Serverless computing has become a transformative cloud paradigm due to its operational simplicity, cost efficiency, and elastic scalability. Unfortunately, it suffers from cold-start problems caused by container initialization during the first function invocation or after container reclamation. Container caching (i.e., keep-alive policies) is an effective and widely adopted solution to mitigate cold-starts. However, existing keep-alive policies rely on fixed time thresholds or static heuristics, making them inadequate to handle the dynamic and heterogeneous workloads. In this paper, we propose LACE, a learned container caching framework for resource management in serverless computing. Leveraging high-quality features and labels of irregular functions, LACE employs a Mixture-of-Experts (MoE) architecture to model dynamic function invocation patterns. It further incorporates a multi-task learning structure that combines regression for fine-grained interval prediction with auxiliary classification for coarse-grained pattern estimation. Evaluation on Azure workload traces shows that LACE outperforms the state-of-the-art keep-alive policy, reducing cold-start overhead and rate by 32.72% and 22.46%, respectively, while achieving container caching accuracy close to the optimal Belady policy, all without incurring excessive model decision delay or memory footprints.
Byte-addressable Persistent Storage (BPS), such as persistent memory and CXL-enabled SSDs, has become an extension of main memory. This opens up new possibilities for indexes that operate and persist data directly on the memory bus. Recent learned indexes exploit data distribution and have shown great potential for some workloads. Despite some work proposed for integrating learned indexes into BPS, they are mainly based on Intel’s first-generation persistent memory. The current design suffers from the following problems: 1) Excessive storage line accesses due to large node in learned indexes; 2) Inefficient concurrency control due to volatile cache; 3) Write amplification due to mismatch access granularity. To resolve these challenges, we introduce a new learned index named PFLX, featuring three key design improvements: 1) a Storage Line-Friendly Node Layout that enhances search and insertion operations and minimizes storage line accesses through strategic use of pointers; 2) a Persistent Cache-based Lock-free Concurrency Mechanism that utilizes atomic primitives to serialize operations in leaf nodes effectively; 3) a Selective Data Flush Mechanism designed to reduce write amplification. Evaluations show that PFLX outperforms the state-of-the-art persistent indexes by 1.2 ∼ 3 ×.
Large Audio-language Models (LAMs) have recently enabled powerful speech-based interactions by coupling audio encoders with Large Language Models (LLMs). However, the security of LAMs under adversarial attacks remains underexplored, especially through audio jailbreaks that craft malicious audio prompts to bypass alignment. Existing efforts primarily rely on converting text-based attacks into speech or applying shallow signal-level perturbations, overlooking the impact of human speech's expressive variations on LAM alignment robustness. To address this gap, we propose StyleBreak, a novel style-aware audio jailbreak framework that systematically investigates how diverse human speech attributes affect LAM alignment robustness. Specifically, StyleBreak employs a two-stage style-aware transformation pipeline that perturbs both textual content and audio to control linguistic, paralinguistic, and extralinguistic attributes. Furthermore, we develop a query-adaptive policy network that automatically searches for adversarial styles to enhance the efficiency of LAM jailbreak exploration. Extensive evaluations demonstrate that LAMs exhibit critical vulnerabilities when exposed to diverse human speech attributes. Moreover, StyleBreak achieves substantial improvements in attack effectiveness and efficiency across multiple attack paradigms, highlighting the urgent need for more robust alignment in LAMs.
Multi-agent workflows have emerged as an important execution pattern for intelligent applications, where specialized agents collaborate through dependent stages such as planning, retrieval, execution, and verification. When such workflows are deployed over edge-cloud infrastructures, scheduling becomes challenging because task dependencies, heterogeneous resource conditions, and topology-dependent communication overhead must be considered jointly. We study the problem of scheduling multi-agent workflows in edge-cloud environments and propose a graph-aware scheduling method that models workflow execution as a task graph and the underlying infrastructure as a resource graph. The method combines structure-aware task and resource representations, communication-sensitive assignment scoring, and online resource-state updates to improve task placement quality. Experiments across different workflow complexities, system scales, and dynamic operating conditions show that the proposed method achieves a favorable balance among latency, communication overhead, and execution cost, particularly in communication-sensitive and medium-to-large-scale settings. These results suggest that jointly modeling workflow structure and resource topology can improve scheduling quality for multi-agent workflow execution in heterogeneous edge-cloud environments.