Multivariate time-series anomaly detection is central to modern cyber-physical and cloud monitoring systems. While most detectors rely solely on sensor or telemetry streams, operational anomalies are often accompanied by textual evidence such as logs. This paper presents a conditional multimodal framework for time-series anomaly detection that fuses sensor streams with log text through bidirectional gated cross-attention (Bi+Gate). The mechanism models sensor-to-log and log-to-sensor interactions, weighted via temporal mean pooling and a learnable sigmoid gate. Experiments across five datasets (MSL, SMAP, SWAT, SMD, PSM) show that MOMENT achieves the highest F1 on SMD (0.832) and TimesNet on PSM (0.974) and SWAT (0.924). Under the fusion ablation protocol, Bi+Gate is the best fusion variant on SMD (F1 = 0.818, +4.5 pp over uni-directional cross-attention, +2.5 pp over concatenation) and PSM (0.951), though it does not exceed MOMENT on the main benchmark table. Fusion is not universally beneficial: on MSL with cross-domain index-modulo log pairing, all fusion variants obtain F1 ≈ 0.42. A post-hoc structured reporting layer converts detector evidence into operator-facing summaries without altering F1. The results highlight that fusion effectiveness depends critically on log-sensor alignment quality.
Bo-Xuan Wang, Meng-Yao Du, Miao Zhang et al.· 2026 12th International Conf...· 0 citations
Real-world inference tasks for large language models exhibit diverse difficulty levels. Existing LLM serving systems integrate models of different sizes and attempt to route tasks of appropriate difficulty to the most suitable model, aiming to reduce resource waste while guaranteeing service quality. Such systems usually adopt a cascading architecture, which performs inference sequentially from lightweight models to heavyweight models and validates outputs until a model that meets the task requirements is identified. However, when handling complex tasks, the cascading architecture inevitably processes unnecessary small models first, leading to cumulative latency and redundant resource consumption. This paper proposes ParaCascade, a parallel cascading framework that supports early routing. The core idea of ParaCascade is to bypass lightweight models and directly route difficult instances to heavyweight model tiers by pre-estimating task complexity, thus avoiding ineffective computation on lightweight models. In addition, ParaCascade adopts parallel prediction and model parallel inference strategies. At the cost of a slight increase in energy consumption, it significantly reduces the systemic latency caused by sequential processing, thereby improving the overall QoS. Extensive evaluations across diverse workloads on the MMLU-pro and MATH benchmarks show that ParaCascade significantly outperforms both single-model deployments and serial inference serving baselines. While maintaining answer quality, it achieves an inference speedup of 1.16× to 1.51×, demonstrating its superiority in efficient LLM serving systems.
Hao Wei, Lujia Yin, Chen Chen et al.· Fall Joint Computer Conferen...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
