NeurEco, the automatic parsimonious neural networks factory
NeurEco introduces a new generation of neural networks, based on parsimony. Through NeurEco’s parsimonious approach, the resources required for implementing artificial intelligence are reduced by several order of magnitude. This includes reduction in learning data, computing resources, development time, and energy consumption.
NeurEco is an ANN (artificial neural network) factory. It generates parsimonious ANN models automatically, using only the learning data provided by the user.
For now, NeurEco 3.0 allows the user to work with two families of problems:
- Tabular problems: generating static ANN models,
- Dynamic problems: generating dynamic RNN models.
Note that in the near future, a new convolution-based family will be added to the product to handle the problems where the inputs are represented on regular grids (for example regression problems with images as inputs).
NeurEco can be used from the command line or via a user-friendly graphical interface and can accept the following data formats: csv, npy and MATLAB matfile.
It comes with an additional API for Python3, making it much easier to interact with standard AI environments.
It allows to export the generated models generated to C, ONNX or FMU.
This family of solutions handles the problems where samples in the tabular learning dataset all independent from one another: there is no dynamic link between the previous sample and the current sample.
There are currently three templates of tabular solutions:
Tabular Regression: This template of tabular solution is used to create neural network models where the outputs are values at the points of some continuous process (physical data, measured data…). NeurEco will create a regression predictive model that approximate the underlying process by a function (f) from inputs (X) to the output variable (Y).
Tabular Classification: This template of tabular solution is used to create neural network models performing supervised predictive classification. It is used when the output variable is a category and the model attempts to classify the data, that is draw conclusions from observations and predict categorical class labels.
Tabular compression: This template of tabular solution is used to create neural network models where the targets are the same as the inputs. These models will compress the input into a lower dimension representation and then reconstruct the output from this representation. The representation is a compact summary or compression of the input. This solution is a powerful tool to reduce dimensionality without losing the physical meaning of the inputs.
It is an add-on of NeurEco Tabular that handles the problems where the samples in the learning dataset represent a sequence of discrete-time data.
Continuous Dynamic: This template of dynamic solution is used to create neural network models that learn the differential equations that governs the dynamical outputs with respect to the inputs.
We invite you to order your licenses by filling the form below
Several types of licenses are available to meet your usage needs: single-user / floating, annual / perpetual