日本フィジカルAI新聞

世界のフィジカルAIを、日本語で。

群制御arXiv:2605.02965

分散データセンター向けのエネルギー管理と連携AIGCワークロードスケジューリング:拡散モデル支援型報酬整形アプローチ

Joint Energy Management and Coordinated AIGC Workload Scheduling for Distributed Data Centers: A Diffusion-Aided Reward Shaping Approach

シェア:XThreadsFacebookLINEはてブBluesky

AIGCサービス提供者がデータセンターのエネルギーコストを削減しつつ高品質なコンテンツ生成を保証するため、エネルギー管理とワークロードスケジューリングを統合する枠組みを提案し、報酬のスパース性を拡散モデルによる報酬整形で解決する。

詳しい要約

1. どんなもの?

- AIGC workload scheduling for distributed data centers. - Joint energy management and coordinated scheduling framework. - Aims to reduce energy costs while ensuring high-quality content generation. - Addresses model heterogeneity, implicit service quality, and complex inference control. - Uses diffusion model-aided reward shaping for DRL.

2. 先行研究と比べてどこがすごい?

- Explicit mathematical characterization of service quality. - Promotes job transfer among ASPs and fine-grained inference configuration. - Joint consideration of various energy resources for power flexibility. - Overcomes reward sparsity in DRL via diffusion-based reward shaping. - Achieves superior convergence and system utility compared to benchmarks.

3. 技術・手法の肝は?

- System utility maximization problem balancing revenue, penalties, and costs. - Diffusion model-aided reward shaping synthesizes complementary rewards via multi-step denoising. - Integrated with DRL for efficient policy learning under sparse feedback. - Explicit service quality characterization enables job transfer and inference configuration. - Joint energy management enhances power usage flexibility.

4. どうやって有効だと検証した?

- Experiments based on real-world models and datasets. - Demonstrates effective accommodation of electricity price fluctuations and AIGC model heterogeneity. - Shows superior learning convergence and system utility compared with benchmark methods.

5. 議論はある?

- 要旨からは不明

6. 次に読むべき論文は?

- 要旨で参照/比較されている研究は明示されていない。 - 関連手法としてDRL, diffusion model, reward shaping, energy management in data centersが挙げられる。

※ AIが要旨から生成した要約です。正確性は原文をご確認ください。

著者: Yang Fu, Peng Qin, Liming Chen, Zihao Zhang, Hao Yu, Yifei Wang

分類: cs.LG, eess.SP, eess.SY, stat.ML

原文アブストラクト

Artificial intelligence-generated content (AIGC) has emerged as a transformative paradigm for automating the creation of diverse and customized content, giving rise to rapidly growing computational workloads in cloud data centers. It is imperative for AIGC service providers (ASPs) to strategically schedule AIGC workloads to reduce data center energy costs while guaranteeing high-quality content generation. However, the distinctive characteristics of AIGC services pose critical challenges, including model heterogeneity across ASPs, implicit service quality evaluation, and complex inference process control. To tackle these challenges, we propose a joint energy management and coordinated AIGC workload scheduling framework, which introduces an explicit mathematical characterization of service quality to promote both job transfer among ASPs and fine-grained inference process configuration. Moreover, various energy resources within data centers are jointly considered to enhance power usage flexibility. Subsequently, a system utility maximization problem is formulated to balance AIGC service revenue with operational penalties and costs. Nevertheless, the strong coupling among job scheduling decisions induces severe reward sparsity, which limits the effectiveness of existing deep reinforcement learning (DRL) algorithms. To address this issue, we develop a diffusion model-aided reward shaping approach to synthesize complementary reward signals through a multi-step denoising process. This approach is seamlessly integrated with DRL to enable efficient learning of scheduling policies under sparse environmental feedback. Experiments based on real-world models and datasets demonstrate that our scheme effectively accommodates electricity price fluctuations and AIGC model heterogeneity, while achieving superior learning convergence and system utility compared with benchmark methods.

関連論文

  1. 1Boston Dynamics、車輪付きAtlasの研究試作機を紹介研究・技術
  2. 2GXOとApptronik、Apolloの物流現場試験を拡大 最大1,000台の展開を構想量産・展開
  3. 3newmoとKDDIが資本業務提携 2028年に堺市で自動運転タクシー商用化目指す提携・買収
  4. 4NVIDIA、CoRLでGR00T 2とIsaac Lab 3.0を発表新製品・発表
  5. 5UC BerkeleyとNVIDIAの研究者、AIエージェントでロボット制御コードを仮想環境で改善研究・技術

PR・アフィリエイト広告

MINISFORUM(ミニPC)

MINISFORUM(ミニPC)の公式広告バナーMINISFORUM公式ストアを見る ↗
PR本紙発行元 EmplifAI