I’ve run the quantum transfer learning problem of “ants and bees”. The notebook which is provided by Andrea Mari. with epoch=30 which is the case in the paper I got “Best test loss: 0.3091 | Best test accuracy: 0.9412” but in the paper by Maria Schuld it’s mentioned the accuracy is 0.97. Do you know what is happening?
2. I run the same “ansts and bees” transfer learning problem in classical mode code provided by PyTorch. I get “est val Acc: 0.960784”. so I was expecting to get higher accuracy with Quantum than classical . am I missing something?
Hi @Chicago_Girl ,
Both classical and quantum software have evolved since the paper was published, so it can happen that the classical model now performs better.
It’s also important to note that for most problems, classical algorithms will outperform quantum ones. It’s actually an active field of research to find applications where quantum computers can show a significant advantage.
If you’re interested in learning more about this you might like our blog post on benchmarking near-term quantum algorithms, or the one on measuring performance.