?url_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=SWIM%3A+Short-Window+CNN+Integrated+With+Mamba+for+EEG-Based+Auditory+Spatial+Attention+Decoding&rft.creator=Zhang%2C+Z&rft.creator=Thwaites%2C+A&rft.creator=Woolgar%2C+A&rft.creator=Moore%2C+B&rft.creator=Zhang%2C+C&rft.description=In+complex+auditory+environments%2C+the+human+auditory+system+possesses+the+remarkable+ability+to+focus+on+a+specific+speaker+while+disregarding+others.+In+this+study%2C+a+new+model+named+SWIM%2C+a+short-window+convolution+neural+network+(CNN)+integrated+with+Mamba%2C+is+proposed+for+identifying+the+locus+of+auditory+attention+(left+or+right)+from+electroencephalography+(EEG)+signals+without+relying+on+speech+envelopes.+SWIM+consists+of+two+parts.+The+first+is+a+short-window+CNN+(SWCNN)%2C+which+acts+as+a+short-term+EEG+feature+extractor+and+achieves+a+final+accuracy+of+84.9%25+in+the+leave-one-speaker-out+setup+on+the+widely+used+KUL+dataset.+This+improvement+is+due+to+the+use+of+an+improved+CNN+structure%2C+data+augmentation%2C+multitask+training%2C+and+model+combination.+The+second+part%2C+Mamba%2C+is+a+sequence+model+first+applied+to+auditory+spatial+attention+decoding+to+leverage+the+long-term+dependency+from+previous+SWCNN+time+steps.+By+joint+training+SWCNN+and+Mamba%2C+the+proposed+SWIM+structure+uses+both+short-term+and+long-term+information+and+achieves+an+accuracy+of+86.2%25%2C+which+reduces+the+classification+errors+by+a+relative+31.0%25+compared+to+the+previous+state-of-the-art+result.&rft.subject=Training%2C+Accuracy%2C+Source+coding%2C+Benchmark+testing%2C+Brain+modeling%2C+Feature+extraction%2C+Data+augmentation%2C+Electroencephalography%2C+Data+models%2C+Decoding&rft.publisher=IEEE&rft.date=2025-01-16&rft.type=Proceedings+paper&rft.language=eng&rft.source=+++++In%3A++Proceedings+of+2024+IEEE+Spoken+Language+Technology+Workshop%2C+SLT+2024.++(pp.+pp.+1031-1038).++IEEE%3A+Macao.+(2025)+++++&rft.format=application%2Fpdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10205071%2F1%2F2409.19884v2.pdf&rft.identifier=https%3A%2F%2Fdiscovery.ucl.ac.uk%2Fid%2Feprint%2F10205071%2F&rft.rights=open