Knowledge Building

Load

Collect data from your sensors, visualize and annotate

Learn

Train the neurons using single or multiple features extracted from your examples

Validate

Verify the recognition of new datasets, identify cases of uncertainty, etc.

Apply

Save the neurons’ content for backup or export to a different deployment platform

NeuroMem Knowledge Builder

Experiment with a NeuroMem neural network to learn traning sets in a few mouse clicks, qualify the knowledge built by the neurons on new testing sets, validate the relevancy of your datasets and classify new datasets.

The NeuroMem Knowledge Builder is agnostic to data type and loads patterns extracted from text, audio, signal, image, video, etc.

Image Knowledge Builder

Experiment with a NeuroMem neural network and simple feature extractions from the General Vision library to learn discrete objects, surfaces, textures and scenes in images. Validate on new images before deploying your own applcation with the CogniSight SDK.

A simple workflow…

A Knowledge Builder (KB)  lets you train the neurons with annotated data sets and verify that the training is appropriate and sufficient to deliver recognition with acceptable accuracy and throughput for your application. The outputs are knowledge files which can then be loaded on systems equipped with neurons to reproduce the same recognition. Regardless of the type of data to comprehend, the Knowledge Builder features diagnostics guiding you through the process of learning for the best throughput and accuracy.