Products & Solutions / By Solution / NeuroFusion / Feature Set

• Dataset size limited only by available memory
• Input data types: numerical and categorical
• Automatic preprocessing before training
• Automatic numerical columns scaling
• Automatic categorical columns encoding

• Automatic problem type determination (classification or regression)
• Automatic network design and training
• Automatic and manual control of training stopping conditions
• Generalization loss control
• Full control of the training process, including pausing and stopping
• Real-time information about training progress
• 3 types of network errors: AE, MSE, CCR

• The ability to save and load a network at any time you need
• Use a just created or loaded network to apply it to new data and get neural network reply

• Low resource requirements
• Example programs included in source form
• Support of load and save operations
• The Online Manual includes Chapter "Introduction in Neural Networks"