c++ matrix template library need for neural networks computations

Question

hi i'm implementing some neural network algorithms and i'll be needing a matrix library, I've looked and found that there are ones like ( boost::ublas ) , (blitz++), TNT... I need experts opinion which one is suitable for (simple , easy coding, high performance maybe )

Answer 1

Absent some reason to do otherwise, I'd probably go with boost::ublas. To be honest, I doubt that most of the code in any of them is going to contribute a lot toward a neural network, so it probably won't make a huge difference which one you choose. They'll all just be acting as simple containers.