Organizations: School of Computer and Artificial Intelligence, Zhengzhou University, Zhengzhou, 450001, Henan, China · Engineering Research Center of Intelligent Swarm Systems, Ministry of Education, Zhengzhou, 450001, Henan, China · National Supercomputing Center in Zhengzhou, Zhengzhou, 450001, Henan, China · Henan Research Center for Large Model Technology and New Quality Software Engineering, Zhengzhou, 450001, Henan, China
Emergency situations in scheduling systems often trigger local functional failures that undermine system stability and even cause system collapse. Existing methods primarily rely on robust scheduling or reactive scheduling, handling emergencies through predefined rules or rescheduling strategies. However, the diversity and unpredictability of real-world emergencies make them difficult to anticipate, which limits the adaptability of these methods in complex scenarios. Recent studies have shown that Large Language Models (LLMs) possess strong potential for complex scheduling tasks because of their extensive prior knowledge and strong reasoning capabilities. Nevertheless, the high inference latency of LLMs and the lengthy contextual information of scheduling systems significantly hinder their application for emergency handling. To mitigate these issues, we propose the Multi-agent Driven Formal Instruction Generation Framework (MAFIG). The framework constrains the decision scope to local functional modules affected by emergency situations and repairs scheduling logic rapidly by generating formal instructions. MAFIG contains a Perception Agent and an Emergency Decision Agent, which mitigates the adverse impact of lengthy system contexts on emergency decision-making. We further introduce span-focused loss-driven local distillation mechanism (SFL) to transfer the decision-making capability of powerful Cloud Large Language Models (C-LLMs) to lightweight local models, reducing inference latency while preserving decision-making effectiveness. Experiments in the Port, Warehousing, and Deck scheduling datasets show success rates of 98.49%, 94.97%, and 97.50%, with average processing times of 0.33 s, 0.23 s, and 0.19 s. These results demonstrate that MAFIG effectively mitigates the impact of emergencies and improves the robustness and adaptability of scheduling systems.
Figures & tables
Figure 1: Accident types in Mediterranean port areas.
Figure 2: Architecture of MAFIG for emergency decision in scheduling systems. The framework consists of the Perception Agent, the Emergency Decision Agent and the Atomic Function Library. It supports semantic parsing, impact analysis, affected function localization and atomic function revision, which enables rapid recovery of scheduling logic under emergency situations.
Variable
Description
t
Current time step
st
Scheduling environment state at time t
et
Emergency situations at time t
x
Aggregated system context (system rules, module specifications, code context)
fs
Function specifications
DA
Atomic function library
Table 1: Description of variables and their corresponding meanings.
Figure 5: Overview of the EvalDeck evaluation datasets.
Task
Models
Total Time (s)
Avg Time (s)
Success Rate
Port
Qwen3-32B
1763.73
8.86
62.31%
Qwen3-Coder-480B-A35B-Instruct
896.53
4.51
85.93%
DeepSeek-V3.2
1827.28
9.18
86.43%
GLM-4.7
5460.58
27.44
89.95%
Qwen2.5-Coder-7B
69.20
0.35
25.62%
MAFIG(Qwen2.5-Coder-7B)
64.98
0.33
98.49%
Table 2: Performance comparison of different models across tasks.
Task
Model
Method
Total Time (s)
Avg Time (s)
Success Rate
Port
Qwen2.5-Coder-7B
—
69.20
0.35
25.62%
Meta-Llama-3.1-8B-Instruct
—
101.39
0.51
30.15%
GLM-4-9B-Chat
—
94.89
0.48
23.62%
Qwen2.5-Coder-7B
MAFIG(w/o SFL)
63.19
0.32
62.81%
Meta-Llama-3.1-8B-Instruct
MAFIG(w/o SFL)
71.06
0.36
48.24%
GLM-4-9B-Chat
MAFIG(w/o SFL)
86.11
0.43
64.82%
Table 3: Evaluating the Effect of MAFIG Architecture on Backbone Model Performance.
Task
Model
Method
Perception Time (s)
Decision Time (s)
Total Time (s)
Avg Time (s)
Accuracy
Port
Qwen2.5-Coder-7B
—
—
—
69.20
0.35
25.62%
MAFIG(w/o SFL)
18.25
44.94
63.19
0.32
62.81%
MAFIG
18.63
46.35
64.98
0.33
98.49%
Warehousing
Qwen2.5-Coder-7B
—
—
—
151.18
0.38
25.87%
MAFIG(w/o SFL)
23.02
64.85
87.87
0.22
55.53%
MAFIG
22.08
68.19
90.27
0.23
94.97%
Table 4: Latency and Success Rate Analysis of MAFIG With Versus Without SFL Across Different Tasks.
Figure 6: Performance comparison of SFL and LoRA under different training set sizes across tasks.
Figure 7: Case study of MAFIG in the deck scheduling scenario under concurrent emergency situations.
The Dynamic Flexible Job Shop Scheduling Problem (DFJSP) necessitates a trade-off between instant reaction to stochastic disturbances and global optimization of production goals. Conventional priority rules are insufficiently flexible to handle complex disruptions, whereas learning-based approaches often compromise interpretability or fail to generalize across problem scales. Although Large Language Models (LLMs) offer advanced reasoning capabilities to bridge this gap, their substantial inference latency is incompatible with the millisecond-level decision cycles of industrial control systems. To resolve this conflict, we introduce RACE-Sched, an asynchronous agent-based framework that decouples policy execution from logical reasoning via a dual-stream architecture. The Reactive Stream executes low-latency symbolic heuristics to enable real-time dispatching, while the parallel Deliberative Stream leverages an LLM to synthesize, validate, and evolve these rules. Candidate rules undergo rigorous testing in a sandbox and are deployed via atomic updates, ensuring safety without blocking the control loop. Additionally, a semantic rule repository indexes validated heuristics for retrieval-based initialization which enhances transferability across problem scales. Extensive evaluations on GEN-Bench, MK-Bench, and JMS-Bench demonstrate that RACE-Sched outperforms leading Deep Reinforcement Learning and other LLM-based baselines. This approach harmonizes real-time constraints with long-horizon reasoning to achieve superior solution quality and robust adaptation to dynamic events.
Shijie Cao, Yuan Yuan, Jing Liu
School of Computer Science and Engineering, Beihang University, Beijing 100191, China · Shenzhen Loop Area Institute, Shenzhen, China · Qingdao Research Institute, Beihang University +3
Large language models (LLMs) promise to accelerate incident response in production systems, yet single-agent approaches generate vague, unusable recommendations. We present MyAntFarm.ai, a reproducible containerized framework demonstrating that multi-agent orchestration fundamentally transforms LLM-based incident response quality. Through 348 controlled trials comparing single-agent copilot versus multi-agent systems on identical incident scenarios, we find that multi-agent orchestration achieves 100% actionable recommendation rate versus 1.7% for single-agent approaches, an 80 times improvement in action specificity and 140 times improvement in solution correctness. Critically, multi-agent systems exhibit zero quality variance across all trials, enabling production SLA commitments impossible with inconsistent single-agent outputs. Both architectures achieve similar comprehension latency (approx.40s), establishing that the architectural value lies in deterministic quality, not speed. We introduce Decision Quality (DQ), a novel metric capturing validity, specificity, and correctness properties essential for operational deployment that existing LLM metrics do not address. These findings reframe multi-agent orchestration from a performance optimization to a production-readiness requirement for LLM-based incident response. All code, Docker configurations, and trial data are publicly available for reproduction.
Efficient task scheduling in heterogeneous computing systems remains a fundamental challenge in embedded and high-level synthesis flows. This paper presents MoSAIC (Motif-aware Scheduling with AI-driven Cognition), a learning-based framework that bridges the gap between exact and heuristic scheduling approaches for DAG-based applications. The system uses the reasoning capabilities of large language models (LLMs) to learn the relationships between DAG structures and scheduling priority functions. The framework first leverages the OR-Tools CP-SAT solver [4] to generate optimal schedules, which are then used to supervise the learning of compact priority-based scheduling functions through Bayesian Optimization. These learned heuristics capture relationships between scheduling decisions and structural graph features such as rank, depth, and fanout, enabling the construction of a reusable knowledge base across diverse DAG instances. Experimental results across synthetic graph families and real-world benchmarks, including PolyBench and MachSuite, show that the learned heuristics consistently improve over classical approaches such as HEFT [2], while maintaining low runtime overhead. The results indicate that combining optimal supervision with learning-based generalization provides a practical approach for scalable and adaptive task scheduling.
Shiva Ahir, Hashmath Shaik, Alex Doboli +2
Department of Electrical & Computer Engineering, Stony Brook University, New York, United States of America · Barcelona Supercomputing Center, Barcelona, Catalonia, Spain · Department of Computer Science, Hofstra University, New York, United States of America