{"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/multi-level-and-multi-scale-feature","title":"Multi-Level and Multi-Scale Feature Aggregation Using Pre-trained Convolutional Neural Networks for Music Auto-tagging","arxiv_id":"1703.01793","date":"2017-03-06","proceeding":null,"authors":["Jongpil Lee","Juhan Nam"],"abstract":"Music auto-tagging is often handled in a similar manner to image\nclassification by regarding the 2D audio spectrogram as image data. However,\nmusic auto-tagging is distinguished from image classification in that the tags\nare highly diverse and have different levels of abstractions. Considering this\nissue, we propose a convolutional neural networks (CNN)-based architecture that\nembraces multi-level and multi-scaled features. The architecture is trained in\nthree steps. First, we conduct supervised feature learning to capture local\naudio features using a set of CNNs with different input sizes. Second, we\nextract audio features from each layer of the pre-trained convolutional\nnetworks separately and aggregate them altogether given a long audio clip.\nFinally, we put them into fully-connected networks and make final predictions\nof the tags. Our experiments show that using the combination of multi-level and\nmulti-scale features is highly effective in music auto-tagging and the proposed\nmethod outperforms previous state-of-the-arts on the MagnaTagATune dataset and\nthe Million Song Dataset. We further show that the proposed architecture is\nuseful in transfer learning.","url_abs":"http://arxiv.org/abs/1703.01793v2","url_pdf":"http://arxiv.org/pdf/1703.01793v2.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":"multi-level-and-multi-scale-feature","repo_url":"https://github.com/jongpillee/music_dataset_split","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"music-auto-tagging","task_name":"Music Auto-Tagging"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1703.01793","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}