Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks

Abstract

Abstract-Deep convolutional networks have proven to be very successful in learning task specific features that allow for unprecedented performance on various computer vision tasks. Training of such networks follows mostly the supervised learning paradigm, where sufficiently many input-output pairs are required for training. Acquisition of large training sets is one of the key challenges, when approaching a new task. In this paper, we aim for generic feature learning and present an approach for training a convolutional network using only unlabeled data. To this end, we train the network to discriminate between a set of surrogate classes. Each surrogate class is formed by applying a variety of transformations to a randomly sampled 'seed' image patch. In contrast to supervised network training, the resulting feature representation is not class specific. It rather provides robustness to the transformations that have been applied during training. This generic feature representation allows for classification results that outperform the state of the art for unsupervised learning on several popular datasets . While such generic features cannot compete with class specific features from supervised training on a classification task, we show that they are advantageous on geometric matching problems, where they also outperform the SIFT descriptor.

Deconvolution and Checkerboard Artifacts

“…Notably, the discriminator has many less feature maps (512 in the highest layer) compared to K-means based techniques, but does result in a larger total feature vector size due to the many layers of 4 × 4 spatial locations. The performance of DCGANs is still less than that of Exemplar CNNs (Dosovitskiy et al, 2015), a technique which trains normal discriminative CNNs in an unsupervised fashion to differentiate between specifically chosen, aggressively augmented, exemplar samples from the source dataset. Further improvements could be made by finetuning the discriminator's representations, but we leave this for future work.…”

FCSS: Fully Convolutional Self-Similarity for Dense Semantic Correspondence

“…Weakly-Supervised Feature Learning For the purpose of object recognition, Dosovitskiy et al [11] trained the network to discriminate between a set of surrogate classes formed by applying various transformations. For object matching, Lin et al [28] proposed an unsupervised learning to learn a compact binary descriptor by leveraging an iterative training scheme.…”

“…Furthermore, since the pre-learned sampling patterns used in the CSS layers are fixed over an entire image, they may be sensitive to nonrigid deformation as described in [26]. To address this, we perform the max-pooling operation within a spatial window N i centered at a pixel i after the non-linear gating: (11) where W k λ is a learnable parameter for scale k. The maxpooling layer provides an effect similar to using pixelvarying sampling patterns, providing robustness to nonrigid deformation. The descriptor for each pixel then undergoes L 2 normalization.…”

An Efficient CNN a deep learning approach applied on the image matching context

“…The paper [1] discusses training the Conventional Neural Network using only unlabelled data to discriminate between a set of surrogate classes. Each surrogate classes are formed by applying a variety of transformations to a randomly sampled seed image patch.…”