BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Parameter-Efficient Fine-tuning for Audio and Speech Processing - 
 Umberto Cappellazzo\, University of Trento
DTSTART:20240422T110000Z
DTEND:20240422T120000Z
UID:TALK215431@talks.cam.ac.uk
CONTACT:Simon Webster McKnight
DESCRIPTION:Leveraging large pre-trained models for downstream tasks has b
 ecome a cornerstone of several domains like natural language processing an
 d audio/speech processing. The typical paradigm involves adapting the whol
 e model to each downstream task (i.e.\, full fine-tuning). However\, given
  the relentless and unprecedented rise in scale of these foundation models
 \, full fine-tuning becomes often prohibitive\, especially when we deal wi
 th numerous downstream tasks. For this reason\, parameter-efficient fine-t
 uning (PEFT) strategies have emerged\, whereby only a small fraction of pa
 rameters are learned while keeping the backbone model frozen. In the realm
  of audio and speech processing\, PEFT has also gained traction and become
  a valid alternative to full fine-tuning. In this talk\, we first provide 
 an overview of the most common PEFT methods for the efficient adaptation o
 f the Audio Spectrogram Transformer to several tasks and under different s
 cenarios. We then investigate how the paradigm of Mixture of Experts can b
 e harnessed to scale the number of adapters\, leading to enhanced performa
 nce. We conclude by proposing new adapter designs that turn out to beat fu
 ll fine-tuning while adapting only 0.3% of parameters compared to it.
LOCATION:Hybrid: JDB Teaching Room\, Engineering Department or Zoom: https
 ://cam-ac-uk.zoom.us/j/86036389828?pwd=U0dsWis2alp4K0tORHpZa3MrSXZXdz09
END:VEVENT
END:VCALENDAR