Chain(layers...)

Chain multiple layers / functions together, so that they are called in sequence on a given input.

m = Chain(x -> x^2, x -> x+1)
m(5) == 26

m = Chain(Dense(10, 5), Dense(5, 2))
x = rand(10)
m(x) == m[2](m[1](x))

Chain also supports indexing and slicing, e.g. m[2] or m[1:end-1]. m[1:3](x) will calculate the output of the first three layers.

Dense(in::Integer, out::Integer, σ = identity)

Creates a traditional Dense layer with parameters W and b.

y = σ.(W * x .+ b)

The input x must be a vector of length in, or a batch of vectors represented as an in × N matrix. The out y will be a vector or batch of length out.

julia> d = Dense(5, 2)
Dense(5, 2)

julia> d(rand(5))
Tracked 2-element Array{Float64,1}:
  0.00257447
  -0.00449443

Conv(size, in=>out)
Conv(size, in=>out, relu)

Standard convolutional layer. size should be a tuple like (2, 2). in and out specify the number of input and output channels respectively.

Example: Applying Conv layer to a 1-channel input using a 2x2 window size, giving us a 16-channel output. Output is activated with ReLU.

size = (2,2)
in = 1
out = 16 
Conv((2, 2), 1=>16, relu)

Data should be stored in WHCN order (width, height, # channels, # batches). In other words, a 100×100 RGB image would be a 100×100×3×1 array, and a batch of 50 would be a 100×100×3×50 array.

Takes the keyword arguments pad, stride and dilation.

MaxPool(k)

Max pooling layer. k stands for the size of the window for each dimension of the input.

Takes the keyword arguments pad and stride.

MeanPool(k)

Mean pooling layer. k stands for the size of the window for each dimension of the input.

Takes the keyword arguments pad and stride.

DepthwiseConv(size, in)
DepthwiseConv(size, in=>mul)
DepthwiseConv(size, in=>mul, relu)

Depthwise convolutional layer. size should be a tuple like (2, 2). in and mul specify the number of input channels and channel multiplier respectively. In case the mul is not specified it is taken as 1.

Data should be stored in WHCN order. In other words, a 100×100 RGB image would be a 100×100×3 array, and a batch of 50 would be a 100×100×3×50 array.

Takes the keyword arguments pad and stride.

ConvTranspose(size, in=>out)
ConvTranspose(size, in=>out, relu)

Standard convolutional transpose layer. size should be a tuple like (2, 2). in and out specify the number of input and output channels respectively. Data should be stored in WHCN order. In other words, a 100×100 RGB image would be a 100×100×3 array, and a batch of 50 would be a 100×100×3×50 array. Takes the keyword arguments pad, stride and dilation.

RNN(in::Integer, out::Integer, σ = tanh)

The most basic recurrent layer; essentially acts as a Dense layer, but with the output fed back into the input each time step.

LSTM(in::Integer, out::Integer)

Long Short Term Memory recurrent layer. Behaves like an RNN but generally exhibits a longer memory span over sequences.

See this article for a good overview of the internals.

GRU(in::Integer, out::Integer)

Gated Recurrent Unit layer. Behaves like an RNN but generally exhibits a longer memory span over sequences.

See this article for a good overview of the internals.

Recur(cell)

Recur takes a recurrent cell and makes it stateful, managing the hidden state in the background. cell should be a model of the form:

h, y = cell(h, x...)

For example, here's a recurrent network that keeps a running total of its inputs.

accum(h, x) = (h+x, x)
rnn = Flux.Recur(accum, 0)
rnn(2) # 2
rnn(3) # 3
rnn.state # 5
rnn.(1:10) # apply to a sequence
rnn.state # 60

Maxout(over)

Maxout is a neural network layer, which has a number of internal layers, which all have the same input, and the maxout returns the elementwise maximium of the internal layers' outputs.

Maxout over linear dense layers satisfies the univeral approximation theorem.

Reference: Ian J. Goodfellow, David Warde-Farley, Mehdi Mirza, Aaron Courville, and Yoshua Bengio.

1. Maxout networks.

In Proceedings of the 30th International Conference on International Conference on Machine Learning - Volume 28 (ICML'13), Sanjoy Dasgupta and David McAllester (Eds.), Vol. 28. JMLR.org III-1319-III-1327. https://arxiv.org/pdf/1302.4389.pdf

σ(x) = 1 / (1 + exp(-x))

Classic sigmoid activation function.

relu(x) = max(0, x)

Rectified Linear Unit activation function.

leakyrelu(x) = max(0.01x, x)

Leaky Rectified Linear Unit activation function. You can also specify the coefficient explicitly, e.g. leakyrelu(x, 0.01).

elu(x, α = 1) =
  x > 0 ? x : α * (exp(x) - 1)

Exponential Linear Unit activation function. See Fast and Accurate Deep Network Learning by Exponential Linear Units. You can also specify the coefficient explicitly, e.g. elu(x, 1).

swish(x) = x * σ(x)

Self-gated actvation function. See Swish: a Self-Gated Activation Function.

testmode!(m)
testmode!(m, false)

Put layers like Dropout and BatchNorm into testing mode (or back to training mode with false).

BatchNorm(channels::Integer, σ = identity;
          initβ = zeros, initγ = ones,
          ϵ = 1e-8, momentum = .1)

Batch Normalization layer. The channels input should be the size of the channel dimension in your data (see below).

Given an array with N dimensions, call the N-1th the channel dimension. (For a batch of feature vectors this is just the data dimension, for WHCN images it's the usual channel dimension.)

BatchNorm computes the mean and variance for each each W×H×1×N slice and shifts them to have a new mean and variance (corresponding to the learnable, per-channel bias and scale parameters).

See Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift.

Example:

m = Chain(
  Dense(28^2, 64),
  BatchNorm(64, relu),
  Dense(64, 10),
  BatchNorm(10),
  softmax)

Dropout(p)

A Dropout layer. For each input, either sets that input to 0 (with probability p) or scales it by 1/(1-p). This is used as a regularisation, i.e. it reduces overfitting during training.

Does nothing to the input once in testmode!.

AlphaDropout(p)

A dropout layer. It is used in Self-Normalizing Neural Networks. (https://papers.nips.cc/paper/6698-self-normalizing-neural-networks.pdf) The AlphaDropout layer ensures that mean and variance of activations remains the same as before.

LayerNorm(h::Integer)

A normalisation layer designed to be used with recurrent hidden states of size h. Normalises the mean/stddev of each input before applying a per-neuron gain/bias.

Group Normalization. This layer can outperform Batch-Normalization and Instance-Normalization.

GroupNorm(chs::Integer, G::Integer, λ = identity;
          initβ = (i) -> zeros(Float32, i), initγ = (i) -> ones(Float32, i), 
          ϵ = 1f-5, momentum = 0.1f0)

$chs$ is the number of channels, the channel dimension of your input. For an array of N dimensions, the (N-1)th index is the channel dimension.

$G$ is the number of groups along which the statistics would be computed. The number of channels must be an integer multiple of the number of groups.

Example:

m = Chain(Conv((3,3), 1=>32, leakyrelu;pad = 1),
          GroupNorm(32,16)) # 32 channels, 16 groups (G = 16), thus 2 channels per group used          

Link : https://arxiv.org/pdf/1803.08494.pdf