Two of the biggest headline creators in the tech world today are blockchain and Machine Learning, with good reason. Both of these disruptive technologies have already begun to change the sectors in which they are being applied. Individually blockchains and ML have the power to radically alter sectors ranging from finance, robotics, communication and data analysis. So what could they do if combined? How might they be combined?
First lets look at what each technology currently does by itself. As of right now blockchains are mostly used to facilitate the creation and distribution of cryptocurrencies in an attempt to break away from the centralised and regulation heavy control of traditional financial institutions. Other applications for blockchains do exist and new ones are emerging all the time (such as data management systems and legal contracts) but their use in cryptocurrency remains dominant.
Machine Learning (ML) is a technology that has actually been around a long time, with many of the modern approaches finding form in the 1990s when seminal papers on SVMs and Recurrent Neural Networks were written, although it could argued that some form of machine learning has followed in step with computing since the 1940s. However ML really has taken off in the past 10 years or so with the massive improvement in both computing power and data gathering capabilities leading to the advent of "deep learning". Very simply put this means throwing a whole load of data and computing power at a problem until it's solved. This can be applied to problems in data analysis, autonomous vehicles, imaging recognition systems and many more.
(source: XKCD, https://xkcd.com/1838/)
So now we know what each technology does, how can they be combined in a useful way? Well one way they can work together is to solve problems within each technology. For example ML usually necessitates the collection of vasts amount of data in order to be useful. Blockchains are a natural way for ML researchers to gather data. One can imagine a neural network deployed on or piggy backing off a blockchain that has wide distributions in order to gather data for any number of applications. Additionally, the neural network could be deployed analyse the blockchain itself, monitoring for example, network load and perhaps even tapping into dynamic parts of the blockchain code in order to manage changes in the load.
Another scenario might see the blockchain being used as a secure database, constantly collecting real time information ranging from financial information about a business or company, health care information from a hospital, information about electrical grid power consumption etc. Meanwhile the ML program can be analysing this data also in real time and providing useful analytics or motioning information. One can even imagine a scenario where the ML program itself would automatically enact changes in the system in response to this real time information.
These are just a few ways blockchains and ML can be used in conjunction with one another. With increased advancements in both technologies it's certain the possible ways they can interact will also increase. As the overlap between these technologies continues to converge, we may soon see completely decentralised, autonomous blockchain/ML systems that require little to no input from humans, to operate.