[Adapt] [Seminar] Weighting Losses Automatically

[forest_zch]

Hi Adapters, 

Numerous deep learning applications benefit from multitask learning with multiple regression and classification objectives. However, simply add the losses of each task together (weighted uniformly) or tune the weight manually may not achieve the best performance, alongside difficult and expensive. In this seminar I’d like to introduce a simple method which use the estimation of uncertainty to weight losses automatically. It is useful in many multi-task scenarios. 

Time: Wed 10 am. - 11:30 am. 
Meeting link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_M2VmMTU5MzgtODUzOC00NmU4LTg0MzktNGFjNDdiMmIwYTI1%40thread.v2/0?context=%7b%22Tid%22%3a%225cdc5b43-d7be-4caa-8173-729e3b0a62d9%22%2c%22Oid%22%3a%221a8b9fa0-af57-4a1c-9390-22d1c201d622%22%7d

Hope you enjoy it. 

Chunhao
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20230918/b7051919/attachment.htm>