Adversarial Training(AT) on long-tailed distributions triggers catastrophic vulnerability in tail classes. Prior work has predominantly relied on passive loss reweighting to compensate for data imbalance, leaving the underlying scarcity of tail-class samples unaddressed at the data level. In this work, we propose an active manifold-completion strategy that integrates conditional diffusion generation with self-supervised adversarial contrastive learning. Our frame-work introduces three core innovations. First, we exploit diffusion-based generative priors to repair the fragmented feature manifolds of minority classes by synthesizing on-manifold samples conditioned on tail-class labels. Second, we propose LT SSL AT, a dual-path self-supervised adversarial framework that decouples semantic invariance learning (clean-augmented views) from robustness learning (clean vs. adversarial pairs). Third, we introduce a curriculum scheduling weight s(t) that progressively shifts optimization from an Self-Supervised Learning(SSL)-only warm-up phase to joint adversarial alignment, preventing representation collapse. On CIFAR-10-LT (IR=10), CD-SSL achieves 35.12% Balanced Robustness under AutoAttack, surpassing the state-of-the-art TAET baseline by 4.58 percentage points. Extensive experiments across imbalance ratios 10, 20, 50, and 100 on three benchmarks (CIFAR-10-LT, CIFAR-100-LT, and MedMNIST) confirm consistent gains over both long-tailed and adversarial baselines.

The search for big data resources is the core foundational function of big data services in cloud environments. Currently, the search methods include centralized Service Oriented Architecture (SOA), structured Peer-to-Peer (P2P), and unstructured P2P. However, SOA has a single point of failure, structured P2P has a complex maintenance mechanism, and unstructured P2P has network congestion and slow search problems. Therefore, firstly, we adopt a hybrid P2P network as the topology structure of data resource nodes in the cloud environment, encapsulating data resources with services, and simplifying user access to data resources by matching service description information. Secondly, in order to further improve search efficiency and stability, an active replication protocol based on data index between supernodes is proposed. Finally, a flooding-based data resource search method among supernodes is proposed, which achieves scalable resource management and search, ensuring that the system can maintain efficient operation even when scaling up. The combination of these three provides a flexible infrastructure, high search efficiency, and scalability. Experiments have shown that under specific conditions, our proposed method reduces the number of messages to one percent of the Flooding network and reduces the average hop count by about 50% compared to the traditional hybrid P2P network.