Many enthusiasts about Deep Learning, fail to appreciate that DL is an optimisation technique. It costs a lot to run now, but once the best algorithms are determined for any operation, those will get spun out of neural networks into more efficient operational hardware and software.
I come into this discussion with some filthy prejudices, however. About twenty years ago, it occurred to me that humans were generally predictable and unnecessarily complicated.
I am looking forward to seeing this demonstrated in machinery - whether it is I, or anyone else who gets it done.
I have been waiting a long time. :)
No comments :
Post a Comment