there has been a lot written about how to do verification checking and automated 'debugging' of machine learning neura networks. the code, unlike conventional code, is not subject to conventional debugging. perhaps, it is topolical analysis and automated hypothesis testing that will be the key to creating software based checkers and debuggers for neural networks.-----
there has been a lot written about how to do verification checking and automated 'debugging' of machine learning neura networks. the code, unlike conventional code, is not subject to conventional debugging. perhaps, it is topolical analysis and automated hypothesis testing that will be the key to creating software based checkers and debuggers for neural networks.-----
very nice talk, I'd be interested in a video of more possible use cases
Smaller text on the slides is not readable, can you link to a copy of his presentation?
OMG he doesn't have a strong Swedish accent. Good job Gunnar. XD