local DropoutLayer = nerv.class("nerv.DropoutLayer", "nerv.Layer")
function DropoutLayer:__init(id, global_conf, layer_conf)
nerv.Layer.__init(self, id, global_conf, layer_conf)
if self.gconf.dropout_rate == nil then
nerv.warning("[DropoutLayer:propagate] dropout rate is not set")
end
self:check_dim_len(1, 1) -- two inputs: nn output and label
end
function DropoutLayer:bind_params()
-- do nothing
end
function DropoutLayer:init(batch_size, chunk_size)
if self.dim_in[1] ~= self.dim_out[1] then
nerv.error("mismatching dimensions of input and output")
end
if chunk_size == nil then
chunk_size = 1
end
self.mask = {}
for t = 1, chunk_size do
self.mask[t] = self.mat_type(batch_size, self.dim_in[1])
end
end
function DropoutLayer:batch_resize(batch_size, chunk_size)
if chunk_size == nil then
chunk_size = 1
end
for t = 1, chunk_size do
if self.mask[t] == nil or self.mask[t]:nrow() ~= batch_size then
self.mask[t] = self.mat_type(batch_size, self.dim_in[1])
end
end
end
function DropoutLayer:propagate(input, output, t)
if t == nil then
t = 1
end
if self.gconf.dropout_rate then
self.mask[t]:rand_uniform()
-- since we will lose a portion of the actvations, we multiply the
-- activations by 1 / (1 - rate) to compensate
self.mask[t]:thres_mask(self.mask[t], self.gconf.dropout_rate,
0, 1 / (1.0 - self.gconf.dropout_rate))
output[1]:mul_elem(input[1], self.mask[t])
else
output[1]:copy_fromd(input[1])
end
end
function DropoutLayer:update(bp_err, input, output, t)
-- no params, therefore do nothing
end
function DropoutLayer:back_propagate(bp_err, next_bp_err, input, output, t)
if t == nil then
t = 1
end
if self.gconf.dropout_rate then
next_bp_err[1]:mul_elem(bp_err[1], self.mask[t])
else
next_bp_err[1]:copy_fromd(bp_err[1])
end
end
function DropoutLayer:get_params()
return nerv.ParamRepo({}, self.loc_type)
end