28 April 2026 in IEEE Transactions on Computational Social Systems
Multimodal emotion recognition (MER) has been widely adopted in affective computing and human–computer interaction. However, real-world multimodal streams frequently suffer from missing or incomplete modalities due to sensor failures, privacy constraints, and heterogeneous acquisition costs, which often causes severe performance degradation for models trained under complete inputs. To address this issue, we propose hierarchical VAE–diffusion for emotion reconstruction (HVDER), a latent-space generative framework for incomplete MER. HVDER first employs modality-specific variational autoencoders (VAEs) to project language, visual, and audio features into a unified low-dimensional latent space with KL-regularized structure. Then, a two-stage coarse-to-fine conditional diffusion module completes missing modality latents by recovering global emotion semantics followed by refining local discriminative details. Finally, the completed and observed modality representations are fused for emotion prediction, optimized end-to-end with a multiobjective loss integrating reconstruction, diffusion matching, cross-modal alignment, and classification supervision. Extensive experiments on CMU-MOSEI and CMU-MOSI validate the effectiveness and robustness of HVDER. Under fixed modality-missing settings, HVDER achieves average ACC2/F1 of 76.8/75.4 on CMU-MOSEI and 73.4/72.9 on CMU-MOSI, outperforming representative baselines across all modality combinations. Under random missing with a high missing rate of MR = 0.7, HVDER maintains ACC2/F1 of 75.4/72.5 on CMU-MOSEI and 68.0/67.2 on CMU-MOSI, indicating a stronger performance lower bound in severely incomplete scenarios. Ablation studies further confirm that latent-space modeling and the coarse-to-fine diffusion design jointly contribute to the main performance gains.
21 May 2026 in IEEE Transactions on Computational Social Systems
As a core implementation of cyber physical social systems (CPSS) in smart manufacturing, the Industrial Internet of Things (IIoT) is accelerating the shift of industrial intelligence toward collaborative learning and distributed-intelligence paradigms. However, the explosive growth of industrial data traffic and the heterogeneity of edge devices make it difficult for conventional traffic management schemes to provide high-accuracy, low-latency classification under strict resource and energy constraints. To address these challenges, this article proposes a collaborative meta-computing-based traffic classification framework for IIoT-enabled CPSS, which jointly considers lightweight model design and energy-efficient distributed collaboration. First, a depthwise separable convolutional neural network (DSCNN) is adopted as the local classifier on industrial edge gateways, significantly reducing computational complexity while preserving fine-grained spatiotemporal patterns in network flows. Second, a model-quality-driven node collaboration mechanism is developed to adaptively select high-quality edge nodes for participation in each global round, thereby improving collaborative learning efficiency and reducing redundant communication. Third, an energy–delay optimal CPU frequency scheduling scheme is introduced to dynamically adjust the computing frequency of each participating node, balancing training delay and energy consumption across heterogeneous gateways. Experiments conducted on the CICIoT2022 and UNSW-IoTBotnet2018 datasets demonstrate that the proposed method effectively supports collaborative, energy-aware traffic classification in distributed IIoT scenarios. The DSCNN-based classifier achieves over 99% accuracy with low computational overhead; the node selection mechanism accelerates model convergence by reducing training epochs by approximately 15%; and the joint optimization of node collaboration and computing frequency yields an overall energy reduction of about 24% compared with the baseline meta-computing setting.