https://github.com/simon-lc/nablanet.jl

Differentiable neural network with easy access to Jacobians of the output wrt the input and the network parameters.
https://github.com/simon-lc/nablanet.jl

Last synced: about 1 month ago
JSON representation

Differentiable neural network with easy access to Jacobians of the output wrt the input and the network parameters.

• Host: GitHub
• URL: https://github.com/simon-lc/nablanet.jl
• Owner: simon-lc
• License: mit
• Created: 2022-05-07T07:02:35.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-05-13T21:25:34.000Z (over 2 years ago)
• Last Synced: 2024-11-11T08:46:05.241Z (about 2 months ago)
• Language: Julia
• Size: 18.6 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# NablaNet.jl

[![CI](https://github.com/simon-lc/NablaNet.jl/actions/workflows/CI.yml/badge.svg)](https://github.com/simon-lc/NablaNet.jl/actions/workflows/CI.yml)

[![codecov](https://codecov.io/gh/simon-lc/NablaNet.jl/branch/main/graph/badge.svg?token=CHJNI2LRNZ)](https://codecov.io/gh/simon-lc/NablaNet.jl)

With NablaNet.jl you can quickly build fully-connected neural networks.

NablaNet.jl provides methods for fast and allocation-free evaluations of the neural network for a given input and given parameters.

It also provides methods for fast and allocation-free computations of the Jacobians of the output with respect to the input and the network parameters.

## Create a fully connected neural network with nonlinear activations

```

## Dimensions

# input size = 3

ni = 3

# final output size = 4

no = 4

```

## Neural Network

```

# generate a random input and some random parameters

xi = rand(ni)

# create neural network with 4 fully connected layers of sizes 6, 2, 3, no

# the activation functions are tanh, tanh, tanh, and identity for the last layer

net = Net(ni, no, dim_layers=[6,2,3], activations=[x->tanh.(x), x->tanh.(x), x->tanh.(x), x->x])

# generate a random input and some random parameters

θ = rand(NablaNet.parameter_dimension(net))

```

## Computations and Gradients

### Evaluate the output of the neural network for given input and parameters

```

evaluation!(net, xi, θ)

get_output(net)

```

### Jacobian of the output with respect ot the input

```

jacobian_input!(net, xi, θ)

net.jacobian_input

```

### Jacobian of the output with respect ot the input

```

jacobian_parameters!(net, xi, θ)

net.jacobian_parameters

```

See [this example](../main/examples/net.jl) for more details.