DN2A

DN2A is a set of highly decoupled JavaScript modules for Neural Networks and Artificial Intelligence development. Each module is based on injection by configuration.

Clients can use a single module alone, more of them together or just the complete set. DN2A’s main goal is to allow users to design, train and use Single Neural Networks with ease as well as very powerful Neural Networks Chains through which to implement their Artificial Intelligence solutions.

DN2A’s side goals are to simplify integration, to speed up training/querying, to allow clustering and to represent the architecture and the relative data of each Neural Network as a (re)combinable string strain that will be usable within genetics optimization techniques. Some of DN2A’s highlights include Modularized Components which aid the development and the clear separation of concerns with great benefits for those who want to use mixed solutions.

Configurable Precision helps to avoid the noise deriving from operation errors and default system precision limits with great improvement of the learning speed and performance stability.

Configuration Checker helps to write fewer details about configuration and to keep compatibility with older versions while the project evolves. StepByStep Training allows users to train neural networks doing a single iteration over the passed information without trying to reach a specific parametric condition.

StepByGoal Training enables users to train neural networks doing a finite or infinite number of iterations over the passed information unless a specific parametric condition is reached. Finally, Continuous Training helps to train neural networks doing an infinite number of iterations over the passed information.