EPNet: Learning to Exit with Flexible Multi-Branch Network
EPNet: Learning to Exit with Flexible Multi-Branch Network
复制标题
EPNet：学习通过灵活的多分支网络退出
DOI：
10.1145/3340531.3411973
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Guo, Tian
中科院分区：
文献类型：
--
作者：
Dai, Xin;Kong, Xiangnan;Guo, Tian
动态推理是一种新兴的技术，可以在资源受限的场景下降低深度神经网络的计算成本，例如在移动的设备上进行推理。实现动态推理的一种方法是利用多分支神经网络，通过遵循不同的分支对输入数据应用不同的计算。传统的多分支神经网络的研究主要以提高每个分支的精度为目标，并采用人工设计的规则来决定哪一个输入跟随网络的哪一个分支。此外，这些网络往往提供少量出口，限制了它们适应外部变化的能力。在本文中，我们研究的问题，设计一个灵活的多分支网络和早期退出的政策，可以适应资源消耗，以个人的推理请求，而不影响推理精度。我们提出了一个轻量级的分支结构，也提供了细粒度的提前退出的灵活性，并利用马尔可夫决策过程（MDP）自动学习提前退出策略。我们提出的模型，EPNet，是有效的，在不影响准确性的情况下，通过选择最合适的分支出口减少推理成本。我们还观察到，与最先进的方法相比，EPNet在推理预算的情况下实现了高3%的准确性。
Dynamic inference is an emerging technique that reduces the computational cost of deep neural network under resource-constrained scenarios, such as inference on mobile devices. One way to achieve dynamic inference is to leverage multi-branch neural networks that apply different computation on input data by following different branches. Conventional research on multi-branch neural networks mainly targeted at improving the accuracy of each branch, and use manually designed rules to decide which input follows which branch of the network. Furthermore, these networks often provide a small number of exits, limiting their ability to adapt to external changes. In this paper, we investigate the problem of designing a flexible multi-branch network and early-exiting policies that can adapt to the resource consumption to individual inference request without impacting the inference accuracy. We propose a lightweight branch structure that also provides fine-grained flexibility for early-exiting and leverage Markov decision process (MDP) to automatically learn the early-exiting policies. Our proposed model, EPNet, was effective in reducing inference cost without impacting accuracy by choosing the most suitable branch exit. We also observe that EPNet achieved 3% higher accuracy with an inference budget, compared to state-of-the-art approaches.
能够使用深度学习从面部照片中区分阿尔茨海默病
DOI：
--
发表时间：
2020
期刊：
影响因子：
--
作者：
Manabu Kokubo;Akihiro Hirashiki;Takahiro Kamihara;Atsuya Shimizu;Hidenori Arai;亀山祐美,亀山征史,深澤誠,飯塚友道,飯島勝矢,田中友規,矢可部満隆,小島太郎,小川純人,秋下雅弘
通讯作者：
亀山祐美,亀山征史,深澤誠,飯塚友道,飯島勝矢,田中友規,矢可部満隆,小島太郎,小川純人,秋下雅弘