Towards Faster Training of Global Covariance Pooling Networks by Iterative Matrix Square Root Normalization

Abstract

Global covariance pooling in convolutional neural networks has achieved impressive improvement over the classical first-order pooling. Recent works have shown matrix square root normalization plays a central role in achieving state-of-the-art performance. However, existing methods depend heavily on eigendecomposition (EIG) or singular value decomposition (SVD), suffering from inefficient training due to limited support of EIG and SVD on GPU. Towards addressing this problem, we propose an iterative matrix square root normalization method for fast end-toend training of global covariance pooling networks. At the core of our method is a meta-layer designed with loopembedded directed graph structure. The meta-layer consists of three consecutive nonlinear structured layers, which perform pre-normalization, coupled matrix iteration and post-compensation, respectively. Our method is much faster than EIG or SVD based ones, since it involves only matrix multiplications, suitable for parallel implementation on GPU. Moreover, the proposed network with ResNet architecture can converge in much less epochs, further accelerating network training. On large-scale ImageNet, we achieve competitive performance superior to existing counterparts. By finetuning our models pre-trained on ImageNet, we establish state-of-the-art results on three challenging finegrained benchmarks. The source code and network models will be available at http://www.peihuali.org/iSQRT-COV.

Looking for the Devil in the Details: Learning Trilinear Attention Sampling Network for Fine-Grained Image Recognition

“…which is a novel self-supervision mechanism to effectively localize informative regions without the need of bounding-box/part annotations. • iSQRT-COV [18]: Towards faster training of global covariance pooling networks by iterative matrix square root normalization. Implementation: We used open-sourced MXNet [4] as our code-base, and trained all the models on 8 Tesla P-100 GPUs.…”

Compare More Nuanced: Pairwise Alignment Bilinear Network for Few-Shot Fine-Grained Learning

“…However, due to the limited representative capacity of hand-crafted features, the performance of this type of method is moderate. In recent years, deep neural networks have developed advanced abilities in the feature extraction and function approximation [28]- [34], bringing significant progress in the fine-grained image classification task [6]- [10], [24], [35], [35]- [46].…”

“…Bilinear CNN model (BCNN) [8] is the first work that adopts matrix outer product operation on the initial embedded features to generate a second-order representation for fine-grained classification. Li et al [10] (iSQRT-COV) further improve the navie bilinear model by using covariance matrices over the last convolutional features as fine-grained features. iSQRT-COV obtains state-of-the-art performance on both generic and fine-grained datasets.…”

“…Then with the operation of bilinear feature grouping, the PCM model can generate low-rank subtle descriptors of the original image. Most recently, Li et al [19] propose to uses the covariance pooling [10] to distillate the image representation of each category. These matrix-outer-product based bilinear pooling operations could extract the secondorder image features and contains more information than traditional first-order features [24], and thus achieve better performance on fine-grained data than generic ones.…”

Local Temporal Bilinear Pooling for Fine-Grained Action Parsing

“…Bilinear pooling. Bilinear pooling (or second-order pooling) is widely used in fine-grained visual classification [7][8][9][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33], visual questioning answering [34][35][36], feature fusion and disentangling [22,23,[37][38][39][40], action recognition [39][40][41][42][43][44] and other tasks. In deep neural nets, bilinear pooling is mostly used only once before the classification layer, e.g.…”