class NeuralNet

• NeuralNet
• Reference
• Object

Defined in:

neural_net_crystal/neural_net.cr

Constant Summary

DEFAULT_TRAINING_OPTIONS = {max_iterations: 1000, error_threshold: 0.01}

MAX_STEP = 50_f64

MIN_STEP = Math.exp(-6)

ZERO_TOLERANCE = Math.exp(-16)

Constructors

• .new(shape : Array(Int32))

Instance Method Summary

• #outputs : Array(Array(Float64))?
• #run(input : Array(Float64)) : Array(Float64)
• #shape : Array(Int32)
• #train(inputs, expected_outputs, opts = {} of Symbol => Float64 | Int32)
• #weight_update_values : Hash(Int32, Array(Array(Float64)))?
• #weight_update_values=(weight_update_values : Nil | Hash(Int32, Array(Array(Float64))))
• #weights : Hash(Int32, Array(Array(Float64)))?
• #weights=(weights : Nil | Hash(Int32, Array(Array(Float64))))

Constructor Detail

Instance Method Detail