module SHAInet

Included Modules

• Apatite

Defined in:

shainet.cr
shainet/basic/es.cr
shainet/basic/exceptions.cr
shainet/basic/exprimental.cr
shainet/basic/layer.cr
shainet/basic/network_run.cr
shainet/basic/network_setup.cr
shainet/basic/neuron.cr
shainet/basic/synapse.cr
shainet/cnn/cnn.cr
shainet/cnn/cnn_input_layer.cr
shainet/cnn/cnn_synapse.cr
shainet/cnn/conv_layer.cr
shainet/cnn/drop_out_layer.cr
shainet/cnn/fc_layer.cr
shainet/cnn/filter.cr
shainet/cnn/max_pool_layer.cr
shainet/cnn/relu_layer.cr
shainet/cnn/softmax_layer.cr
shainet/data/cnn_data.cr
shainet/data/data.cr
shainet/data/json_data.cr
shainet/data/test_data.cr
shainet/data/training_data.cr
shainet/math/functions.cr
shainet/math/random_normal.cr

Constant Summary

NEURON_TYPES = ["memory", "eraser", "amplifier", "fader", "sensor"]

Each type of neuron uses and propogates data differently

Class Method Summary

• ._bp_sigmoid(value : GenNum) : Float64
• ._bp_sigmoid_prime(value : GenNum) : Float64
• ._cross_entropy_cost(expected : Float64, actual : Float64) : Float64
• ._cross_entropy_cost_derivative(expected : Float64, actual : Float64) : Float64
• ._l_relu(value : GenNum, slope : Float64 = 0.01) : Float64
• ._l_relu_prime(value : GenNum, slope : Float64 = 0.01) : Float64
• ._log_sigmoid(value : GenNum) : Float64
• ._log_sigmoid_prime(value : GenNum) : Float64
• ._quadratic_cost(expected : Float64, actual : Float64) : Float64
• ._quadratic_cost_derivative(expected : Float64, actual : Float64) : Float64

  Derivatives of cost functions #

• ._relu(value : GenNum)
• ._relu_prime(value : GenNum) : Float64
• ._sigmoid(value : GenNum) : Float64
• ._sigmoid_prime(value : GenNum) : Float64
• ._tanh(value : GenNum) : Float64
• ._tanh_prime(value : GenNum) : Float64
• .bp_sigmoid : ActivationFunction
• .cross_entropy_cost : CostFunction
• .l_relu : ActivationFunction
• .log_sigmoid : ActivationFunction
• .log_softmax(array : Array(GenNum)) : Array(Float64)

  Not working yet, do not use

• .none : ActivationFunction
• .normalize_stcv(payloads : Array(String))

  translate an array of strings to one-hot vector matrix and hash dictionary

• .quadratic_cost : CostFunction
• .relu : ActivationFunction
• .sigmoid : ActivationFunction
• .sign(input : GenNum)

  Used in Rprop

• .softmax(array : Array(GenNum)) : Array(Float64)
• .softmax_prime(array : Array(GenNum)) : Array(Float64)

  The input array in this case has to be the output array of the softmax function

• .tanh : ActivationFunction

Class Method Detail