An Effort to Democratize Networking Research in the Era of AI/ML
An Effort to Democratize Networking Research in the Era of AI/ML
复制标题
AI/ML 时代网络研究民主化的努力
DOI：
10.1145/3365609.3365857
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
W. Willinger
中科院分区：
文献类型：
--
作者：
Arpit Gupta;Chris Mac;W. Willinger
当今网络社区越来越担心的是，随着人工智能/机器学习（AI/ML）技术的普及，缺乏对现实世界生产网络的访问使学术研究人员处于明显的劣势。事实上，与少数几个可以利用其全球规模的生产网络来开发和评估学习模型的工业研究小组相比，学术研究人员不仅难以获得真实世界的数据集，而且发现几乎不可能在现实条件下充分训练和评估他们的学习模型。在本文中，我们认为，当适当的仪器和妥善管理，企业网络的形式，大学或校园网络可以作为现实世界的生产网络，并可以，因为它们的无处不在，有助于创造一个更公平的竞争环境，为学术研究人员。尽管它们有各种限制，但作为现实世界的生产网络，这样的企业网络可以（i）作为一些丰富数据的独特来源，使这些研究人员能够影响或推进当前最先进的人工智能/机器学习网络，（ii）也可以作为急需的测试床，在那里可以评估或“道路测试”新开发的基于人工智能/机器学习的工具在实际部署到生产网络之前。我们讨论了校园网络的双重角色所带来的新的研究挑战，并评论了我们的建议为学术和行业研究人员提供的机会，使他们能够从各自生产环境的优势和局限性中受益，共同寻求推进基于AI/ML的工具的开发和评估，使其能够在实践中部署。
A growing concern within today's networking community is that with the proliferation of Artificial Intelligence/Machine Learning (AI/ML) techniques, a lack of access to real-world production networks is putting academic researchers at a significant disadvantage. Indeed, compared to a select few research groups in industry that can leverage access to their global-scale production networks in their data-driven efforts to develop and evaluate learning models, academic researchers not only struggle to get their hands on real-world data sets but find it almost impossible to adequately train and assess their learning models under realistic conditions. In this paper, we argue that when appropriately instrumented and properly managed, enterprise networks in the form of university or campus networks can serve as real-world production networks and can, because of their ubiquity, help create a more level playing field for academic researchers. Their various limitations notwithstanding, as real-world production networks, such enterprise networks can (i) serve as unique sources for some of the rich data that will enable these researchers to influence or advance the current state-of-the-art in AI/ML for networking and (ii) also function as much-needed test beds where newly developed AI/ML-based tools can be evaluated or "road-tested" prior to their actual deployment in the production network. We discuss new research challenges that arise from this proposed dual role of campus networks and comment on the opportunities our proposal affords for both academic and industry researchers to benefit from the advantages and limitations of their respective production environments in their common quest to advance the development and evaluation of AI/ML-based tools to the point where they can be deployed in practice.