While trying to decide a topic for my ML research paper, I stumbled across an Idea to evaluate optimizer algorithms performance across multiple losses such as mean squared error binary crossentropy etc.
Can anyone tell me whether this would be a good idea or not and if yes what should be my approach with the experimentation method.
#Need help with research on optimizers
velvet shard
celest edge
I wouldn’t consider this research to be particularly interesting. Maybe for an undergrad project? But I doubt this would get published anywhere