-
Notifications
You must be signed in to change notification settings - Fork 2
/
train_vol_autoencoder.lua
243 lines (219 loc) · 7.67 KB
/
train_vol_autoencoder.lua
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
--this script trains a volumetric auto-encoder for 10 classes
--with pure SGD. default training parameters are already entered.
--if you do not have cutorch or CUDA installed, comment the cuda
--source code.
require 'torch'
require 'nn'
require 'nnx'
require 'optim'
require 'cutorch'
require 'cunn'
require 'pl'
require 'paths'
local matio = require 'matio'
----------------------------------------------------------------------
-- parse command-line options
local opt = lapp[[
-s,--save (default "mul-class/AE_6912_.1_10class_r_dummy/") subdirectory to save logs
-p,--plot plot while training
-o,--optimization (default "SGD") optimization: SGD | LBFGS
-r,--learningRate (default 0.1) learning rate, for SGD only
-b,--batchSize (default 1) batch size
-m,--momentum (default 0.9) momentum, for SGD only
]]
-- fix seed
torch.manualSeed(1234)
torch.setdefaulttensortype('torch.FloatTensor')
--define the voxel/input resolution
local inD = 30
local featuresOut = inD * inD* inD
local cube_size = inD
-- define the data-struct for stroing training and test data
trainData = {
data = {},
labels = {},
size = function() return trsize end
}
testData = {
data = {},
labels = {},
size = function() return tesize end
}
-- load the input data
trainData.data = matio.load('Data/mul-class/mul-class_tr_10_reduced.mat', 'tr_data_10_class_r')
trsize = trainData.data:size()[1]
print ('no of training exmaple ='..trsize)
trainData.labels = torch.reshape(trainData.data,trsize ,featuresOut)
testData.data = matio.load('Data/mul-class/mul-class_te_10_reduced.mat', 'te_data_10_class_r')
tesize = testData.data:size()[1]
print( 'no of testing exmaple ='..tesize)
testData.labels = torch.reshape(testData.data,tesize ,featuresOut)
-- define the path for log files
trainLogger = optim.Logger(paths.concat(opt.save, 'train.log'))
testLogger = optim.Logger(paths.concat(opt.save, 'test.log'))
--define the model parameters
desc_dims = {6912,6912} --dims of desc
fSize = {1,64,256,256,64,1} -- no of feature maps at each layer
filtsize = {9,4,5,6} --size of filters in conv-deconv layers
local dT = {2,3} --stride for deconv
local kT= 3 --upsampling (local outD = (5-1) * dT + kT)
dropout_p = .5
model= nn.Sequential()
model:add(nn.Dropout(dropout_p))
model:add(nn.VolumetricConvolution(fSize[1], fSize[2], filtsize[1], filtsize[1], filtsize[1], 3, 3, 3)) -- (30 - 9 + 3)/3 = 8
model:add(nn.ReLU(true))
--features:add(nn.SpatialMaxPooling(2,2,2,2)) -- 26
model:add(nn.VolumetricConvolution(fSize[2], fSize[3], filtsize[2], filtsize[2], filtsize[2], 2, 2, 2)) -- 3^3
model:add(nn.ReLU(true))
model:add(nn.Reshape(desc_dims[1]))
model:add(nn.Linear(desc_dims[1],desc_dims[2]))
model:add(nn.ReLU(true))
model:add(nn.Dropout(dropout_p))
model:add(nn.Reshape(fSize[4],3,3,3 ))
--Deconvolutional layers
model:add(nn.VolumetricFullConvolution(fSize[4], fSize[5], filtsize[3], filtsize[3], filtsize[3], dT[1], dT[1], dT[1] ))
model:add(nn.ReLU(true))
model:add(nn.VolumetricFullConvolution(fSize[5], fSize[6], filtsize[4], filtsize[4], filtsize[4], dT[2], dT[2], dT[2] ))
model:add(nn.Reshape(featuresOut))
model:add(nn.Sigmoid())
print(model)
----------------------------------------------------
-- loss function: negative log-likelihood
criterion = nn.BCECriterion()
model:cuda()
criterion:cuda()
----------------------------------------------------------------------
print('loading params from the NN')
parameters,gradParameters = model:getParameters()
--define the optimizer
optimState = {
learningRate = opt.learningRate,
--weightDecay = opt.weightDecay,
momentum = opt.momentum,
learningRateDecay = 5e-7
}
optimMethod = optim.sgd
-- training function
function train(dataset)
-- epoch tracker
model:training()
epoch = epoch or 1
-- local vars
local time = sys.clock()
local shuffle = torch.randperm(trsize)
print '==> defining some tools'
-- do one epoch
print('<trainer> on training set:')
print("<trainer> online epoch # " .. epoch .. ' [batchSize = ' .. opt.batchSize .. ']')
for t = 1,dataset.data:size()[1],opt.batchSize do
-- create mini batch
local inputs = torch.Tensor(opt.batchSize,1,cube_size ,cube_size , cube_size )
local targets = torch.Tensor(opt.batchSize, cube_size*cube_size*cube_size)
inputs = inputs:cuda()
targets = targets:cuda()
local k = 1
for i = t,math.min(t+opt.batchSize-1,dataset.data:size()[1]) do
-- load new sample
local input = dataset.data[i]
local target = dataset.labels[i]
input = input:cuda()
target = target:cuda()
--target = target:squeeze()
inputs[k] = input
targets[k] = target
k = k + 1
end
-- create closure to evaluate f(X) and df/dX
local feval = function(x)
-- just in case:
--collectgarbage()
-- get new parameters
if x ~= parameters then
parameters:copy(x)
end
-- reset gradients
gradParameters:zero()
-- evaluate function for complete mini batch
local outputs = model:forward(inputs)
local f = criterion:forward(outputs, targets)
-- estimate df/dW
local df_do = criterion:backward(outputs, targets)
model:backward(inputs, df_do)
-- return f and df/dX
return f,gradParameters
end
-- optimize on current mini-batch
-- Perform SGD step:
optimMethod(feval, parameters, optimState)
-- disp progress
--xlua.progress(t, dataset.data:size()[1])
end
-- time taken
time = sys.clock() - time
time = time / dataset.data:size()[1]
print("<trainer> time to learn 1 sample = " .. (time*1000) .. 'ms')
-- save/log current net
local filename = paths.concat(opt.save, 'model.net')
os.execute('mkdir -p ' .. sys.dirname(filename))
if paths.filep(filename) then
os.execute('mv ' .. filename .. ' ' .. filename .. '.old')
end
if epoch%5== 0 then
print('<trainer> saving network to '..filename)
torch.save(filename, model)
end
-- next epoch
epoch = epoch + 1
end
-- test function
if 0 then
function test(dataset)
-- local vars
local time = sys.clock()
-- test over given dataset
print('<trainer> on testing Set:')
for t = 1,dataset:size(),opt.batchSize do
-- disp progress
xlua.progress(t, dataset:size())
-- create mini batch
local inputs = torch.Tensor(opt.batchSize,1,geometry[1],geometry[2])
local targets = torch.Tensor(opt.batchSize)
local k = 1
for i = t,math.min(t+opt.batchSize-1,dataset:size()) do
-- load new sample
local sample = dataset[i]
local input = sample[1]:clone()
local _,target = sample[2]:clone():max(1)
target = target:squeeze()
inputs[k] = input
targets[k] = target
k = k + 1
end
-- test samples
local preds = model:forward(inputs)
-- confusion:
for i = 1,opt.batchSize do
confusion:add(preds[i], targets[i])
end
end
-- timing
time = sys.clock() - time
time = time / dataset:size()
print("<trainer> time to test 1 sample = " .. (time*1000) .. 'ms')
end
end
----------------------------------------------------------------------
-- and train!
--
while true do
-- train/test
train(trainData)
--test(testData)
-- plot errors
if opt.plot then
trainLogger:style{['% mean class accuracy (train set)'] = '-'}
--testLogger:style{['% mean class accuracy (test set)'] = '-'}
trainLogger:plot()
--testLogger:plot()
end
end