{"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/max-ast-combining-convolution-local-and","title":"MAX-AST: COMBINING CONVOLUTION, LOCAL AND GLOBAL SELF-ATTENTIONS FOR AUDIO EVENT CLASSIFICATION","arxiv_id":null,"date":"2024-04-14","proceeding":"ICASSP 2024 4","authors":["Tony Alex","Sara Ahmed","Armin Mustafa","Muhammad Awais","Philip JB Jackson"],"abstract":"In the domain of audio transformer architectures, prior research\r\nhas extensively investigated isotropic architectures that capture the\r\nglobal context through full self-attention and hierarchical architectures that progressively transition from local to global context utilising hierarchical structures with convolutions or window-based attention. However, the idea of imbuing each individual block with\r\nboth local and global contexts, thereby creating a hybrid transformer\r\nblock, remains relatively under-explored in the field.\r\nTo facilitate this exploration, we introduce Multi Axis Audio\r\nSpectrogram Transformer (Max-AST), an adaptation of MaxViT\r\nto the audio domain. Our approach leverages convolution, local\r\nwindow-attention, and global grid-attention in all the transformer\r\nblocks. The proposed model excels in efficiency compared to prior\r\nmethods and consistently outperforms state-of-the-art techniques,\r\nachieving significant gains of up to 2.6% on the AudioSet full set.\r\nFurther, we performed detailed ablations to analyse the impact of\r\neach of these components on audio feature learning. The source\r\ncode is available at https://github.com/ta012/MaxAST.git","url_abs":"https://cmsworkshops.com/ICASSP2024/view_paper.php?PaperNum=7388","url_pdf":"https://cmsworkshops.com/ICASSP2024/view_paper.php?PaperNum=7388","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":"max-ast-combining-convolution-local-and","repo_url":"https://github.com/ta012/MaxAST","is_official":0,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"audio-classification","task_name":"Audio Classification"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/audio-classification-on-audioset","task":"Audio Classification","dataset":"AudioSet","model":"MAX-AST (Single)","rank_in_archive_order":25,"of":51,"metrics":{"Test mAP":"0.481"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}