arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2102.03705cs.ROcs.SYeess.SY
收藏

一种应用于机器人学的解析型分层深度学习框架

An Analytic Layer-wise Deep Learning Framework with Applications to Robotics

Nanyang Technological University(南洋理工大学);Yonsei University(延世大学)

Huu-Thiet Nguyen, Chien Chern Cheah, Kar-Ann Toh

AI总结:

该研究提出一种适用于全连接网络的解析型分层深度学习框架,含两种可分析收敛性的分层算法,在基准数据集上兼顾性能与可解释性,并成功应用于UR5e机器人在线运动学学习。

摘要:

深度学习(DL)已在诸多应用中取得巨大成功,但从理论角度对其的分析仍较为欠缺。黑箱深度学习模型难以解释的成功引发了科研人员的疑问,也推动了可解释人工智能(XAI)领域的兴起。在机器人学领域,由于机器人是需要与物理世界安全交互的主动智能体,以可预测、稳定的方式部署深度学习算法尤为重要。本文提出了一种适用于全连接神经网络的解析型深度学习框架,可同时应用于回归问题与分类问题,回归与分类问题的应用场景包括在线机器人控制与机器人视觉。我们提出了两种分层学习算法,可对学习系统的收敛性进行分析。首先,提出了一种适用于多层网络的逆分层学习算法,并对每一层进行了收敛性分析,以厘清分层深度学习的相关问题。其次,开发了一种前向渐进学习算法,通过使用单隐藏层网络逐步构建深度网络,以实现更高的精度。研究表明,从收敛性角度来看,渐进学习方法可用于权重微调。我们通过使用MNIST和CIFAR-10数据集的经典基准识别任务验证了所提框架的有效性,结果显示该框架在性能与可解释性之间实现了良好的平衡。随后,我们将所提方法应用于机器人运动学的在线学习,并给出了在模型未知的UR5e机器人运动控制上的实验结果。

英文摘要:

Deep learning (DL) has achieved great success in many applications, but it has been less well analyzed from the theoretical perspective. The unexplainable success of black-box DL models has raised questions among scientists and promoted the emergence of the field of explainable artificial intelligence (XAI). In robotics, it is particularly important to deploy DL algorithms in a predictable and stable manner as robots are active agents that need to interact safely with the physical world. This paper presents an analytic deep learning framework for fully connected neural networks, which can be applied for both regression problems and classification problems. Examples for regression and classification problems include online robot control and robot vision. We present two layer-wise learning algorithms such that the convergence of the learning systems can be analyzed. Firstly, an inverse layer-wise learning algorithm for multilayer networks with convergence analysis for each layer is presented to understand the problems of layer-wise deep learning. Secondly, a forward progressive learning algorithm where the deep networks are built progressively by using single hidden layer networks is developed to achieve better accuracy. It is shown that the progressive learning method can be used for fine-tuning of weights from convergence point of view. The effectiveness of the proposed framework is illustrated based on classical benchmark recognition tasks using the MNIST and CIFAR-10 datasets and the results show a good balance between performance and explainability. The proposed method is subsequently applied for online learning of robot kinematics and experimental results on kinematic control of UR5e robot with unknown model are presented.

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。

↑