xtractis products & services
The xtractis AI brings you all functions to efficiently carry out your predictive modeling projects: discovery and maintenance of your models and their use in production, without programming nor framework.
You are attended by INTELLITECH at all stages of your project, from the preparation of data to the deployment of your predictive decision-making systems and their certification.
A unique technology in a fully automated solution
Key features of xtractis
SELF-LEARNING
GENERATE robots automatically discovers predictive models that are robust and explainable, from multidimensional datasets, through a collective and evolutionary machine learning.
Fully automated solution: no programming, nor framework. Software interface controlled by mouse click.
Because of the intensive assessment of robustness and the continuous improvement of learning strategies, it takes a few days to a few weeks before converging to the most robust models. However, the more computing energy is available, the more this generation time is reduced.
PREDICTIONS
PREDICT uses the xtractis models to deduce predicted responses of the process in real time, for a new real or simulated input situation.
Inference engine delivers prediction instantaneously.
PRESCRIPTIONS
OPTIMIZE uses the xtractis models to discover the optimal solutions that satisfy your multi-objective request.
A few tens of seconds are enough to find solutions in multi-dimensional space.
AUTO-MAINTENANCE
MONITOR continuously monitors the performance of the deployed models, builds the new learning datasets and requires from GENERATE to regenerate the expired models.
Features
Automatic discovery of best learning strategies
to automatically build best predictive models
Automatic selection of the predictive variables, Determination of the influence of each variable
Multicore CPU Massively Parallel Calculations to manage large datasets or to launch multiple robots simultaneously
Intensive analysis of the robustness of the generated models by different cross-validation techniques (LOO, MC, Validation/Testing)
Regression, Multiclassification
or Scoring Modelling
Clustering
by unsupervised learning
Visualization of the decision rules, visualization of the individual contributions of the variables for each prediction
Instant and simultaneous Predictions for several studied variables with several models
Real time and simultaneous Predictions for several studied variables with several models
Definition of multi-objective requests on the studied variables and constraints on the predictive variables
Simultaneous inversion of several models.
Discovery of exact optimal solutions
Dynamic exploration of the decision surface
(2D/3D cross sections)
Performance monitoring of models used in production,
Detection of behavioral change, Building of new learning datasets
Monitoring of GENERATE Robots for the regeneration of expired models
Reporting of decision making traceability
Web Service through query by client applications running on any type of OS
Temporary License
Operating System
xtractis
GENERATE
Single-station
multi-session
WINDOWS
/ WINDOWS SERVER
xtractis
PREDICT
Single-station single-session / Floating
or
Webservice (PREDICT Server)
WINDOWS
/ WINDOWS SERVER
xtractis
OPTIMIZE
Single-station single-session / Floating
or
Webservice (OPTIMIZE Server)
WINDOWS
/ WINDOWS SERVER
xtractis
MONITOR
Single-station
mono-session
WINDOWS
/ WINDOWS SERVER
xtractis SERVER version
High level services
to support you in your AI digital transformation
TRAINING
Personalized training on predictive data mining techniques, on the xtractis Augmented Fuzzy AI approach, and on the deployment of xtractis models in your business processes.
COACHING
The objective of the coaching sessions is to verify that your modeling problems are posed correctly, to analyze with you your data and the results obtained...
ASSISTANCE
In certain cases, INTELLITECH can assist you in auditing your predictive systems with the regulator or the certifying authority.
PILOT PROJECT
Minimum service to test xtractis on your data. We mobilize our resources to present to you in less than 3 months the maximum performance obtained on a selected issue.
BENCHMARKS
Full benchmarking against other statistical techniques and Open Source AI: PLS, Logistic Regression, Polynomial Regression, CART Decision Tree, Random Forest, Boosted Trees, KSVM, Deep Learning / Neural Networks...
