{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/learning-convolutional-neural-networks-using","title":"Learning Convolutional Neural Networks using Hybrid Orthogonal Projection and Estimation","arxiv_id":"1606.05929","date":"2016-06-20","proceeding":null,"authors":["Hengyue Pan","Hui Jiang"],"abstract":"Convolutional neural networks (CNNs) have yielded the excellent performance\nin a variety of computer vision tasks, where CNNs typically adopt a similar\nstructure consisting of convolution layers, pooling layers and fully connected\nlayers. In this paper, we propose to apply a novel method, namely Hybrid\nOrthogonal Projection and Estimation (HOPE), to CNNs in order to introduce\northogonality into the CNN structure. The HOPE model can be viewed as a hybrid\nmodel to combine feature extraction using orthogonal linear projection with\nmixture models. It is an effective model to extract useful information from the\noriginal high-dimension feature vectors and meanwhile filter out irrelevant\nnoises. In this work, we present three different ways to apply the HOPE models\nto CNNs, i.e., {\\em HOPE-Input}, {\\em single-HOPE-Block} and {\\em\nmulti-HOPE-Blocks}. For {\\em HOPE-Input} CNNs, a HOPE layer is directly used\nright after the input to de-correlate high-dimension input feature vectors.\nAlternatively, in {\\em single-HOPE-Block} and {\\em multi-HOPE-Blocks} CNNs, we\nconsider to use HOPE layers to replace one or more blocks in the CNNs, where\none block may include several convolutional layers and one pooling layer. The\nexperimental results on both Cifar-10 and Cifar-100 data sets have shown that\nthe orthogonal constraints imposed by the HOPE layers can significantly improve\nthe performance of CNNs in these image classification tasks (we have achieved\none of the best performance when image augmentation has not been applied, and\ntop 5 performance with image augmentation).","url_abs":"http://arxiv.org/abs/1606.05929v4","url_pdf":"http://arxiv.org/pdf/1606.05929v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"learning-convolutional-neural-networks-using","repo_url":"https://github.com/mowangphy/HOPE-CNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"image-augmentation","task_name":"Image Augmentation"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}