[论文解读] From Cloud to Edge: Rethinking Generative AI for Low-Resource Design Challenges

Sai Krishna Revanth Vuruma, Ashley Margetts+3 more|arXiv (Cornell University)|Feb 20, 2024

Manufacturing Process and Optimization被引用 5

一句话总结

本文主张在资源受限环境中离线、边缘部署的生成式AI解决方案用于设计，并探讨压缩、边缘计算，以及受 TinyML 启发的方法，以在发展中国家实现实际应用。

ABSTRACT

Generative Artificial Intelligence (AI) has shown tremendous prospects in all aspects of technology, including design. However, due to its heavy demand on resources, it is usually trained on large computing infrastructure and often made available as a cloud-based service. In this position paper, we consider the potential, challenges, and promising approaches for generative AI for design on the edge, i.e., in resource-constrained settings where memory, compute, energy (battery) and network connectivity may be limited. Adapting generative AI for such settings involves overcoming significant hurdles, primarily in how to streamline complex models to function efficiently in low-resource environments. This necessitates innovative approaches in model compression, efficient algorithmic design, and perhaps even leveraging edge computing. The objective is to harness the power of generative AI in creating bespoke solutions for design problems, such as medical interventions, farm equipment maintenance, and educational material design, tailored to the unique constraints and needs of remote areas. These efforts could democratize access to advanced technology and foster sustainable development, ensuring universal accessibility and environmental consideration of AI-driven design benefits.

研究动机与目标

• 激励并概述在偏远且资源受限的环境中支持设计所需的离线、边缘启用的生成式AI。
• 识别将大型多模态模型适配到具有有限内存、计算能力、能源和连通性的边缘设备所面临的挑战。
• 调研并提出适用于低资源设计任务的模型压缩、边缘计算和评估指标的实际方法。

提出的方法

• 评阅当前的多模态和生成式AI模型（LLMs、扩散模型、VLMs）及其在边缘部署方面的局限性。
• 讨论包括剪枝、量化和知识蒸馏在内的模型压缩技术，这些技术与 LLMs 及扩散/VLMs 相关。
• 描述边缘计算和 TinyML 概念，作为离线、低资源在设备上的 AI 的推动因素。
• 提出适用于离线、低资源设计任务的评估指标，强调低成本与性能。
• 概述潜在解决方案，包括硬件-软件协同设计、分布式计算和本地数据中心。

Figure 1: Images and captions generated by CoDi given a wooden greenhouse design request.

实验结果

研究问题

• RQ1 在资源受限环境中将当前生成式AI模型部署到边缘设备的关键障碍是什么？
• RQ2 哪些模型压缩与边缘计算策略最能实现实际的离线生成式设计工具？
• RQ3 离线、在地训练的模型如何在偏远社区支持设计任务（医疗、农业、教育），同时确保可持续性与可访问性？

主要发现

• 在本地数据上训练的离线ML模型可以让设计工具针对独特的区域制约进行定制。
• 剪枝、量化和蒸馏等压缩技术可以降低模型规模和边缘部署的推理需求。
• 边缘计算和 TinyML 概念对于在偏远地区实现响应迅速、私密且低带宽的AI至关重要。
• 硬件进步、定制加速器和分布式计算的结合可以缓解离线部署生成模型时的资源限制。
• 需要合适的评估指标，以反映低资源设计情境和实际性能。

Figure 2: Images and captions generated by CoDi given a design replacement intent.