A growing body of recent researches have migrated graph structure learning (GSL) to the multivariate time series forecasting (MTSF), which lays the foundation for the promotion of ''Generalized Graph'' for multimedia MTSF applications. In other words, we expect generalized graph to encompass the learning of inter-variable, inter-temporal and latent correlations, becoming a universal tool for multivariate correlations learning. However, due to the heterogeneity of multivariate time series in distribution, graph learning inevitably captures inaccurate relationships, which requires the quality of graph learning; Meanwhile MTSF often requires instant predictions for decision-making in real-world, which also challenges the speed of GSL. To solve these challenges, we propose AGGA-MVFLN, namely Adaptive Generalized Graph Accompanied Multi-View Frequency Learning Network. Specifically, we introduce an adaptive generalized graph structure from multi-view (global and local) to capture diverse ''spatio-temporal patterns''. Subsequently, we utilize the Fast Fourier Transform to map them into the frequency domain, and enhance the quality of the generalized graph by collaboratively learning the complementarities and differences through reconstructed ''spatio-temporal patterns'' and error-driven supervised training of adaptive graph. The benefits are: (1) The frequency domain can disentangle complex temporal patterns, making the process of learning multivariate relationships more robust. (2) Multi-view learning can significantly reduce training time by preset and seamless integration (i.e., the multi-task loss form). (3) ''Generalized Graph'' can be regarded as universal component for multivariate correlation learning. Evaluation of 9 real-world datasets confirms the superiority of AGGA-MVFLN over SOTA benchmark.

Quantum computing has been accumulating tremendous attention in recent years. In current superconducting quantum processors, each qubit can only be connected with a limited number of neighbors. Therefore, the original quantum circuit should be converted to a hardware-dependent circuit, and this process is called qubit mapping and routing, in which typically extra SWAP gates need to be inserted. Due to a limited qubit lifetime, one of the main objectives of qubit mapping and routing is to minimize the circuit depth, which is a time-consuming process. By studying several existing greedy mappers, we extract and analyze two patterns that significantly impact the mapping and routing performance. Then, we propose a sliding window method named SWin, which dramatically reduces the computational cost with negligible performance degradation. For devices with constrained executable circuit depth, we propose SWin+, which introduces adaptive circuit slicing methods with VF$2+ {+}$subgraph isomorphism initial mapping methods. Compared with the state-of-the-art greedy methods, SWin can find an effective result by up to 39% depth decrease, on average of 16% for large-scale circuits. Moreover, SWin can be easily modified to be noise-aware, while the depth reduction will yield better performance for real execution. Furthermore, SWin still performs well for various chip couplings. SWin+ significantly enhances processing efficiency, achieving improvements up to$22.3\times $, with an average increase of$6.1\times $. Concurrently, it maintains the effectiveness of the transformed circuit depth.

System Protection Communication Network (SPCN) is a new type of high-speed, real-time, secure and reliable communication network proposed in China supporting services such as AC/DC control, pumped storage control etc. In order to reduce the impact of SPCN failure on electric power system, this paper proposes a risk modeling and optimization approach. Firstly, we build a risk model to analyze the dynamic link and service risk from aspects of failure probability and its impact value. Then, we construct a risk optimization problem aiming at minimizing the link risk balance degree with service quality and risk constraints, and propose improved genetic algorithm to solve it. Based on part of network topology from a Chinese province, simulation results show that the proposed approach can make SPCN more reliable comparing to other methods when link failure occurs.