Open-set signal recognition provides a new approach for verifying the robustness of models by introducing novel unknown signal classes into the model testing and breaking the conventional closed-set assumption, which has become very popular in real-world scenarios. In the present work, we propose an efficient open-set signal recognition algorithm, which contains three key sub-modules: the signal representation sub-module based on a vision transformer (ViT) structure, a set distance metric sub-module based on Wasserstein distance, and a class space compression sub-module based on reciprocal point separation and central loss. In this algorithm, the representing features of signals are established based on transformer-based neural networks, i.e., ViT, in order to extract global information about time series-related data. The employed reciprocal point is used in modeling the potential unknown space without using the corresponding samples, while the distance metric between different class spaces is mathematically modeled in terms of the Wasserstein distance instead of the classical Euclidean distance. Numerical experiments on different open-set signal recognition tasks show that the proposed algorithm can significantly improve the recognition efficiency in both known and unknown categories.

Federated learning has been a hot topic in the field of artificial intelligence in recent years due to its distributed nature and emphasis on privacy protection. To better align with real-world scenarios, federated class incremental learning (FCIL) has emerged as a new research trend, but it faces challenges such as heterogeneous data, catastrophic forgetting, and inter-client interference. However, most existing methods enhance model performance at the expense of privacy, such as uploading prototypes or samples, which violates the basic principle of only transmitting models in federated learning. This paper presents a novel selective knowledge fusion (FedSKF) model to address data heterogeneity and inter-client interference without sacrificing any privacy. Specifically, this paper introduces a PIT (projection in turn) module on the server side to indirectly recover client data distribution information through optimal transport. Subsequently, to reduce inter-client interference, knowledge of the global model is selectively absorbed via knowledge distillation and an incomplete synchronization classifier at the client side, namely an SKS (selective knowledge synchronization) module. Furthermore, to mitigate global catastrophic forgetting, a global forgetting loss is proposed to distill knowledge from the old global model. Our framework can easily integrate various CIL methods, allowing it to adapt to application scenarios with varying privacy requirements. We conducted extensive experiments on CIFAR100 and Tiny-ImageNet datasets, and the performance of our method surpasses existing works.