Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/neuralegion/crystal-fann
FANN (Fast Artifical Neural Network) binding in Crystal
https://github.com/neuralegion/crystal-fann
crystal fann machine-learning neural-network
Last synced: 22 days ago
JSON representation
FANN (Fast Artifical Neural Network) binding in Crystal
- Host: GitHub
- URL: https://github.com/neuralegion/crystal-fann
- Owner: NeuraLegion
- License: mit
- Created: 2017-06-30T14:17:57.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2021-11-28T14:12:37.000Z (about 3 years ago)
- Last Synced: 2024-05-28T20:14:47.532Z (7 months ago)
- Topics: crystal, fann, machine-learning, neural-network
- Language: Crystal
- Size: 43.9 KB
- Stars: 80
- Watchers: 11
- Forks: 5
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# crystal-fann
[](https://gitter.im/crystal-fann/Lobby?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge)
[](https://travis-ci.org/NeuraLegion/crystal-fann)
Crystal bindings for the FANN C lib
## Installation
Add this to your application's `shard.yml`:
```yaml
dependencies:
crystal-fann:
github: NeuraLegion/crystal-fann
```
## Usage
Look at the spec for most functions
```crystal
ann = Fann::Network::Standard.new(2, [2], 1)
ann.randomize_weights(0.0, 1.0)
3000.times do
ann.train_single([1.0, 0.0], [0.5])
end
result = ann.run([1.0, 0.0])
# Remember to close the network when done to free allocated C memory
ann.close
(result < [0.55] && result > [0.45]).should be_true
```
```crystal
# Work on array of test data (batch)
ann = Fann::Network::Standard.new(2, [3], 1)
input = [[0.0, 0.0], [0.0, 1.0], [1.0, 0.0], [1.0, 1.0]]
output = [[0.0], [1.0], [1.0], [0.0]]
train_data = Fann::TrainData.new(input, output)
data = train_data.train_data
ann.randomize_weights(0.0, 1.0)
if data
ann.train_batch(data, {:max_runs => 8000, :desired_mse => 0.001, :log_each => 1000})
end
result = ann.run([1.0, 1.0])
ann.close
(result < [0.15]).should be_true
```
```crystal
# Work on array of test data using the Cascade2 algorithm (no hidden layers, net will build it alone)
ann = Fann::Network::Cascade.new(2, 1)
input = [[0.0, 0.0], [0.0, 1.0], [1.0, 0.0], [1.0, 1.0]]
output = [[0.0], [1.0], [1.0], [0.0]]
train_data = Fann::TrainData.new(input, output)
data = train_data.train_data
ann.train_algorithm(LibFANN::TrainEnum::TrainRprop)
ann.randomize_weights(0.0, 1.0)
if data
ann.train_batch(data, {:max_neurons => 500, :desired_mse => 0.001, :log_each => 10})
end
result = ann.run([1.0, 1.0])
ann.close
(result < [0.1]).should be_true
```
## Development
All C lib docs can be found here -> http://libfann.github.io/fann/docs/files/fann-h.html
- [x] Add TrainData class
- [x] Add network call method to train on train data
- [x] Add binding to the 'Parallel' binding to work on multi CPU at same time
- [ ] Clean unneeded bindings in the LibFANN binding
- [ ] Add specific Exceptions
- [ ] Add binding and checks for lib errors
I guess more stuff will be added once more people will use it.
## Contributing
1. Fork it ( https://github.com/NeuraLegion/crystal-fann/fork )
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create a new Pull Request
## Contributors
- [bararchy](https://github.com/bararchy) - creator, maintainer
- [libfann](https://github.com/libfann/fann) - c lib creators