arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2007.00062cs.CVcs.CGcs.LG
收藏

深特征空间的几何透视

Deep Feature Space: A Geometrical Perspective

Democritus University of Thrace(色雷斯德谟克利特大学)

Ioannis Kansizoglou, Loukas Bampis, Antonios Gasteratos

摘要:

神经网络(NN)最突出的属性之一是它们能够学习从高维数据(如图像)中提取鲁棒和描述性特征。因此,这种能力使得它们在大量的现代推理系统中作为特征提取器被特别频繁地利用。它们的应用范围主要包括复杂的级联任务,如多模态识别和深度强化学习(RL)。然而,神经网络引起的隐式偏见是难以避免或处理,并没有满足在传统的图像描述符。此外,缺乏描述层内属性的知识-因此,它们的一般行为-限制了提取的特征的进一步适用性。在此基础上,提出了一种在神经网络输出层之前可视化和理解向量空间的新方法,旨在启发分类任务下深层特征向量的性质。主要关注的是在特征空间中的过拟合的性质及其对进一步开发的不利影响。我们提出的研究结果,可以从我们的模型的制定,我们评估他们在现实的识别场景,证明其突出的改善所获得的结果。

英文摘要:

One of the most prominent attributes of Neural Networks (NNs) constitutes their capability of learning to extract robust and descriptive features from high dimensional data, like images. Hence, such an ability renders their exploitation as feature extractors particularly frequent in an abundant of modern reasoning systems. Their application scope mainly includes complex cascade tasks, like multi-modal recognition and deep Reinforcement Learning (RL). However, NNs induce implicit biases that are difficult to avoid or to deal with and are not met in traditional image descriptors. Moreover, the lack of knowledge for describing the intra-layer properties -- and thus their general behavior -- restricts the further applicability of the extracted features. With the paper at hand, a novel way of visualizing and understanding the vector space before the NNs' output layer is presented, aiming to enlighten the deep feature vectors' properties under classification tasks. Main attention is paid to the nature of overfitting in the feature space and its adverse effect on further exploitation. We present the findings that can be derived from our model's formulation, and we evaluate them on realistic recognition scenarios, proving its prominence by improving the obtained results.

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。

↑