synthetic data


for ai applications

Unlocking game-changing computing capacity for wide AI adoption

Notice: During recess Neuromation will continue the token sale with selected partners, remaining tokens will be available in the public token sale .
60M NTK sold
< 7M NTK remain
Neuromation sold over half of its available tokens in the pre-sale
* numbers projected to the end of the pre-sale (Noon GMT, Jan 1st 2018)
synthetic data
Neurotoken (NTK) pre sale ended
public sale starts
7 Jan, 12.00 GMT
Unlocking game-changing computing capacity for wide AI adoption
Unlocking game-changing computing capacity for wide AI adoption

Get social with Us!


"This project is unique, revolutionary, and cheaper/faster than every competitor. The science they are accelerating with these large data mines could change the world in a lot of ways. I feel very comfortable recommending this ICO. I wouldn’t sell it anytime soon either. 7.8 out 10 "

Joshua Larson , Hacked.com .

Our Partners

38 359 447

Labeled Synthetic images have been created

Why Synthetic Data?

The Neuromation Platform will use distributed computing along with blockchain proof of work tokens to revolutionize AI model development.
The revolution is long overdue: deep learning employs artificial neural networks of extremely large capacity and, therefore, requires highly accurate labeling. Collecting large datasets of images, text and sound is easy, but describing and annotating data to make it usable has traditionally been challenging and costly. Crowdsourcing was applied to the problem of dataset creation and labeling a few years ago, employing large numbers of humans to correct mistakes and improve accuracy. It proved slow, expensive and introduced human bias. Besides, there were tasks that humans simply could not do well, such as estimating distances between objects, quantifying lighting in a scene, accurately translating text, and so on.
We propose a solution whose accuracy is guaranteed by construction: synthesizing large datasets along with perfectly accurate labels. The benefits of synthetic data are manifold. It is fast to synthesize and render, perfectly accurate, tailored for the task at hand, and can be modified to improve the model and training itself. It is important to note that real data with accurate labels is still required for evaluating models trained on synthetic data, in order to guarantee acceptable performance at inference time. However, the amount of validation data required is orders of magnitude smaller than training data!
Neuromation is creating a distributed platform to service all aspects of future synthetic data ecosystem. The platform will allow users to create dataset generators, generate massive datasets, train deep learning models. Users will also be able to trade datasets and models in the platform marketplace.
We plan to take advantage of the latest trends in blockchain cryptocurrency. All transactions in the Neuromation will be done using our Ethereum extended ECR-20 compliant token. We will engage crypto-currency miners in computationally intensive tasks of data generation and model training. They will be mining our Tokens by performing these tasks instead of mining cryptocurrency.
The enormous computing capacity that will become available on the platform will be game-changing for wide AI adoption by the Enterprise.

Industrial automation

Neuromation plans to open an industrial automation lab where synthetic data will be applied to various problems. Typical difficulty with industrial environment is highly dynamic nature of the performed tasks. The model will need to analyze real time data and produce highly sophisticated predictions and calculations on the fly. It is very difficult to properly teach models to operate with dynamic data. Sometimes it is impractical to construct even the smallest dataset from real data. Neuromation will be using our proprietary Sensor Emulation Sandbox to procedurally create virtual environments where the models would be trained. Areas we are planning to tackle are: drone autopilots, industrial process monitoring, object manipulation, and others.

Our operating procedure is to focus on finding an industry partner who would take advantage of the capabilities Neuromation will provide and spearhead adoption of our methods. The goal then is to open up our sandboxes to 3rd parties to allow them to buy our data and train their models. Operating procedure is to focus on finding an industry partner who would take advantage of the capabilities Neuromation will provide and spearhead adoption of our methods.

Biotech and Pharma

There are a lot of interesting targets for the Neuromation approach in Pharma / Biotech. In general any problem where construction is easier than recognition (output is easy to calculate from input, but the reverse is hard; like computer science hash function) will yield well to our synthetic data approach. We are currently looking for partners in diagnostics and drug discovery to start a pilot project and open up a lab. We estimate that a lot of hypothesis building in drug discovery and biological system dynamics can be handled with Deep Learning models. If you work in the field and have good ideas please write to us: partnerships@neuromation.io

Our ideal pilot partner will bring a concrete problem that can be readily deployed in the industry. The most interesting case being where Neuromation would need to model not only the output of the system for proper classification, but also some underlying processes to create a full scale simulation that would extend to other problems and domains. We look forward to your suggestions!

Data Set Generation

Neuromation sells synthetic datasets for AI training. We create a sandbox environment that will generate virtually unlimited set of well-labeled examples for you.

We charge per image / data point.The generation tool can be deployed into a web service API.In general you only pay after some agreed degree of validation has been achieved. The model will train and be cross-validated on real-world data. Partnership opportunities are also available and are evaluated on a case by case basis.

Please write to us: sales@neuromation.io for more information.
AI Model As a Service

Neuromation can build your back-end AI engine. We will develop a model from scratch taking care of data generation, model training and tuning for you. Neuromation can also do retraining and maintenance as required. Neuromation can also help if you need to improve the existing model with additional data points (synthetically generated).

The result can be deployed as an API on the cloud or embedded into a device. Our preferred model is to charge per processed request. However, we can also hand the model off to you and deploy on your infrastructure with full release of the IP.

We welcome other partnerships suggestions.
Please write to us: sales@neuromation.io for more information.
Basic Research

Neuromation is engaged in basic research on the role of synthetic data in Deep Learning. We would love to share our findings and also collaborate with other teams.

If you think synthetic data can benefit your research effort Neuromation can engage our staff scientists to help you realize your goals. We can organize research teams for both short and long term projects. There is no specific business model to follow for this type of service. If the research subject is interesting enough Neuromation can sponsor the effort.

We are open to partnership ideas please write to us: partnerships@neuromation.io



Neuromation will distribute 10% of its token sale proceeds in computing power to startups that contribute to the development of neural networks.
Read more news

Contact Us

Kai 1-5M,
10111 Estonia