Independent Components Analysis (ICA) and its variants have been successfully used for unsupervised feature learning. However, standard ICA requires an orthonoramlity constraint to be enforced, which makes it difficult to learn overcomplete features. In addition, ICA is sensitive to whitening. These properties make it challenging to scale ICA to high dimensional data. In this paper, we propose a robust soft reconstruction cost for ICA that allows us to learn highly overcomplete sparse features even on unwhitened data. Our formulation reveals formal connections between ICA and sparse autoencoders, which have previously been observed only empirically. Our algorithm can be used in conjunction with off-the-shelf fast unconstrained optimizers. We show that the soft reconstruction cost can also be used to prevent replicated features in tiled convolutional neural networks. Using our method to learn highly overcomplete sparse features and tiled convolutional neural networks, we obtain competitive performances on a wide variety of object recognition tasks. We achieve state-of-the-art test accuracies on the STL-10 and Hollywood2 datasets.

In this paper, we present a large database of over 50,000 user-labeled videos collected from YouTube. We develop a compact representation called "tiny videos" that achieves high video compression rates while retaining the overall visual appearance of the video as it varies over time. We show that frame sampling using affinity propagation-an exemplar-based clustering algorithm-achieves the best trade-off between compression and video recall. We use this large collection of user-labeled videos in conjunction with simple data mining techniques to perform related video retrieval, as well as classification of images and video frames. The classification results achieved by tiny videos are compared with the tiny images framework [24] for a variety of recognition tasks. The tiny images data set consists of 80 million images collected from the Internet. These are the largest labeled research data sets of videos and images available to date. We show that tiny videos are better suited for classifying scenery and sports activities, while tiny images perform better at recognizing objects. Furthermore, we demonstrate that combining the tiny images and tiny videos data sets improves classification precision in a wider range of categories.

This paper presents a new method for video and image categorization based on a database of over 50,000 videos collected from YouTube and down-sampled to tiny size. The categorization results achieved by tiny videos are compared with the tiny images framework for a variety of recognition tasks. The tiny images dataset consists of 80 million images collected from the Internet. These are the largest labeled research datasets of videos and images available to date. We show that tiny videos are better suited for classifying sports activities and scenery, while tiny images perform better at recognizing objects. Furthermore, we demonstrate that combining the tiny images and tiny videos datasets improves categorization precision in a wider range of categories.

This work extends the tiny images techniques developed by Torralba et al. to videos. A dataset of 6,612 videos was collected from YouTube in the Sports and News sections. We present a method for compressing the temporal dimension nonuniformly using affinity propagation. We show that nonuniform sampling using affinity propagation outperforms temporal sampling at uniform intervals, because it covers a greater range of visual appearances in the video for the same number of samples. We examine two main applications for the tiny video dataset: duplicate video detection and related video retrieval. We also show that the scope of text-based searches on YouTube can be significantly increased by incorporating visual similarity.