Neuron – Neural Network Support
Build a neural network over Go.
Installation
gvt fetch github.com/cacilhas/neuron
Use
sensors := []string{
"distance",
"height",
}
actions := []string{"jump"}
buildNeuron := func(length int) neuron.Neuron {
neu, err := neuron.NewNeuron(length)
if err != nil {
panic(err)
}
return neu
}
// 2 sensors:
front := neuron.Layer{buildNeuron(2), buildNeuron(2), buildNeuron(2)}
// 3 front neurons:
middle := neuron.Layer{buildNeuron(3), buildNeuron(3)}
// 2 middle neurons, one back one for each action:
back := neuron.Layer{buildNeuron(2)}
net := neuron.NewNeuralNet(sensors, actions, []neuron.Layer{front, middle, back})
Using the network:
params := map[string]float{
"distance": -12,
"height": 246.128,
}
res, _ := net.Compute(params)
if res["jump"] {
// Jump
}
Saving and retrieving
Save to file:
fp, err := os.Create("net.dna")
if err != nil {
panic(err)
}
defer fp.Close()
err = net.Save(fp)
if err != nil {
panic(err)
}
Load from file:
fp, err := os.Open("net.dna")
if err != nil {
panic(err)
}
defer fp.Close()
var net neuron.NeuralNet
net, err := neuron.LoadNet(fp)
if err != nil {
panic(err)
}
API
Neuron
(neuron
is the instance):
NewNeuron([]byte) (Neuron, error)
NewNeuron(*bytes.Buffer) (Neuron, error)
- Build a neuron from its binary representation.
NewNeuron([]int) (Neuron, error)
- Build a new neuron from the
int
array. Each integer represents a gene.
- Build a new neuron from the
NewNeuron(int) (Neuron, error)
- Build a new random neuron,
int
is the amount of genes.
- Build a new random neuron,
NewNeuron(io.Reader) (Neuron, error)
- Load a neuron from a stream.
NewNeuron(Neuron) (Neuron, error)
- Clone the neuron supplied.
NewNeuron(string) (Neuron, error)
- Deserialise a neuron.
neuron.Compute(...float64) int
- Compute the output from a list of parameters. There must be supplied as many parameters as genes.
neuron.Equals(Neuron) bool
- Check whether two neurons have the save genetic pool.
neuron.GetSize() int
- Return the neuron size (amount of genes).
neuron.GetGene(int) int
- Return the value of the gene in the index
int
.
- Return the value of the gene in the index
neuron.Marshal() <-chan byte
- Return a channel that supplies the neuron binary representation byte by byte.
neuron.Child(int) Neuron
- Return a new random child neuron, with the deviation
int
.
- Return a new random child neuron, with the deviation
neuron.String() string
- Return the neuron binary representation encoded on base32hex.
NeuralNet
(net
is the instance):
NewNeuralNet(sensors, actions []string, neurons []Layer) (NeuralNet, error)
- Create a new neural network given the parameters.
LoadNet(io.Reader) (NeuralNet, error)
- Load a neural network from a stream.
net.GetActions() []string
- Return the neural network’s actions.
net.Compute(map[string]float64) (map[string]bool, error)
- Compute the processing. The
map[string]float64
parameter must supply one key for each network’s sensor, and themap[string]bool
brings if each action must be performed.
- Compute the processing. The
net.GetChild(int) NeuralNet
- Return a new random child neural network, with the deviation
int
.
- Return a new random child neural network, with the deviation
net.GetSensors() []string
- Return the neural network’s sensors.
net.Neurons(index) []Neuron
- Return the neural network’s
int
layer of neurons (nil
ifint
is too big).
- Return the neural network’s
net.Save(io.Writer) error
- Save the neural network into a stream.
net.String() string
- Return the neural network serialisation.
neuron.Layer
is a layer of neurons:
type Layer []Neuron
License
Copyright 2019 Rodrigo Cacilhas batalema@cacilhas.info
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
-
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
-
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
-
Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.