The Future Of Machine Learning – We want to solve real-world problems with tools that don’t actually work like our world. That might be a problem that quantum computers solve in the near future. Just imagine a stick. This stick represents a classical bit. It can point towards two directions. Now imaging catching that stick in the center and trying to move it as fast as possible in all possible directions. If you do it fast enough, you will see a sphere. This sphere is a quantum bit.
The problem with quantum bits is they are constantly in motion and whenever you try to explore their state, it will be different even just after your previous observation. Moreover, all of the other quantum bits will simultaneously change their state. Hmm, it sounds like a serious problem. Fortunately, you can use this mechanism to calculate probabilities for particular states and it makes everyone hoping that there will be some way to use it in practice. There’s another problem however.
Not everyone is able to spend approximately 15 million dollars on a quantum computer just to have a chance to try to use it. It sounds like a horrible idea – even for someone who’s got lots of money. As it happens, it’s not necessary for some time and that’s why TensorFlow Quantum cooperates with Python library Cirq. With Cirq’s help, you can simulate quantum circuits on your own machine and when the time comes, you should be able to implement your code to the real quantum computer.
It sounds really cool, doesn’t it? It sounds cool, but it’s not that easy. Let me show you the code responsible for “Hello World”, but for quantum computing. It’s called “Hello Many Worlds”. I think about TensorFlow Quantum as a bridge between quantum computing and classical Machine Learning. This notebook shows exactly how to join these two worlds.
Elaborating further on such an advanced topic wouldn’t make much sense for such a short article. What’s important is that you can find further information about possible applications of TensorFlow Quantum in their documentation. You can read about “simple” MNIST classification, how to calculate gradients, and quantum CNNs.
The last one is my favorite notebook, and I do recommend for everyone to go through it. All of the concepts are beautifully explained. Just don’t get me wrong. I’m not a quantum mechanic wizard. I just did a couple of days of research and I’ve implemented a few basic applications using TensorFlow Quantum. It’s been a lot of fun and it required me to read about many different issues that I haven’t heard about before.
Give it a try and I promise you will not regret it, as long as you’ve got some time to catch up where it’s needed. The whole library was open-sourced with the hope that somewhere there will be good and talented people willing to contribute to it. Maybe it’s you?
Web enthusiast. Thinker. Evil coffeeaholic. Food specialist. Reader. Twitter fanatic. Music maven. AI and Machine Learning!