Multiteacher Multiobjective Meta-Learning for Zero-Shot Heterogeneous Hyperspectral Band Selection

Band selection plays a crucial role in hyperspectral image classification by removing redundant and noisy bands and retaining discriminative ones. However, most existing cross-scene band selection methods are targeted at specific homogeneous datasets. In heterogeneous datasets with different numbers of spectral bands and wavelength ranges, the previous methods exhibit limited ability to generalize to unseen target datasets.To address this issue, a novel multi-teacher multi-objective meta-learning network (M3BS) is proposed for zero-shot hyperspectral band selection and classification. In M3BS, heterogeneous spectral information is effectively modeled through a spatial-spectral graph, enabling a generalizable graph convolution network (GCN) is convolutional to generate dataset-agnostic base, and extract compatible meta-knowledge from multiple band selection tasks. To enhance the ability of meta-knowledge extraction, multiple band selection teachers are introduced to provide diverse high-quality experiences. Finally, subsequent classification tasks are attached and jointly optimized with multi-teacher band selection tasks through multi-objective meta-learning in an end-to-end trainable way. Multi-objective meta-learning guarantees to coordinate diverse optimization objectives automatically and adapt to various datasets simultaneously. Once the optimization is accomplished, the acquired meta-knowledge can be directly transferred to unseen datasets without any retraining or fine-tuning. Experimental results demonstrate that our proposed method outperforms state-of-the-art baselines for zero-shot hyperspectral band selection.