STARTUPS, TECHNOLOGY

NeuroSeed and neural networks learning

 NeuroSeedTechnologies which is currently going through a private presale of its tokens, claims to hold 1% to 10% of the machine leaning market by 2025. That was announced by a company founder, Oscar Braverman, at Moscow for Blockchain/Blockchain for Moscow conference. At the current stage, the company has produced project’s MVP (Minimum Viable Product).

NeuroSeed founders

NeuroSeedTechnologies is registered in Tallinn (Estonia). Its two founders are Oscar Braverman and Ivan Suvorov, both nationals of Israel. In Russia, the interests of the company are represented by Ruslan Okhrimchuk, branch office CEO and project’s economics advisor.

Project founders have already invested $600K. They intend to raise no less than $3.2 mio at the private presale and $17.5 mio at an ICO. The ICO date has not been set yet.

Nowadays, the company has other branch offices in Kharkov, where a team of developers is located, as well as in Israel and Estonia.

NeuroSeed purposes

In the age of BigData, IoT, Virtual Reality and Augmented Reality, machine learning is in a daily growing demand. Yet teaching ML models is a lengthy and costly process. A subsequent use of the already taught ML models or employing them as the basis for new developments may help solving the problem. Nevertheless, there is never a guaranty the models are absolutely correct and have not been altered by some intruders. Even at Github, the largest web service for IT projects hosting and for their development, such deficient programs may be found. Google has developed an AI which can remove any image protection and even clean watermarks off photos.

NeuroSeed platform aims to resolve the problem of trust between owners of data adapted for neural networks teaching, and ML solutions developers. NeuroSeed has introduced a validation algorithm which allows monitoring the entire neural networks development cycle and saves hashtags of the entire process in a blockchain. For that, external validators (i.e. special people validating collected data accuracy), digital signature of the data scientist who composed a data set, as well as validation algorithms are used. Models will be able to learn by using encrypted data thus protecting copyrights. To keep the information on a neural network development intact, a public blockchain will be used. Blockchain is also required for mutual settlements when readymade solutions and data are leased.

NeuroSeed will use ML solutions stored at Github. At the platform, NeuroSeed stored its own products as well, including, for instance, a program which distinguishes cats from dogs, a program which recognize music styles, etc.

NeuroSeed aims to become a marketplace with readymade machine learning solutions available. The platform stores models which resolve practical business problems and anyone can employ an already taught model. All platform participants such as data scientists, validators, developers, computation capacities suppliers, get remunerations according to the API references.

NeuroSeed, just like Google, wants to make ML algorithms available to everyone. Such solutions will soon be in demand everywhere where process computerization is required, ranging from companies to smart homes.

We do not create Artificial Intelligence, yet we build algorithms or bricks an Artificial Intelligence is made of”, Oscar Braverman notes.

That is the distinction between NeuroSeed and the companies which develop centralized or decentralized AI. Meanwhile, Google has already developed a set of simple instruments for working with an Artificial Intelligence. Thus, UrbanOutfitters has employed a package provided for computerization of collection of data on the opinions of buyers of various goods.

NeuroSeed ambitions

According to Statista, by 2025 global machine learning market will reach $60 bln. NeuroSeed strives to get hold of up to 10% of it. The company does not task itself with developing ML solutions, whereas it intends to design a universal instrument for cheapening and simplifying ML algorithms development processes.

Ruslan Okhrimchuk, NeuroSeed economics advisor, says that “NeuroSeed is a unique platform, and its uniqueness is the foundation for its future success. Our program reduces the costs of programs and their learning time, since mainly long established and well tested neural networks are being taught”.

Nowadays, neural networks are employed virtually everywhere. In Moscow underground, a face recognition system is used. It is based on neural networks and has helped arrest lots of criminals. Robots are now employed to pick up and pack apples, peppers, flowers. Their operation is based on neural networks taught to recognize ripeness of fruit and flowers. Robots handle calls at call centers, fix appointments with doctors and hairdressers, cook food. Neural networks are able to count bees in a hive, which helps in predicting the volume of honey to be produced, etc.

Aeroflot will probably also use ML algorithms based on NeuroSeed platform. They need a neural network to analyze their air traffic in order to prevent emergencies in the air and on the ground. The neural network will help dispatchers in performing their duties. The details of the negotiations on the partnership between the air carrier and NeuroSeed are not disclosed. The air company management did not reply to a comment request forwarded by Invest Foresight.

Logitek Plus, a logistics company in St Petersburg, intends to order a program to monitor vehicles and drivers to see the drivers’ performance and to swiftly respond to breakages and accidents. The offered program was priced at $50K. If NeuroSeed platform’s resources are used, the program will be several times cheaper.

By Natalia Kuznetsova

Previous ArticleNext Article