Issue description

• Hi! I have been going through the .hpp files in src/mlpack/methods/ann/layer and found some discrepancies regarding accessor implementations (getter and setter methods for the private variables). Specifically, in quite a large number of files, some of these methods are missing and thus makes the files a bit incomplete. I am listing out the ones I found below (there might be a few that I missed).
• Also, I am not exactly sure if all the stuff I am listing below need to have accessor implementations (for example, some variable might have been left private without an accessor on purpose.) So, I would really appreciate if someone provides some clue as to accessor methods I should go about implementing for which of the variables from the list below.
• I intend to implement all those methods and submit a PR very soon with the changes.

---

List :

add_merge.hpp

• OutputDataType gradient

add.hpp

• size_t outSize

atrous_convolution.hpp

• size_t batchSize
• arma::cube weight

batch_norm.hpp

• size_t size
• double eps
• bool loading
• OutputDataType gamma
• OutputDataType beta
• OutputDataType mean
• OutputDataType variance
• OutputDataType normalize
• OutputDataType inputMean

bilinear_interpolation.hpp

• size_t inRowSize
• size_t inColSize
• size_t outRowSize
• size_t outColSize
• size_t depth
• size_t batchSize

concat_performance.hpp

• OutputLayerType outputLayer
• size_t inSize

concat.hpp

• size_t axis
• bool useAxis
• bool model
• size_t channels

concatenate.hpp

• size_t inRows

constant.hpp

• size_t inSize
• size_t outSize
• OutputDataType constantOutput

convolution.hpp

• size_t batchSize
• arma::cube weight

dropconnect.hpp

• double scale
• OutputDataType mask
• OutputDataType denoise

dropout.hpp

• double scale
• OutputDataType mask

elu.hpp

• bool deterministic
• arma::mat derivative

fast_lstm.hpp

• size_t inSize
• size_t outSize
• size_t forwardStep
• size_t backwardStep
• size_t gradientStep
• OutputDataType prevOutput
• size_t batchSize
• size_t batchStep
• size_t gradientStepIdx
• size_t rhoSize
• size_t bpttSteps

flexible_relu.hpp

• possible error on lines 134-136

glimpse.hpp

• size_t inSize
• size_t size
• size_t depth
• size_t scale
• size_t inputDepth

gru.hpp

• size_t inSize
• size_t outSize
• size_t batchSize
• size_t forwardStep
• size_t backwardStep
• size_t gradientStep
• arma::mat prevError

highway.hpp

• size_t inSize
• bool model
• bool reset
• OutputDataType transformWeight
• OutputDataType transformBias
• OutputDataType transformGate
• OutputDataType transformGateActivation
• OutputDataType transformGateError
• size_t width
• size_t height
• OutputDataType networkOutput

join.hpp

• size_t inSizeRows
• size_t inSizeCols

layer_norm.hpp

• size_t size
• double eps
• bool loading
• OutputDataType gamma
• OutputDataType beta
• OutputDataType normalized
• OutputDataType inputMean

lookup.hpp

• size_t inSize
• size_t outSize

lstm.hpp

• size_t inSize
• size_t outSize
• size_t rho
• size_t forwardStep
• size_t backwardStep
• size_t gradientStep
• OutputDataType prevOutput
• size_t batchSize
• size_t gradientStepIdx
• size_t rhoSize
• size_t bpttSteps

max_pooling.hpp

• bool reset
• size_t offset
• size_t batchSize
• OutputDataType gradient

mean_pooling.hpp

• bool reset
• size_t offset
• size_t batchSize
• OutputDataType gradient

minibatch_discrimination.hpp

• size_t batchSize
• OutputDataType weight
• arma::cube M
• arma::cube distances

multiply_merge.hpp

• bool model
• bool run
• bool ownsLayer

recurrent_attention.hpp

• size_t outSize
• size_t rho
• size_t forwardStep
• size_t backwardStep
• bool deterministic
• std::vectorarma::mat feedbackOutputParameter
• std::vectorarma::mat moduleOutputParameter
• arma::mat recurrentError
• arma::mat actionError
• arma::mat actionDelta
• arma::mat rnnDelta
• arma::mat initialInput
• arma::mat attentionGradient

recurrent.hpp

• size_t rho
• size_t forwardStep
• size_t backwardStep
• size_t gradientStep
• bool deterministic
• bool ownsLayer
• std::vectorarma::mat feedbackOutputParameter
• arma::mat recurrentError

reinforce_normal.hpp

• double stdev

reparametrization.hpp

• bool stochastic
• bool includeKl
• double beta
• OutputDataType gaussianSample
• OutputDataType mean
• OutputDataType preStdDev
• OutputDataType stdDev

select.hpp

• size_t index
• size_t elements

sequential.hpp

• bool model
• bool reset

subview.hpp

• size_t inSize
• size_t beginRow
• size_t endRow
• size_t beginCol
• size_t endCol

transposed_convolution.hpp

• size_t aW
• size_t aH

virtual_batch_norm.hpp

• size_t size
• double eps
• bool loading
• OutputDataType gamma
• OutputDataType beta
• OutputDataType referenceBatchMean
• OutputDataType referenceBatchMeanSquared
• double oldCoefficient
• double newCoefficient
• OutputDataType mean
• OutputDataType variance
• OutputDataType inputParameter
• OutputDataType normalized
• OutputDataType inputSubMean

vr_class_reward.hpp

• std::vector<LayerTypes<> > network
• double reward
• bool sizeAverage
• double scale

weight_norm.hpp

• size_t biasWeightSize
• size_t layerWeightSize
• OutputDataType scalarParameter
• OutputDataType vectorParameter
• OutputDataType layerGradients
• OutputDataType layerWeights