Coredump: Provable ML
Mar 5, 2022
Dump 05.03.2022.09.23.am
Welcome to Core dump, the one stop for techies, in this episode, we have Dr bheemaiah talk about provable ML.
Gecko: so all NP complete problems are Z Complete, and hence tensorizable, based on the machine dreaming, so essentially ML problems are provable in tensorization, and complexity.
Dr Bheemaiah:
Supervised and unsupervised learning is encapsulated in neural architectures, tensorized to formally define the complexity, trainable and transfer learnt to solve all NP complete problems.
Gecko:
Have we found any exceptions, any unsolvable problem that is not Z Complete?
Dr bheemaiah:
None so far.