How to rapidly adapt to new tasks and improve model generalization through few-shot learning remains a significant challenge in meta-learning. Model-Agnostic Meta-Learning (MAML) has become a powerful approach, with offers a simple framework with excellent generality. However, the requirement to compute second-order derivatives and retain a lengthy calculation graph poses considerable computational and memory burdens, limiting the practicality of MAML. To address this issue, we propose Evolving MAML (Evo-MAML), an optimization-based meta-learning method that incorporates evolving gradient within the inner loop. Evo-MAML avoids the second-order information, resulting in reduced computational complexity. Experimental results show that Evo-MAML exhibits higher generality and competitive performance when compared to existing first-order approximation approaches, making it suitable for both few-shot learning and meta-reinforcement learning settings.

Deep reinforcement learning (DRL) typically requires reinitializing training for new tasks, limiting its generalization due to isolated knowledge transfer. Meta-reinforcement learning (Meta-RL) addresses this by enabling rapid adaptation through prior task experiences, yet existing gradient-based methods like MAML suffer from poor out-of-distribution performance due to overfitting narrow task distributions. To overcome this limitation, we propose Evolving Gradient Regularization MAML (ER-MAML). By integrating evolving gradient regularization into the MAML framework, ER-MAML optimizes meta-gradients while constraining adaptation directions via a regularization policy. This dual mechanism prevents overparameterization and enhances robustness across diverse task distributions. Experiments demonstrate ER-MAML outperforms state-of-the-art baselines by $14.6\%$ in out-of-distribution success rates. It also achieves strong online adaptation performance in the MetaWorld benchmark. These results validate ER-MAML's effectiveness in improving meta-RL generalization under distribution shifts.

In few-shot learning, methods based on graph neural network (GNN) are proposed for full exploitation of relations between samples. However, most GNN-based methods only explore a very rough relation among instances, ignoring that the relation between two samples is more complicated than a similarity value can convey. In this paper, we propose a model named GNN-based Relations Extraction for local features (GNN-RE in short) to explore the complex relations between instances. Specifically, we represent each image with multiple local semantic features, with each representing one type of semantic information. By exploring the relations among local semantic features in different images, we can learn more complex relations between images. To evaluate the effectiveness of the proposed GNN-RE, we conduct extensive experiments on two benchmark datasets. Our method achieves better or competitive performance comparing with previous work.