arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.03331cs.CVcs.AIcs.DC
收藏

面向跨场景在线推理的场景感知模型适应方案

A Scene-aware Models Adaptation Scheme for Cross-scene Online Inference on Mobile Devices

Department of Computer Science and Engineering, Shanghai Jiao Tong University(计算机科学与工程系,上海交通大学);Huawei Cloud(华为云);School of Computer Science and Technology, Donghua University(计算机科学与技术学院,东华大学)

Yunzhe Li, Hongzi Zhu, Zhuohong Deng, Yunlong Cheng, Zimu Zheng, Liang Zhang, Shan Chang, Minyi Guo

AI总结:

本文提出Anole方案,通过自适应选择场景特定DNN模型提升移动设备上跨场景在线推理的准确性、速度和能耗效率。

摘要:

随着人工智能物联网(AIoT)应用的发展,人们希望在移动设备上使用深度神经网络(DNN)模型进行在线预测。然而,由于设备的移动性,不熟悉的测试样本不断出现,显著影响了预训练DNN的预测准确性。此外,不稳定的网络连接要求进行本地模型推理。本文提出了一种轻量级方案,称为Anole,以应对移动设备上的本地DNN模型推理。Anole的核心思想是首先建立一支紧凑的DNN模型队伍,然后自适应地选择最适合当前测试样本的模型进行在线推理。关键在于自动识别适合模型的场景以训练场景特定的DNN模型。为此,我们通过结合人类启发法和特征相似性来分离场景,设计了一种弱监督的场景表示学习算法。此外,我们进一步训练了一个模型分类器,以预测每个测试样本的最佳适配场景特定的DNN模型。我们将在不同类型的移动设备上实现Anole,并基于无人机(UAVs)进行广泛的轨迹驱动和现实世界实验。结果表明,Anole在预测准确性(高4.5%)、响应时间(快33.1%)和能耗(低45.1%)方面优于使用通用大DNN的方法。

英文摘要:

Emerging Artificial Intelligence of Things (AIoT) applications desire online prediction using deep neural network (DNN) models on mobile devices. However, due to the movement of devices, unfamiliar test samples constantly appear, significantly affecting the prediction accuracy of a pre-trained DNN. In addition, unstable network connection calls for local model inference. In this paper, we propose a light-weight scheme, called Anole, to cope with the local DNN model inference on mobile devices. The core idea of Anole is to first establish an army of compact DNN models, and then adaptively select the model fitting the current test sample best for online inference. The key is to automatically identify model-friendly scenes for training scene-specific DNN models. To this end, we design a weakly-supervised scene representation learning algorithm by combining both human heuristics and feature similarity in separating scenes. Moreover, we further train a model classifier to predict the best-fit scene-specific DNN model for each test sample. We implement Anole on different types of mobile devices and conduct extensive trace-driven and real-world experiments based on unmanned aerial vehicles (UAVs). The results demonstrate that Anole outwits the method of using a versatile large DNN in terms of prediction accuracy (4.5% higher), response time (33.1% faster) and power consumption (45.1% lower).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑