A Learning Method with Gap-Aware Generation for Heterogeneous DAG Scheduling

2026-03-24

AI Summary

This work addresses the challenges of DAG scheduling in heterogeneous environments, where resource constraints, task dependencies, and the need for rapid schedule generation often lead traditional methods—limited by restrictive construction strategies—into suboptimal solutions. To overcome this, we propose WeCAN, an end-to-end reinforcement learning framework that generates complete schedules via a single two-stage forward pass: first predicting task–resource pool scores and global parameters, then constructing the final mapping without repeated network invocations. We introduce a novel order-space analysis based on feasible scheduling sequences to diagnose suboptimality gaps and design a skip-extended mechanism with analytically parameterized decreasing skip rules, expanding the reachable schedule space while preserving inference efficiency. Integrating a weighted cross-attention encoder with compatibility-coefficient gating to model task–resource interactions, and embedding list-scheduling principles for scale-invariant generation, WeCAN significantly outperforms strong baselines on both synthetic compute graphs and real-world TPC-H DAGs, achieving shorter makespans, inference speeds comparable to classical heuristics, and faster execution than multi-round neural schedulers.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Efficient scheduling of directed acyclic graphs (DAGs) in heterogeneous environments is challenging due to resource capacities and dependencies. In practice, the need for adaptability across environments with varying resource pools and task types, alongside rapid schedule generation, complicates these challenges. We propose WeCAN, an end-to-end reinforcement learning framework for heterogeneous DAG scheduling that addresses task--pool compatibility coefficients and generation-induced optimality gaps. It adopts a two-stage single-pass design: a single forward pass produces task--pool scores and global parameters, followed by a generation map that constructs schedules without repeated network calls. Its weighted cross-attention encoder models task--pool interactions gated by compatibility coefficients, and is size-agnostic to environment fluctuations. Moreover, widely used list-scheduling maps can incur generation-induced optimality gaps from restricted reachability. We introduce an order-space analysis that characterizes the reachable set of generation maps via feasible schedule orders, explains the mechanism behind generation-induced gaps, and yields sufficient conditions for gap elimination. Guided by these conditions, we design a skip-extended realization with an analytically parameterized decreasing skip rule, which enlarges the reachable order set while preserving single-pass efficiency. Experiments on computation graphs and real-world TPC-H DAGs demonstrate improved makespan over strong baselines, with inference time comparable to classical heuristics and faster than multi-round neural schedulers.

Problem

• heterogeneous DAG scheduling
• optimality gap
• generation-induced gap
• task-pool compatibility
• schedule generation

Innovation

• gap-aware generation
• heterogeneous DAG scheduling
• order-space analysis
• single-pass reinforcement learning
• compatibility-aware attention