DC goes Quantum

Naar blogoverzicht
3 March 2018

Since everyone seems to be talking about bots, AI and Blockchain nowadays, we at DotControl need to be sure that we stay next level and technically awesome by being (at least) one step ahead. That's why we decided to get our hands dirty with what we think will be the next big thing (or more precisely enable the next big things!). It's called quantum computing and is a completely new paradigm compared to "normal" computers.

 

Traditional computers have not fundamentally changed since their inception during the first half of the previous century. Modern computers are still constructed using the same 0's and 1's as Alan Turing's machine. Quantum computers make use of so called qubits. Whereas a traditional bit could only be in one of two states, true/false, on/off, a qubit can be in multiple states at the same time. To be quite honest, we also don't fully understand the underlying theoretical aspects (you should probably have an IQ that approaches Stephen Hawking's), but what we do know is that it will improve performance of algorithms to a scale that we have not yet seen before.

 

Quantum computing is often feared as a bitcoin killer as it will significantly reduce the amount of time needed to solve the mining hash algorithm that's required to be solved to add a block to a blockchain. Luckily quantum computing is not at the level needed for factoring large numbers.... yet. On the other end one could argue that quantum computers will eventually be able to help solve the current performance/network congestion issues blockchains are facing.

 

And as you might know, AI algorithms like deep learning are very compute intense. Recent advances in cloud technology have kind of hidden the hard steel for us, but you still need a lot of GPU's to train adeep learning model. GPU stands for Graphical Processing Unit and is a special kind of computer chip. Originally designed for 3D rendering, it turned out these chips where ideal for machine learning and blockchain mining algorithms) And to make our deep learning models even better, we need to increase the amount of inputs and number of layers in neural networks to go from lizard/cat brain level to super human level, which in  turn will exponentially lengthen the time it will take to train our models. Quantum computing will enable scientist to build weather forecasting models that are as complex as the world it self.

 

For long quantum computing has been a theoretical idea amongst scientists. Microsoft recognized the huge potential over 20 years ago and invested heavily in both building a real physical quantum computer as well as developing the tools needed to control this machine. Imagine having to build a machine that is needs to keep qubits in a stable state at near absolute zero temparature, while still being able to communicate and being controlled by a normal computer at room temparature. Microsoft developed a great toolkit that allows developers to learn how to programm this new paradigm using a quantum computer emulator while the quantum computer it self is still under development.

 

Quantum computing originated from the scientific community, so a lot of potential applications lie in the fields of for example physics modelling. One of the areas where we expect quantum computing will make a great impact is in the field of machine learning. Machine Learning and quantum computing mix very well since quantum computing in its current form is very well suited for the matrix algebra most machine learning algorithms use. We see quantum computing as a sort of GPU's on steroids.

 

Since the field of quantum computing is so new and literally offers (almost) endless possibilities it's very hard to predict what big things will be enabled by quantum computing. We will just make sure we have the knowledge and experience to be able to build those next things for our clients. See you at the other side...

Share this post on social media

Van Nelleweg 1, 3044 BC Rotterdam
+31 (0)10 71 44 646 info@dotcontrol.nl