Bifalgorithm post

One of the best projects you can put on your resume is combining machine learning with one of the fastest growing fields, Quantum, and here's exactly how you're going to do it. First, pick your framework. For this, I'd use Penny Lane because it turns a Quantum Circuit into a layer that can be dropped directly into PyTorch. Then take a normal neural network and swap out one layer for a Quantum Circuit. And you can keep everything else the same, so there's no need to change the optimizer, loss function, or anything else. The Circuit is 4 qubits, your features go in as rotation angles, then a block of trainable rotations, then a ring of C-knock gates that entangle the qubits together. You measure at the end and those measurements are what the next layer sees. For me, I ran it on 2 moons so you can actually watch decision boundary form while it trains. But to actually prove the quantum layer works, you need a task that classical machine learning can cheat on. You need something where the two features only mean something together. So train a classical machine learning version right next to it, then a third one with the entangling gates deleted. So I'm testing it on a ZOR where the label depends on both features at once. So a model that can combine the features together will fail. And if all three score the same, your quantum layer didn't really do that much. And you basically just built a slower neural network. So pick a data set, swap out one layer and improve it with quantum.

One of the best projects you can put on your resume is combining machine learning with one of the fastest growing fields, Quantum, and here's exactly how you'r…