Abstract
We address plasticity loss in mixture-of-experts under QoE shifts in adaptive video streaming, with theoretical justification and large-scale empirical validation.
We mitigate plasticity loss in mixture-of-experts under shifting QoE objectives, with theoretical justification.