131 papers
This paper presents the design and implementation of a monitoring system for distributed clouds that utilizes node-based agents to collect, aggregate, and persist machine, container, and application metrics, making them accessible via multiple APIs to ensure the necessary observability for real-world viability.
This paper presents a scalable, AI-driven Intelligent Transportation System that orchestrates edge-cloud resources to process thousands of city-scale CCTV streams in real time, utilizing DNNs for detection, Spatio-Temporal GNNs for forecasting, and continuous federated learning to maintain accuracy under strict latency and bandwidth constraints.
This paper provides a complete characterization of the conditions under which randomized Election algorithms exist in anonymous networks with shared or unshared randomness, generalizing previous deterministic results and applying these findings to various levels of structural knowledge.
This paper presents PromptTuner, an SLO-aware elastic system that optimizes LLM prompt tuning through a Prompt Bank for efficient convergence and a Workload Scheduler for rapid resource allocation, significantly reducing SLO violations and operational costs compared to existing solutions.
This paper presents FluxSieve, a unified architecture that embeds lightweight in-stream precomputation and filtering into the data ingestion path to reconcile streaming and analytical data planes, thereby achieving orders-of-magnitude improvements in query performance for large-scale cloud observability with negligible overhead.
This final paper unifies philosophical, engineering, and physical perspectives through the "Leibniz Bridge" to propose that mutual information conservation, rather than forward temporal flow, is the fundamental principle enabling semantic consistency and dissolving classic distributed computing impossibility theorems.
This paper argues that the "FITO category mistake"—the failure to distinguish between mere temporal succession and meaningful causal integration due to the absence of a mandatory reflecting phase—systematically corrupts meaning across diverse domains including file synchronization, email, and human and artificial memory.
This study evaluates Python 3.14.2's experimental free-threaded build, revealing that while it significantly improves execution time and energy efficiency for parallelizable workloads, it incurs higher memory usage and increased energy consumption for sequential or highly contended tasks, indicating that its adoption depends on specific workload characteristics rather than offering a universal performance boost.
This paper argues that RDMA's high-performance data movement relies on a "completion fallacy"—a category mistake where delivery guarantees are mistaken for semantic commitment—creating a dangerous temporal gap between data placement and application integration that can only be resolved by a protocol architecture featuring a mandatory reflecting phase.
This paper presents a redesigned, GPU-accelerated solver for large-scale linear programs that decouples problem specification from the optimization engine using composable primitives, achieving at least a 10x speedup over prior CPU-based implementations while maintaining convergence guarantees.
This paper presents a distributed library prototype implementing Breadth-First Search, PageRank, and Triangle Counting using the HPX runtime system, demonstrating that its asynchronous execution and unified programming model significantly outperform conventional frameworks like GraphX and PBGL by effectively overcoming latency-bound limitations through fine-grained parallelism and overlapping communication with computation.
This paper introduces AMV-L, a value-driven memory lifecycle framework for long-running LLM agents that replaces age-based retention with utility-based tiering to bound retrieval workloads, thereby achieving significantly improved tail-latency control and throughput compared to traditional TTL and LRU policies.
FedEMA-Distill is a robust and communication-efficient federated learning framework that leverages server-side exponential moving average smoothing and ensemble knowledge distillation from compressed client logits to achieve superior accuracy, faster convergence, and Byzantine resilience under non-IID data conditions without requiring client-side software modifications.
This paper presents a one-round randomized distributed algorithm that 2-colors cycles with an expected monochromatic edge fraction below 0.24118 (improving the previous 0.25 bound) and establishes a new lower bound of 0.23879, with both results discovered by large language models and formalized in Lean 4.
This paper presents the design, implementation, and evaluation of a CPU-free MPI GPU communication abstraction leveraging HPE Slingshot 11 capabilities, which significantly reduces latency and improves strong scaling performance for GPU-based HPC applications by eliminating CPU involvement in the communication fast path.
This paper presents a complete classification of the distributed computational complexity for local optimization problems in directed cycles within both deterministic and randomized LOCAL models, identifying four distinct complexity classes and providing an efficient meta-algorithm to automatically determine the complexity and synthesize optimal distributed algorithms for any given problem.
This paper investigates "modality inflation" in multimodal large language models by providing the first stage-level energy analysis across vision encoding, prefill, and decoding, revealing significant energy overheads and GPU underutilization while demonstrating that stage-wise dynamic voltage and frequency scaling (DVFS) offers an effective optimization strategy for energy-efficient inference.
This paper introduces Cylon, a high-performance distributed data frame solution that leverages a serverless communicator using NAT Traversal TCP Hole Punching to enable direct communication between AWS Lambda functions, achieving scaling efficiency within 6.5% of traditional EC2 clusters for data-intensive machine learning applications.
This paper demonstrates that oblivious robots operating under sequential schedulers possess computational power orthogonal to and often superior to those under the fully synchronous scheduler (FSYNC), proving that Universal Pattern Formation is solvable in the former without additional assumptions while remaining unsolvable in the latter even with strong capabilities, with the exception that Gathering requires weak multiplicity detection in the sequential setting.
This paper introduces Parallel Split Learning with Global Sampling (GPSL), a server-driven scheme that fixes the global batch size and uses pooled-level proportions to draw local samples without replacement, thereby eliminating rounding bias, stabilizing optimization under non-IID data, and achieving centralized-like accuracy with negligible overhead.