{"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/hyperspectral-image-classification-in-the","title":"Hyperspectral Image Classification in the Presence of Noisy Labels","arxiv_id":"1809.04212","date":"2018-09-12","proceeding":null,"authors":["Junjun Jiang","Jiayi Ma","Zheng Wang","Chen Chen","Xian-Ming Liu"],"abstract":"Label information plays an important role in supervised hyperspectral image\nclassification problem. However, current classification methods all ignore an\nimportant and inevitable problem---labels may be corrupted and collecting clean\nlabels for training samples is difficult, and often impractical. Therefore, how\nto learn from the database with noisy labels is a problem of great practical\nimportance. In this paper, we study the influence of label noise on\nhyperspectral image classification, and develop a random label propagation\nalgorithm (RLPA) to cleanse the label noise. The key idea of RLPA is to exploit\nknowledge (e.g., the superpixel based spectral-spatial constraints) from the\nobserved hyperspectral images and apply it to the process of label propagation.\nSpecifically, RLPA first constructs a spectral-spatial probability transfer\nmatrix (SSPTM) that simultaneously considers the spectral similarity and\nsuperpixel based spatial information. It then randomly chooses some training\nsamples as \"clean\" samples and sets the rest as unlabeled samples, and\npropagates the label information from the \"clean\" samples to the rest unlabeled\nsamples with the SSPTM. By repeating the random assignment (of \"clean\" labeled\nsamples and unlabeled samples) and propagation, we can obtain multiple labels\nfor each training sample. Therefore, the final propagated label can be\ncalculated by a majority vote algorithm. Experimental studies show that RLPA\ncan reduce the level of noisy label and demonstrates the advantages of our\nproposed method over four major classifiers with a significant margin---the\ngains in terms of the average OA, AA, Kappa are impressive, e.g., 9.18%, 9.58%,\nand 0.1043. The Matlab source code is available at\nhttps://github.com/junjun-jiang/RLPA","url_abs":"http://arxiv.org/abs/1809.04212v2","url_pdf":"http://arxiv.org/pdf/1809.04212v2.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":"hyperspectral-image-classification-in-the","repo_url":"https://github.com/junjun-jiang/RLPA","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"hyperspectral-image-classification","task_name":"Hyperspectral Image Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1809.04212","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}