For non-circular signals, the existing robust widely linear (WL) beamforming algorithms either suffer from performance degradation at high signal-to-noise ratio or have high computational complexities. In order to cope with this problem, three low-complexity robust WL beamformers, i.e., the WL-IQCQP, the WL-RBLW, and the WL-OAS, are proposed in this paper. First, based on the spatial spectrum of non-circularity coefficient, the augmented interference-plus-noise covariance matrix is reconstructed to remove the signal-of-interest’s component from the augmented sample covariance matrix. Then, the method of iterative quadratically constrained quadratic programming, the modified Rao-Blackwell Ledoit-Wolf estimator, and oracle approximating shrinkage estimator are developed, respectively, to estimate the desired signal’s extended steering vector. Only the prior knowledge of the antenna array geometry and the angular sector in which the desired signal is located are utilized in the proposed algorithms. Compared with several representative robust WL beamformers, numerical simulations demonstrate that the proposed beamformers can achieve better performances.
Meta-learning is a machine learning paradigm that extracts crosstask knowledge by learning a large number of subtasks, to fast adapt to new tasks. Many meta-learning methods are widely applied in few-shot classification. These methods adopt an episodic training strategy, and the learning subtasks are sampled uniformly from the task distribution. In this paper, we explore the effect of the order of training subtasks on the performance of different meta-learning algorithms and propose a curriculum learning framework to improve the generalization performance. We define the hardness of subtasks at the class level and guide the model to learn training subtasks from easy to hard. We evaluate our curriculum learning framework on two few-shot classification benchmarks (mini-ImageNet and FC100), and it achieves improvements across different meta-learning algorithms and datasets. In the cross-domain scenario, we compare the performance of different meta learning algorithms under three curriculum settings. The results show that our CL approach improves significantly the generalization performance of different meta-learning methods.