Machine Learning

We use the most advanced Machine Learning
and Artificial Intelligence
tools to provide our
clients with ground-breaking analytics and
forecasting
solutions.

Omnys is the first Amazon Forecast partner in the EMEA region

Omnys has been officially acknowledged by Amazon as the 1st European partner who remarkably used, in an appropriate and skillfull way, the Amazon Forecast service on real-life industrial use cases.

Visit the Amazon Forecast page >

Machine Learning and Big Data

We have used BIG DATA infrastructures in a number of scenarios: from IoT to Machine Learning, from Social Networks to Data Analysis in order to:

  • store and process a huge amount of data gathered from sensors (over 100 thousand collections a day per device);
  • develop dashboards aimed at analysing and displaying data in real-time;
  • design predictive systems based on Machine Learning algorithms (Multi-class, Regression- or Binary-based) able to generate predictions in real-time;
  • manage multi-level relationships (friendship, kinship, interests on more than 4 million users) so as to provide social stream, likes, comments, search results, in real-time.

We exploited our know-how and expertise in Artificial Intelligence by developing a platform performing Deep Learning algorithms to train models and provide predictions

READY-TO-USE
It is a platform designed to be used and managed by non-expert users.


HIGHLY SCALABLE
It is built on application layers that are horizontally scalable and able to recognize patterns by analyzing huge amount of data.

MODEL SELF-DISCOVERY
It can find out by itself the best model to be applied to data under analysis, while user doesn't need to make any choice or setting.

MULTI DOMAIN
It is not focused on nor targeted to only one specific domain: it is able to analyze and manage any kind of data, regardless of the domain the data belong to.

MODEL SELF-IMPROVEMENT AND CLEANING
It embeds a model self-improving capability, for the sake of a feature that allows to add authomatically new data and to clear the data-set from fake data.

Supported Algorithms:
Neural Networks, Random Forest, SVM and Logistic Regression, Probabilistic Classifier, Linear regression, Isotonic regression, Survival regression, Gradient boost regressor, LSTM Neural Networks, IsolationForest, LGBRegressor, LGBClassifier.