-
Notifications
You must be signed in to change notification settings - Fork 0
test
双线性汇合(bilinear pooling)在细粒度图像分析及其他领域的进展综述
原创: 张皓 SigAISIGAI推荐SIGAI 资源大汇总
作者简介:张皓
南京大学计算机系机器学习与数据挖掘所(LAMDA)
研究方向:计算机视觉和机器学习(视觉识别和深度学习)
个人主页:goo.gl/N715YT
细粒度图像分类旨在同一大类图像的确切子类。由于不同子类之间的视觉差异很小,而且容易受姿势、视角、图像中目标位置等影响,这是一个很有挑战性的任务。因此,类间差异通常比类内差异更小。双线性汇合(bilinear pooling)计算不同空间位置的外积,并对不同空间位置计算平均汇合以得到双线性特征。外积捕获了特征通道之间成对的相关关系,并且这是平移不变的。双线性汇合提供了比线性模型更强的特征表示,并可以端到端地进行优化,取得了和使用部位(parts)信息相当或甚至更高的性能。
在本文,我们将对使用双线性汇合进行细粒度分类的方法发展历程予以回顾。研究方向大致分为两类:设计更好的双线性汇合过程,以及精简双线性汇合。其中,对双线性汇合过程的设计主要包括对汇合结果规范化过程的选择及其高效实现,以及融合一阶和二阶信息。精简双线性汇合设计大致有三种思路:利用PCA降维、近似核计算、以及低秩双线性分类器。此外,双线性汇合的思想也被用于其他计算机视觉领域,例如风格迁移、视觉问答、动作识别等。我们也将介绍双线性汇合在这些领域的应用。
1. 数学准备
在本节,我们介绍本文用要用到的符号和在推导时需要的数学性质。
1.1 符号
深度描述向量(deep
descriptor)。其中1<=i<=HW
。例如对VGG-16网络,我们通常使用relu5-3层的特征提取图像的深度描述向量,那么H=W=14,D=512。
描述矩阵(descriptor
matrix)。定义为
均值向量(mean vector)μ∈RD。定义为
格拉姆矩阵(Gram
matrix)。定义为
协方差矩阵(covariance
matrix)。定义为
其中
是中心化矩阵(centering matrix)。
分数向量(score
vector),softma层的输入,k是分类任务的类别数。
1.2 数学性质
由于双线性汇合相关的论文涉及许多矩阵运算,尤其是迹运算。如下性质在推导时将有帮助。这些性质在使用时即可以从左向右使用,也可以从右向左使用。
向量化操作的迹等价
弗罗贝尼乌斯范数(Frobenius norm)的迹等价
矩阵迹满足交换率和结合率
矩阵转置不改变迹
矩阵幂等价于特征值分别取幂
1.3 双线性
对函数分f(x,y),双线性(bilinear)是指当固定其中一个参数(例如x)时,f(x,y)对另一个参数(例如y)是线性的。在这里,研究的双线性函数是形如f(x,y)=XTAy这样的形式。本文关注的双线性汇合叫双线性这个名字是受历史的影响,在早期两个分支是不同的,现在主流做法是两个分支使用相同的输入,整个操作将变为非线性而不是双线性,但这个名称沿用至今。
2. 双线性汇合
双线性汇合在深度学习复兴前就已经被提出,随后,在细粒度图像分类领域得到了广泛使用。本节将介绍双线性汇合及其变体。
2.1 细粒度分类中的双线性汇合
Tsung-Yu Lin, Aruni RoyChowdhury, and Subhransu Maji. Bilinear CNN models for fine-grained visual recognition. ICCV 2015: 1449-1457.
Tsung-Yu Lin, Aruni RoyChowdhury, and Subhransu Maji. Bilinear convolutional neural networks for fine-grained visual recognition. TPAMI 2018, 40(6): 1309-1322.
双线性汇合操作通过计算深度描述向量的格拉姆矩阵G捕获特征通道之间成对的相关关系。随后,将格拉姆矩阵展成向量
并进行规范化(normalization)
得到最终的双线性特征。
在原文中,Lin等人使用了两个不同的网络得到双线性汇合的不同分支,动机是希望一个分支学到位置(where)信息,而另一个分支学到外观(what)信息。事实上,这是不可能的。
Mohammad Moghimi, Serge J. Belongie, Mohammad J. Saberian, Jian Yang, Nuno Vasconcelos, and Li-Jia Li. Boosted convolutional neural networks. BMVC 2016.
Moghimi等人提出BoostCNN,利用多个双线性CNN的boosting集成来提升性能,通过最小二乘目标函数,学习boosting权重。然而这会使得训练变慢两个量级。
Tsung-Yu Lin and Subhransu Maji. Improved bilinear pooling with CNNs. BMVC 2017.
Lin和Maji探索了对格拉姆矩阵不同的规范化方案,并发现对格拉姆矩阵进行0.5矩阵幂规范化压缩格拉姆矩阵特征值的动态范围
后结合逐元素平方开根和规范化可以得到2-3
%的性能提升。此外,由于GPU上没有SVD的高效实现,Lin和Maji使用牛顿迭代法的变体计算G1/2 ,并通过解李雅普诺夫方程(Lyapunov
equation)来估计G1/2的梯度进行训练。
Peihua Li, Jiangtao Xie, Qilong Wang, and Wangmeng Zuo. Is second-order information helpful for large-scale visual recognition? ICCV 2017: 2089-2097.
Li等人提出MPN-COV,其对深度描述向量的协方差矩阵∑进行0.5矩阵幂规范化
以得到双线性汇合特征。
Peihua Li, Jiangtao Xie, Qilong Wang, and Zilin Gao. Towards faster training of global covariance pooling networks by iterative matrix square root normalization. CVPR 2018: 947-955.
由于在GPU上没有特征值分解和SVD的高效实现,相比Lin和Maji在反向解李雅普诺夫方程时仍需进行舒尔分解(Schur decomposition)或特征值分解,Li等人前向和反向过程都基于牛顿迭代法,来对矩阵进行0.5矩阵幂规范化。
![](data:image/jpeg;base64,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