11 May 2018

The DeepBrain Chain Making Inroads in the NEO Blockchain

An Interview With DeepBrain Chain CEO

Tony Tian a cryptocurrency investor that is involved in long-term investment and swing trading recently conducted a live video interview with the CEO and founder of DeepBrain Chain – Yong He. A lot of questions were thrown at the CEO mainly regarding the history, present activities and possible future plans of DeepBrain Chain. As well as how DBC intends optimising there presence on the NEO blockchain.

One of the telling plans of DeepBrain Chain involves AI training on NEO blockchain. This will revolutionise the AI industry and could only be made possible with the use of the NEO blockchain which is a decentralised, open source and flexible network.

The questions asked by the interviewer bothered on the issues that DeepBrain Chain may need to resolve in their quest to conduct AI training using the NEO blockchain network. Check out excerpts from the interview below.

The excerpts include questions on AI training and Yong He’s response. (Note that the information provided for both the questions and answers are not entirely word-for-word or verbatim and only represent a summarised report of the actual interview.) For more information you can watch the entire interview here.


Tony Tian’s Question:

Tony Tian

What problems do you foresee DeepBrain Chain encountering in its quest to conduct AI training on the NEO blockchain?

Response from Yong He:


Using a case sample: AI Facial Recognition

If 99.99% accuracy in facial recognition is the desired target, hundreds of millions of data will be required if this target is to be attained. However, this is also under the premise that the AI algorithm used is a complete one.

Still on facial recognition – there will be need for hundreds of millions of images and this would require a significant amount of data. The data required would be thousands of gigabytes and not a few gigabytes. Lots of money would be required to buy this volume of data and this is just the first problem that has to be sorted out.

The second issue is the need to have the right machines to train the model before it can be utilised. Imagine if just one machine has a size of between 30 and 50 square meters, this takes up a lot of space and you will need several of such machines. The machine itself needs to train the data and would take two weeks to do so. This is before you have a working AI application in operation. Imagine what just one machine would cost and you will need several other similar machines.

The reason why many machines will be needed is because it is expected that when AI training takes off fully on the NEO blockchain there will be a vast number of users from the community due to the sheer nature of the open source network. As users on the NEO blockchain grow, more machines would be required to do even more training to improve the model as well as carry out calculations at the same time in order to better the algorithm.

It’s the same thing with facial recognition, as there are more users on the NEO blockchain, there will be need for more computing power to calculate the vast volume of data. So the AI industry has the problem of high operational costs. To obtain the machines and to keep upgrading their number as the growth of users on the NEO blockchain increases.


Tony Tian’s Question:

What would be the role of miners, buyers and sellers in the NEO ecosystem?


Response from Yong He:

At the basic level on the decentralised NEO ecosystem there will be data sellers that are directly from the AI industry. These data sellers could be professional data IoT or AI companies themselves. The data buyers are the AI companies obviously because the data is needed for training. Buying and selling data will be important roles of the DBC.


Tony Tian’s Question:

How do you intend recruiting miners?

AI companies can be miners because they have the machines to do so. These machines don’t operate on a 24/7 basis, so they can be made available on the DBC ecosystem and offer their computing power when not in use. Individual investors can also buy machines and get returns on their investment when they sublet the use of these machines to AI companies. The open source benefit of the NEO blockchain also makes it easy to have incentivised professional mining pools.



Leave Comment

Your email address will not be published. Required fields are marked *