+ Tools
Detego - Intelligent Solutions Uncovered
Introducing Detego
Detego was developed out of a need from NORAD USSPACECOM to rapidly implement reconfigurable sensor anomaly detectors that could train on extensive historical data sets and monitor multiple real-time live data feeds. The result is an easy to use neural network tool that operates in three modes: initialization, training, and testing. The Initialization Mode allows the user to configure neural networks using any number of inputs and output nodes, along with zero to two hidden layers.
The Training Mode allows the neural network to learn the relationships specified in the training data file, and is optimized via data encoding schemes and learning optimizations. The GUI provides real-time training performance updates and is highly interactive.
The Testing Mode allows the user to evaluate new data and perform statistical analysis on the solution.
Features At a Glance
- GUI driven engine allows users to effortlessly create, train, and utilize neural networks for classification and prediction
- Monitors training performance in real-time via graphs that display the RMS error of the entire training set and each training vector
- Real-time interaction allows users to modify training parameters while the neural network is training
- Statistical library analyzes the neural network solution
- Supports the use of neural network architectures with zero to two hidden layers
- Provides data encoding schemes and learning optimizations that significantly reduce training time