As training scales grow, collective communication libraries (CCL) increasingly face anomalies arising from complex interactions among hardware, software, and environmental factors. These anomalies typically manifest as slow/hang communication, the most frequent and time-consuming category to diagnose. However, traditional diagnostic methods remain inaccurate and inefficient, frequently requiring hours or even days for root cause analysis. To address this, we propose CCL-D, a high-precision diagnostic system designed to detect and locate slow/hang anomalies in large-scale distributed training. CCL-D integrates a rank-level real-time probe with an intelligent decision analyzer. The probe measures cross-layer anomaly metrics using a lightweight distributed tracing framework to monitor communication traffic. The analyzer performs automated anomaly detection and root-cause location, precisely identifying the faulty GPU rank. Deployed on a 4,000-GPU cluster over one year, CCL-D achieved near-complete coverage of known slow/hang anomalies and pinpointed affected ranks within 6 minutes—substantially outperforming existing solutions.

Deep neural networks (DNNs) are susceptible to Universal Adversarial Perturbation (UAP), which significantly increases the likelihood of deceiving DNNs. Current UAP generation methods are categorized into data-dependent, relaxed-data-free, and data-free attacks based on different data dependencies. However, both strategies exhibit poor transferability in the black-box settings. To address this limitation, we propose BTUAP, a novel UAP generation method designed to enhance the transferability of UAP in the black-box setting. BTUAP employs an ensemble strategy with min-max weight adjustment mechanisms to reduce the impact of model characteristics and introduces a self-supervised optimization strategy to maximize the distance of predicted logits between benign samples and adversarial examples. Experimental results demonstrate that BTUAP significantly improves transferability in different data dependency settings under black-box constraints. We also quantify the impact of the distribution shift and provide a new metric to measure the robustness of models. The source code is available at htps://github.com/RaymondDawn/BTUAP.

Deep neural networks are vulnerable to adversarial examples, where decision-based attacks can generate adversarial examples based solely on the predicted labels. However, these attacks typically require excessive queries to attack one example. Considering this challenge, we propose FBA (Frequency based Boundary Attack), a decision-based attack against the limitation of query efficiency. FBA incorporates a novel search process, utilizing high-frequency based importance sampling for efficient gradient estimation. Empirical results confirm the superior query efficiency of our method. Specifically, FBA surpasses SOTA attacks by achieving a 54% average improvement in query efficiency, quantified by the reduction in perturbation size within the same number of queries.

Deep neural networks are vulnerable to adversarial attacks. We study such threats in the decision-based black-box setting where the adversary could obtain only the predicted labels of the victim classifier within limited queries and aims at performing targeted and untargeted adversarial attacks under different perturbation constraints. In this paper, we propose BounceAttack as a query-efficient attack method. We propose the bounce vector which encourages the iterations to maximally explore the adversarial space towards the optimal adversarial example within limited queries to improve the query efficiency. We perform extensive experiments on various benchmark datasets and models. Experimental results show that BounceAttack achieves both high query efficiency and small perturbation size. BounceAttack outperforms existing attack methods. For example, BounceAttack achieves 48.1% smaller perturbation compared to the state-of-the-art attack methods on average using the same number of model queries.