I'm not too familiar with few-shot learning, so the TADAM explanation was unclear to me. Specifically, I wasn't sure what a "Prototypical Networks -like setup" was and what a "support set" is. Is there a way to explain the use of FiLM without using these terms? Or (perhaps less preferably) using these terms, but explaining them a bit more?
I'm not too familiar with few-shot learning, so the TADAM explanation was unclear to me. Specifically, I wasn't sure what a "Prototypical Networks -like setup" was and what a "support set" is. Is there a way to explain the use of FiLM without using these terms? Or (perhaps less preferably) using these terms, but explaining them a bit more?