Graph-based multi-view clustering aims to leverage the consistency and complementarity of multiple information sources (views) to enhance clustering performance. The introduction of multi-order graphs has brought significant performance gains by mitigating the sparsity of first-order graphs. However, different views, along with their derived high-order graphs, inevitably contain noise and view-specific information (i.e., diversity), which may hinder the learning of a consensus graph. To address this critical issue while retaining the benefits of multi-order structures, this paper proposes a novel framework, termed Consistency driven Decomposition for Multi-view Multi-order Graph Clustering (CDMMGC). In CDMMGC, multi-order graphs are utilized to mitigate the sparsity problem of first-order graphs and each multi-order graph from each view is decomposed into a consistency and a diversity component. Accordingly, the consensus graph is learned on the consistency component from multi-view multi-order graphs, which is expected to be more accurate in capturing the data semantics. Experiments on extensive datasets demonstrate the effectiveness and superiority of the proposed CDMMGC model in data clustering compared with the state-of-the-art methods. The code is available athttps://github.com/SunseaIU/CDMMGC

In the dynamic field of film and game development, the emergence of human motion synthesis methods has revolutionized avatar animation. Traditional methodologies, typically reliant on single modality inputs like text or audio, employ modality-specific model frameworks, posing challenges for unified model deployment and application. To address this, we propose Everything2Motion, a unified model framework. Everything2Motion consists of three key modules. The Input-Output Modality Modulation module tailors structures for specific multimodal inputs, eliminating the need for modality-specific frameworks. The Query-aware Autoencoder, based on the transformer encoder-decoder architecture, enables efficient latent motion generation. Lastly, the Prior Motion Distillation Decoder, a pretrained module, enhances the final skeleton sequence's naturalness and fluidity. Comprehensive experiments on several public datasets demonstrate the effectiveness of Everything2Motion, highlighting its potential for practical applications and setting a new benchmark in human motion synthesis.