven Feature Learning

Authors: Daixun Li, Weiying Xie, Jiaqing Zhang, Yunsong Li

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental results on three multi-modal remote sensing datasets show that MDFL reaches an average overall accuracy of 98.25%, outperforming various state-of-the-art baseline schemes. Code available at https://github.com/LDXDU/ MDFL-AAAI-24.
Researcher Affiliation Academia State Key Laboratory of Integrated Services Networks, Xidian University, Xi an 710071, China
Pseudocode No The paper describes the model architecture and mathematical formulations but does not provide pseudocode or a clearly labeled algorithm block.
Open Source Code Yes Code available at https://github.com/LDXDU/ MDFL-AAAI-24.
Open Datasets Yes To validate the effectiveness of the proposed method in analyzing high-dimensional data, three remote sensing multimodal datasets with hyperspectra, namely the Houston2013 dataset, the Trento dataset, and the MUUFL dataset, are selected for verification of the proposed classification model.
Dataset Splits No The paper mentions "training samples" and "test images" but does not explicitly describe a validation dataset split, specific percentages for validation, or how a validation set was used.
Hardware Specification Yes These experiments are conducted on a machine equipped with an NVIDIA A100 Tensor Core GPU.
Software Dependencies No The paper mentions the use of "Adam optimizer" and a "step scheduler" but does not specify version numbers for any software dependencies like deep learning frameworks or libraries.
Experiment Setup Yes The training samples are randomly cropped to a size of 7 × 7. The Adam optimizer is employed with an initial learning rate set to 1e-3, and a weight decay of 5e-3 was applied. The training process spans 1000 epochs. In addition, a step scheduler with a step size of 50 and gamma value of 0.9 is utilized. The batch size is set to 64. Additionally, for incorporating noise, a total step size of 500 was used, and the values of t = 0, 50, 100, 200, 400 are selected.