Scholarlink

<p>Recent advances in AI culminate a shift in science and engineering away from strong reliance on algorithmic and symbolic knowledge towards new data-driven approaches. How does the emerging intelligent data-centric world impact research on real-time and embedded computing? We argue for two effects: (1) new challenges in embedded system contexts, and (2) new opportunities for community expansion beyond the embedded domain. First, <i>on the embedded system side</i>, the shifting nature of computing towards <i>data-centricity</i> affects the types of bottlenecks that arise. At training time, the bottlenecks are generally <i>data-related</i>. Embedded computing relies on <i>scarce</i> sensor data modalities, unlike those commonly addressed in mainstream AI, necessitating solutions for <i>efficient learning</i> from scarce sensor data. At inference time, the bottlenecks are <i>resource-related</i>, calling for <i>improved resource economy</i> and <i>novel scheduling policies</i>. Further ahead, the convergence of AI around large language models (LLMs) introduces additional <i>model-related</i> challenges in embedded contexts. Second, <i>on the domain expansion side</i>, we argue that community expertise in handling resource bottlenecks is becoming increasingly relevant to a new domain: the <i>cloud</i> environment, driven by AI needs. The paper discusses the novel research directions that arise in the data-centric world of AI, covering data-, resource-, and model-related challenges in embedded systems as well as new opportunities in the cloud domain.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The bottlenecks of AI: challenges for embedded and real-time research in a data-centric age

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月1日

查看全文
  • Tarek Abdelzaher,
  • Yigong Hu,
  • Denizhan Kara,
  • Tomoyoshi Kimura,
  • Ashitabh Misra,
  • Vishakha Ramani,
  • Olivier Tardieu,
  • Tianshi Wang,
  • Maggie Wigness,
  • Alaa Youssef

摘要

Recent advances in AI culminate a shift in science and engineering away from strong reliance on algorithmic and symbolic knowledge towards new data-driven approaches. How does the emerging intelligent data-centric world impact research on real-time and embedded computing? We argue for two effects: (1) new challenges in embedded system contexts, and (2) new opportunities for community expansion beyond the embedded domain. First, on the embedded system side, the shifting nature of computing towards data-centricity affects the types of bottlenecks that arise. At training time, the bottlenecks are generally data-related. Embedded computing relies on scarce sensor data modalities, unlike those commonly addressed in mainstream AI, necessitating solutions for efficient learning from scarce sensor data. At inference time, the bottlenecks are resource-related, calling for improved resource economy and novel scheduling policies. Further ahead, the convergence of AI around large language models (LLMs) introduces additional model-related challenges in embedded contexts. Second, on the domain expansion side, we argue that community expertise in handling resource bottlenecks is becoming increasingly relevant to a new domain: the cloud environment, driven by AI needs. The paper discusses the novel research directions that arise in the data-centric world of AI, covering data-, resource-, and model-related challenges in embedded systems as well as new opportunities in the cloud domain.