Our technology is unique. It stems from evolutionary algorithms. Moreover, this unique technology is integrated into a powerful platform: TADA, which allows you to predict the future based on past data. It is performant, starting with small quantities of data (a few hundred records) to millions. Lightweight, quick, light in energy consumption, our algorithms are the flagships of their category.
We made the bold choice of using genetic algorithms to create our models, we have fine-tuned our engine to reach the perfect compromise between complexity and speed. We do control the complexity of our models by choice.
On top of it, our algorithms are very efficient, starting on small data sets (a few hundred records) and staying very accurate when moving to big data sets (millions).
The combination of genetic algorithms with controlled complexity and often small data sets results in a platform that generates a model in a few minutes.
When using neural networks, big data is necessary for training. Therefore, the training usually requires data sets with a few million records.
Genetic algorithms learn efficiently, starting from a few hundred records. It means that the range of data sets usable with our technology is much broader than the range of data sets usually considered for AI platforms.
For example, 85% of the data generated in companies are so-called ‘small’ data sets. Typically, these data sets contain a few hundred to a few thousand records.
Using an efficient AI on small and large datasets alike unleashes the potential to address both the 85% of ‘small’ data sets and the 15% of big data.
We use symbolic regression to create models which are expressed as a mathematical formula. The resulting formula is readable by users. They can understand it. What they see is what they get. Next, the algorithm selects a few criteria out of the whole data set to create the formula. And these criteria appear in the resulting mathematical formula.
Furthermore, we provide several visualization tools. One of them displays the critical criteria selected in the data set to create the model and their respective importance (here). Another tool allows to dynamically vary a criterion and see its impact on the result (live predict).
Our AI provides the users with the means to understand the model, explore it. It is a white box model, a transparent model.
As a result, it creates trust between the users and the AI.
Our AI platform has shown exceptional results on data sets related to real life. For example, we get fantastic accuracy when modeling and making predictions based on medical data sets, industrial ones, customer behavior ones.
MyDataModels algorithms are the result of five years of research and development. Our technology is patented multiple times.
Half of our team is composed of data scientists. Recognized companies such as Biogen or Thales use our AI technology daily.
Our technology has been challenged and approved by European academic teams and renowned scientists such as Marc Schoenauer.
Big data and AI often rhyme with neural networks. Neural networks are a great technology, yet they are computationally intensive, power greedy. They also provide black box results, or results that are not readable, cannot be challenged or interpreted.
Our AI technology stems from genetic algorithms. It is less computationally greedy than neural networks. Therefore it consumes less power. Moreover, our AI platform is environmentally friendly.
Read more White Papers and Publications about our unique Technology