arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1904.09201cs.CVcs.AI
收藏

在深度神经决策森林中可视化决策过程

Visualizing the decision-making process in deep neural decision forest

Hong Kong University of Science and Technology(香港科技大学)

Shichao Li, Kwang-Ting Cheng

摘要:

深度神经决策森林(NDF)通过结合决策树和深度表示学习,在各种视觉任务上取得了显着的性能。在这项工作中,我们首先跟踪该模型的决策过程,并可视化显着性图,以了解输入的哪一部分对分类和回归问题的影响更大。然后,我们将NDF应用于多任务坐标回归问题,并展示了路由概率的分布,这对于解释NDF是至关重要的,但没有显示回归问题。用于可视化的预训练模型和代码将在https://github.com/Nicholasli1995/VisualizingNDF上提供

英文摘要:

Deep neural decision forest (NDF) achieved remarkable performance on various vision tasks via combining decision tree and deep representation learning. In this work, we first trace the decision-making process of this model and visualize saliency maps to understand which portion of the input influence it more for both classification and regression problems. We then apply NDF on a multi-task coordinate regression problem and demonstrate the distribution of routing probabilities, which is vital for interpreting NDF yet not shown for regression problems. The pre-trained model and code for visualization will be available at https://github.com/Nicholasli1995/VisualizingNDF

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。

↑