Paper Search
Found 124 papers
-
HALO: Semantic-Aware Distributed LLM Inference in Lossy Edge Network
Peirong Zheng, Wenchao Xu, Haozhao Wang, Jinyu Chen, Xuemin Shen
January 16, 2026
cs.DC arXiv: 2601.11676The deployment of large language models' (LLMs) inference at the edge can facilitate prompt service responsiveness while protecting user privacy
distributed llm inference edge computing lossy networks packet loss rate tensor parallelism semantic-aware prediction activation sparsity neuron groups relaxed synchronization parallel execution load balancing heterogeneous edge devices scheduler min-max scheduling plr-aware scheduling overlap scheme end-to-end speedup model accuracy gsm8k dataset arc dataset smart home raspberry pi cluster quantization time per token semantic imbalancecs.NI cs.DC cs.AI -
HybridFlow: Adaptive Task Scheduling for Fast and Token-Efficient LLM Inference in Edge-Cloud Collaboration
Jiangwen Dong, Jiayu Li, Tianhang Zheng, Wanyu Lin
December 11, 2025
cs.DC arXiv: 2512.22137Large language models (LLMs) exhibit impressive reasoning and problem-solving abilities, yet their substantial inference latency and token consumption pose major
hybridflow edge-cloud collaboration llm inference adaptive task scheduling token-efficient inference task decomposition dependency-aware dag parallel execution resource-aware routing utility modeling learned router knapsack formulation lagrangian relaxation online adaptation dual thresholding contextual bandit fine-grained parallelism budget-aware routing subtask routing benefit-cost utility normalized cost gpqa mmlu-pro aime24 livebench-reasoningcs.DC -
Computing Embeddings and Isomorphisms of Finite Semigroups
James East, Attila Egri-Nagy, Andrew R. Francis, James D. Mitchell
March 20, 2016
math.GR arXiv: 1603.06204Semigroup theory is a branch of abstract algebra, and it provides mathematical tools for the theory of computation. Finite semigroups can describe state
semigroup theory finite semigroups isomorphism embedding backtrack search abstract algebra transformation semigroups diagram semigroups minimal degree representation partitioned backtrack index-period element profile brauer monoid partition monoid cayley table graph isomorphism algebraic automata theory computational experiments generator sets conjugation subsemi package kigen package parallel execution 2-generated subsemigroups symmetry groupsmath.GR -
HD-GEN: A High-Performance Software System for Human Mobility Data Generation Based on Patterns of Life
Hossein Amiri, Joon-Seok Kim, Hamdi Kavak, Andrew Crooks, Dieter Pfoser, Carola Wenk, Andreas Züfle
January 03, 2026
cs.SE arXiv: 2601.01219Understanding individual-level human mobility is critical for a wide range of applications. As such, real-world trajectory datasets provide valuable insights into actual
human mobility patterns of life synthetic data generation trajectory simulation agent-based modeling genetic algorithm calibration openstreetmap geolife dataset software pipeline data processing visual analytics scalability privacy preservation behavioral modeling checkpointing log streaming parallel execution anomaly detection spatiotemporal trajectories urban planning social network analysis mason toolkit reproducibility hd-gencs.SE -
Monadic Context Engineering
Yifan Zhang, Yang Yuan, Mengdi Wang, Andrew Chi-Chih Yao
December 27, 2025
cs.AI arXiv: 2512.22431The proliferation of Large Language Models (LLMs) has catalyzed a shift towards autonomous agents capable of complex reasoning and tool use
monadic context engineering large language models autonomous agents monads applicative functors monad transformers agentmonad functional programming category theory state management error handling concurrency model context protocol meta-agents generative orchestration metaprogramming asynchronous execution parallel execution statet eithert io monad compositionality agent architecture tool use short-circuitingcs.CL cs.FL cs.AI -
Introducing Instruction-Accurate Simulators for Performance Estimation of Autotuning Workloads
Rebecca Pelke, Nils Bosbach, Lennart M. Reimann, Rainer Leupers
May 19, 2025
cs.AR arXiv: 2505.13357Accelerating Machine Learning (ML) workloads requires efficient methods due to their large optimization space
autotuning instruction-accurate simulators performance estimation machine learning workloads simulator interface score predictor gem5 simulator tvm cache optimization parallel execution multiple linear regression deep neural networks bayesian optimization xgboost x86 architecture arm architecture risc-v architecture conv2d+bias+relu kernels resnet full system simulator atomic mode timing mode cache hierarchy prediction metrics pre-silicon software developmentcs.LG cs.AR -
The Six Sigma Agent: Achieving Enterprise-Grade Reliability in LLM Systems Through Consensus-Driven Decomposed Execution
Khush Patel, Siva Surendira, Jithin George, Shreyas Kapale
January 29, 2026
cs.AI arXiv: 2601.22290Large Language Models demonstrate remarkable capabilities yet remain fundamentally probabilistic, presenting critical reliability challenges for enterprise
six sigma agent consensus voting task decomposition micro-agent sampling dynamic scaling llm systems enterprise-grade reliability error compounding exponential reliability gains dpmo atomic actions parallel execution embedding-based clustering voting judge heterogeneous execution temperature variation world state manager theoretical analysis enterprise use cases financial document processing customer support routing contract document analysis fault tolerance high-reliability organizations consensus-driven executioncs.AI -
Failing to Explore: Language Models on Interactive Tasks
Mahdi JafariRaviz, Keivan Rezaei, Arshia Soltani Moakhar, Zahra Sodagar, Yize Cheng, Soheil Feizi
January 29, 2026
cs.LG arXiv: 2601.22345We evaluate language models on their ability to explore interactive environments under a limited interaction budget. We introduce three parametric tasks with
language models interactive tasks exploration exploitation interaction budget hillsearch treesearch maxsatsearch parametric tasks controllable difficulty under-exploration premature commitment parallel execution periodic summarization explore-exploit strategies baseline comparisons performance scaling agentic ai black-box environments oracle queries reward maximization systematic evaluation lightweight interventions context summarization self-correctioncs.LG -
Business as Rulesual: A Benchmark and Framework for Business Rule Flow Modeling with LLMs
Chen Yang, Ruping Xu, Ruizhe Li, Bin Cao, Jing Fan
May 24, 2025
cs.CL arXiv: 2505.18542Extracting structured procedural knowledge from unstructured business documents is a critical yet unresolved bottleneck in process automation. While prior work has focused on
business rule extraction procedural knowledge extraction process automation logic-aware language models brex benchmark exide framework large language models rule flow modeling condition-action rules dependency graph reconstruction conditional branching parallel execution sequential dependencies prompt engineering executable grounding pseudo-code generation structure-aware reasoning chain-of-thought prompting reasoning-optimized models instruction-tuned models named entity recognition multiclass classification inter-annotator agreement intraclass correlation coefficient neural-symbolic reasoningcs.CL -
ParaCook: On Time-Efficient Planning for Multi-Agent Systems
Hai Zhao, Zouying Cao, Haobo Yuan, Shiqi Zhang, Zhuosheng Zhang, Ming-Hsuan Yang, Pengrui Lu, Xinbei Ma, Yunqing Xu, Tiancheng Shen
October 13, 2025
cs.AI arXiv: 2510.11608Large Language Models (LLMs) exhibit strong reasoning abilities for planning long-horizon, real-world tasks, yet existing agent benchmarks focus on task
multimodal large language models multi-agent systems collaborative planning time efficiency parallel execution asynchronous operations paracook benchmark task decomposition directed acyclic graph scheduling resource contention intra-agent parallelism inter-agent parallelism embodied agents grid world overcooked chain-of-thought success rate order completion time agent utilization movement distance abstract planning tasks coordination long-horizon reasoningcs.AI -
Tool-Aware Planning in Contact Center AI: Evaluating LLMs through Lineage-Guided Query Decomposition
Ayush Kumar, Varun Nathan, Shreyas Guha
February 16, 2026
cs.CL arXiv: 2602.14955We present a domain-grounded framework and benchmark for tool-aware plan generation in contact centers, where answering a query for business insights, our target use case
contact center ai large language models tool-aware planning query decomposition plan lineage text2sql retrieval-augmented generation agentic planning snowflake structured tools unstructured tools dependency graph parallel execution iterative refinement evaluator-optimizer loop benchmark data curation tool-prompt alignment step executability query adherence one-shot evaluation metric-wise evaluation claude-3-7-sonnet o3-mini data analyticscs.CL cs.SE -
SIT-LMPC: Safe Information-Theoretic Learning Model Predictive Control for Iterative Tasks
Rahul Mangharam, Truong X. Nghiem, Zirui Zang, Ahmad Amine, Nick-Marios T. Kokolakis, Ugo Rosolia
February 18, 2026
cs.RO arXiv: 2602.16187Robots executing iterative tasks in complex, uncertain environments require control strategies that balance robustness, safety, and high
model predictive control iterative learning control information-theoretic control stochastic systems nonlinear systems safe learning normalizing flows adaptive penalty method mppi optimal control robotics autonomous racing value function estimation constrained optimization real-time optimization gpu acceleration parallel execution uncertainty modeling robustness sampling-based control terminal cost function safe set cross-entropy method neural spline flows off-road vehiclescs.AI cs.RO eess.SY cs.SY -
DistributedEstimator: Distributed Training of Quantum Neural Networks via Circuit Cutting
Rajkumar Buyya, Adel N. Toosi, Prabhjot Singh
February 18, 2026
cs.DC arXiv: 2602.16233Circuit cutting decomposes a large quantum circuit into a collection of smaller subcircuits. The outputs of these subcircuits are then
quantum neural networks circuit cutting distributed training expectation value estimation classical reconstruction hybrid quantum-classical algorithms variational quantum algorithms qiskit distributed estimator subcircuit decomposition scheduling policies straggler mitigation parallel execution wall-clock time bottleneck analysis sampling complexity binary classification mnist iris dataset adversarial robustness parameter-shift rule staged pipeline tail latency quantum machine learning nisqquant-ph cs.DC cs.LG -
Framework of Thoughts: A Foundation Framework for Dynamic and Optimized Reasoning based on Chains, Trees, and Graphs
Felix Fricke, Simon Malberg, Georg Groh
February 18, 2026
cs.AI arXiv: 2602.16512Prompting schemes such as Chain of Thought, Tree of Thoughts, and Graph of Thoughts can significantly enhance the
framework of thoughts large language models prompting schemes chain of thought tree of thoughts graph of thoughts dynamic reasoning execution graph reasoning graph hyperparameter optimization prompt optimization parallel execution persistent caching probtree automated reasoning problem decomposition test-time compute inference cost reduction dspy optuna game of 24 hotpotqa musique multi-hop question answering foundation frameworkcs.AI -
Accelerating Mobile Inference through Fine-Grained CPU-GPU Co-Execution
Zhuojin Li, Marco Paolieri, Leana Golubchik
October 24, 2025
cs.LG arXiv: 2510.21081Deploying deep neural networks on mobile devices is increasingly important but remains challenging due to limited computing resources. On the other hand, their unified
mobile inference cpu-gpu co-execution deep neural networks unified memory architecture inference latency workload partitioning synchronization overhead shared virtual memory opencl latency prediction machine learning models gpu kernels dispatch times convolutional layers linear layers tensorflow lite xnnpack heterogeneous computing gradient-boosted decision trees lightgbm winograd algorithm cache coherence active polling parallel execution mobile platformscs.DC cs.LG cs.PF -
The Tragedy of Chain Commons
Ignacio Amores-Sesar, Mirza Ahad Baig, Seth Gilbert, Ray Neiheiser, Michelle X. Yeo
February 23, 2026
cs.DC arXiv: 2602.20341Byzantine Fault Tolerant (BFT) consensus forms the foundation of many modern blockchains striving for both high throughput and low
blockchain bft consensus decoupled design gaslighting attack resource capacity utilization reward distribution fairness partial coupling transaction execution modular blockchains throughput latency formal framework byzantine fault tolerance gas pricing mempool block creation execution bottleneck resource allocation strategic behavior dirty ledger parallel execution leader-based protocols cost enforceability state dependencies distributed systems securitycs.DC -
Search More, Think Less: Rethinking Long-Horizon Agentic Search for Efficiency and Generalization
Shu Xu, Ge Zhang, Jiayu Zhang, Jian Yang, Xin Gui, Minghao Liu, Wangchunshu Zhou, Piaohong Wang, Jiaheng Liu, He Zhu, Tianyu Zheng, Maojia Song, Yuqing Wang, Jiaqi Wu, Qiexiang Wang, Tianrui Qin, Yuchen Eleanor Jiang, Xinpeng Liu, Qianben Chen, King Zhu, Chengjun Yu, Jingyi Cao, Dingfeng Shi, Tiannan Wang
February 26, 2026
cs.CL arXiv: 2602.22675Recent deep research agents primarily improve performance by scaling reasoning depth, but this leads to high inference cost and latency in search-intensive scenarios
agentic search long-horizon reasoning smtl parallel evidence acquisition efficiency generalization data synthesis pipeline supervised fine-tuning reinforcement learning rloo context management browsecomp gaia xbench deepresearch bench knowledge graph subgraph extraction multi-hop reasoning tool-augmented llm plan refinement inference latency information density deterministic question answering open-ended research parallel executioncs.CL -
NovaLAD: A Fast, CPU-Optimized Document Extraction Pipeline for Generative AI and Data Intelligence
Aman Ulla
February 23, 2026
cs.CV arXiv: 2603.00122Document extraction is an important step before retrieval-augmented generation (RAG), knowledge bases, and downstream generative AI can work. It turns unstructured
document extraction retrieval-augmented generation yolov10 layout detection object detection vision transformer optical character recognition easyocr vision-language models cpu optimization parallel execution dp-bench tree edit distance-based similarity normalized indel distance document parsing knowledge graphs markdown conversion image classification semantic element detection reading order serialization rule-based grouping dbscan data intelligence unstructured data multimodal aics.AI cs.CV cs.IR -
OMNIINTENT: A Trusted Intent-Centric Framework for User-Friendly Web3
Zhuoran Pan, Yue Li, Zhi Guan, Jianbin Hu, Zhong Chen
March 04, 2026
cs.CR arXiv: 2603.04168The increasingly complex Web3 ecosystem and decentralized finance (DeFi) landscape demand ever higher levels of technical expertise and financial literacy from
web3 intent-centric decentralized finance defi domain-specific language icl trusted execution environment tee intel sgx enclave compiler transaction dependency graph parallel execution execution optimizer mempool-aware feasibility checker remote attestation smart contracts blockchain ethereum asset management liquidity provision staking nft trading throughput optimizationcs.CR -
HiMAP-Travel: Hierarchical Multi-Agent Planning for Long-Horizon Constrained Travel
The Viet Bui, Wenjun Li, Yong Liu
March 05, 2026
cs.AI arXiv: 2603.04750Sequential LLM agents fail on long-horizon planning with hard constraints like budgets and diversity requirements. As
himap-travel hierarchical multi-agent planning long-horizon planning constrained travel planning coordinator-executor architecture transactional monitor bargaining protocol group relative policy optimization grpo role conditioning travelplanner flextravelbench constraint drift parallel execution context isolation synchronized global state hard constraints resource allocation autonomous agents tool-augmented planning itinerary generation reinforcement learning multi-turn adaptation qwen3 combinatorial optimizationcs.CL cs.AI