The ABC of Blockchain and How Machine Learning can benefit from it

The Blockchain - record-keeping technology behind the most popular cryptocurrency in the world - Bitcoin. There is a good chance that all of the hype around it confuses you, but the good news is that the blockchain technology is a lot easier to understand than it sounds.

Quick jump to the beginning of all. Blockchain technology was first outlined in 1991. Researchers, Stuart Haber and W. Scott Stornetta, wanted to implement a system where document timestamps could not be tampered with. A real breakthrough happened almost two decades later, with the launch of Bitcoin in January 2009. This was the first real-world application of a blockchain network.

After the explanation of the technology itself, we will tackle the topic of convergence between the blockchain and Machine Learning and why is that important.

Blockchain - hero under the mask

When we remove all the hype and buzzy words, Blockchain is just a chain of digital information blocks (the “block”) stored in a public dataset (the “chain”).

The blocks have three parts:

1. Information about transactions - the time, date, and amount of your purchase.
2. Information about who is participating in transactions - instead of the actual names, a purchase is recorded without any identifying information using a unique “digital signature.”
3. Information or unique code that distinguishes them from other blocks - a “hash” - hashes are cryptographic codes created by unique algorithms that set the blocks apart because of their unique code values.

Let’s remove the mask

Blockchain consists of multiple data blocks strung together. So for a block to be added to the chain and glue with others, four things must happen:

1. The transaction must occur - a single block can store from one to few thousands of transactions under one roof. So your transaction is packed in the block along with other users’ transaction information as well.
2. The transaction must be verified - this task is left to a network of computers, and that is why no one controls a blockchain. When you make the transaction, the network of computers rushes to check that your transaction happened in the way you said it did. They confirm the details - the transaction’s time, amount, and participants. This is what we meant by the decentralized network.
3. The transaction must be stored in a digital block - after the verification, it gets the green light, and it can join hundreds, or thousands, of others like it in a block.
4. The block must be given a hash - after all of a block’s verification, it must be given a unique code - hash. The block also receives the hash of the latest block added to the Blockchain. Once they connected (hashed), the block can be added to the Blockchain.

After this process, it becomes publicly available along with the information about when, where, and by whom the digital block was added.

The goal of Blockchain is to allow digital information to be recorded and distributed but not edited.

If anyone can see it, is the Blockchain private?

Indeed, anyone can see the content of a blockchain. More, users can connect their computer to the blockchain network and receive a copy of a blockchain that is updated automatically when a new block is added. Think of this as a News Feed on Social Networks when new status is posted.

So if I can have a copy, anyone can have a copy, which means there could be millions of copies of the same network. Regardless of identical copies, spreading the information across the network of computers makes the manipulation of the data almost impossible. Hackers would need to manipulate every single copy of a network on all the computer devices. This is the main reason why we call the blockchain “distributed leader” and why, even though the information is public, they are safe from manipulation.

However, it is essential to mention that you do not have access to information about the users making transactions because personal data about users is limited to their digital signature. So, if you cannot know who is adding blocks to the Blockchain, how can we trust Blockchain?

What about Security?

The security topic has been tackled in several ways. Let’s dive straight into it.

1. New blocks are always stored chronologically - they are continually added to the “end” of the Blockchain.
2. It would be tough to go back and change the contents of the block after it has already been added to the Blockchain. The main reason is that each block contains its hash code, along with the hash of the block before it which connecting the two. A math function creates the codes. If that information is edited, the hash code changes as well.
3. The test has been implemented for computers that want to join and add blocks to the chain - “consensus models.” Users need to “prove” themselves before they can participate in this process. One of the most famous examples is connected with Bitcoin and is called “proof of work.” With this test, computers must “prove” that they have done “work” by solving a complex computational math problem. When a computer solves the problem, it can add a block to the Blockchain. Be aware that the process of adding blocks to the Blockchain is not easy.

Statistically, the odds of solving one of these problems were about 1 in 15.5 trillion in January 2020. To solve complex math problems, computers must run programs that cost them significant amounts of power and energy.

Example:

Hacker tries to edit your transaction so that you have to pay double the price. As soon as they edit the amount, the block’s hash changed, but the next block in the chain still contains the old hash. That block would also need to be updated to cover their tracks. Doing so, he would change that block’s hash to. So to change a single block, every single block after it on the Blockchain has to be changed. Changing all of those hash codes would take an enormous amount of computing power.

In other words, once a block is added to the Blockchain, it becomes tough to edit and impossible to delete.

How could the Blockchain unlock the full power of Machine Learning?

Apart from Blockchain, Machine Learning is an innovation that is changing the world. Machine Learning revolution is here, with all of new research and applications coming to reality every single day. Not soo long ago, Machine Learning seemed like some science fiction novel, but today, it is all around us.

On the other hand, the progress of this technology brings unrealistically high expectations. Machine and Deep Learning expends a considerable measure of processing power. That is causing huge expenses for models to keep running on servers. Furthermore, this technology nowadays is under the control of massive corporations that have financial plans to resources into innovations. Truth to be told, algorithms are open sources, but computing power and data are not, which put us all in a highly dependent state of the big ones.

With that being said, blockchain technology could enable an infrastructure that allows Machine Learning to reach its full potential with decentralizing the computing power with intent to decrease computing costs and conveys transparency and security.

Computing power

We have to be aware that CPU data centers are not enough to meet the growing demand for Machine and Deep Learning. On the other hand, GPU computing power leads to increased speed and less energy consumption while distributing processing, perfect for Machine Learning tasks. Blockchain may provide the computational resources needed to utilize the computing power of machines that hold non-utilized GPU computing power.

“If we can tokenize value, can’t be also tokenize computing power?”

If we can create a peer-to-peer network of connected computers allowing individuals to rent resources out from each other, these resources can be used to complete tasks requiring any amount of computation time and capacity.

Computing costs

Decentralized marketplaces for GPU computing power could quickly compress the cost curves and unleash the power that Machine Learning task needs. The idea could be to have connected platform members who share their system resources to complete a given task. With blockchain technology, Machine Learning innovations can dramatically reduce its computing costs by accessing the globally distributed GPUs. Crypto miners could use these GPUs, and then make them available to Machine Learning and Deep Learning companies.

Transparency and Data Integrity

Blockchains can create an environment where data is immutable, transparent, distributed, and free to operate. With this approach, the model efficiently preserves its validity. Furthermore, we could count on adding structure and accountability to Machine Learning algorithms, as well as the quality and usefulness of the solutions they produce.

Below is the classification list of current Blockchain projects for Machine Learning.

1. Blockchains for Machine Learning in Trading and Finance
2. Blockchains for Machine Learning in Medicine
3. Blockchains for Machine Learning in Supply Chain
4. SingularityNET
5. DeepBrainChain
6. MatrixAINetwork

All that awesomeness of Blockchain & Machine Learning Projects

Blockchain has the potential to act as the backbone for Machine Learning infrastructure. By commoditizing and tokenizing raw computational processing power, we can expand the development of our models beyond imaginable. Any interested party would be free to access GPU computing power, which reduces our dependence on big tech giants.

On the other hand, the combination of Machine Learning and Blockchain is defacto fueling the “Fourth Industrial Revolution”. We are reinventing economics and information exchange. From fintech to healthcare, the combination of Machine Learning and Blockchain is slowly but surely transforming industries. The only question is, would you take an active role in it, or be a passive watcher.