CNNSupervisedTrainer_CifarClassifierNetwork.py 19.1 KB
Newer Older
Nicola Gatto's avatar
Nicola Gatto committed
1
2
3
4
5
6
import mxnet as mx
import logging
import numpy as np
import time
import os
import shutil
Christian Fuß's avatar
Christian Fuß committed
7
import pickle
Sebastian Nickels's avatar
Sebastian Nickels committed
8
9
import math
import sys
Nicola Gatto's avatar
Nicola Gatto committed
10
11
from mxnet import gluon, autograd, nd

Eyüp Harputlu's avatar
Eyüp Harputlu committed
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
class CrossEntropyLoss(gluon.loss.Loss):
    def __init__(self, axis=-1, sparse_label=True, weight=None, batch_axis=0, **kwargs):
        super(CrossEntropyLoss, self).__init__(weight, batch_axis, **kwargs)
        self._axis = axis
        self._sparse_label = sparse_label

    def hybrid_forward(self, F, pred, label, sample_weight=None):
        pred = F.log(pred)
        if self._sparse_label:
            loss = -F.pick(pred, label, axis=self._axis, keepdims=True)
        else:
            label = gluon.loss._reshape_like(F, label, pred)
            loss = -F.sum(pred * label, axis=self._axis, keepdims=True)
        loss = gluon.loss._apply_weighting(F, loss, self._weight, sample_weight)
        return F.mean(loss, axis=self._batch_axis, exclude=True)

Eyüp Harputlu's avatar
Eyüp Harputlu committed
28
29
30
31
32
33
34
35
36
class LogCoshLoss(gluon.loss.Loss):
    def __init__(self, weight=None, batch_axis=0, **kwargs):
        super(LogCoshLoss, self).__init__(weight, batch_axis, **kwargs)

    def hybrid_forward(self, F, pred, label, sample_weight=None):
        loss = F.log(F.cosh(pred - label))
        loss = gluon.loss._apply_weighting(F, loss, self._weight, sample_weight)
        return F.mean(loss, axis=self._batch_axis, exclude=True)

Sebastian Nickels's avatar
Sebastian Nickels committed
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
class SoftmaxCrossEntropyLossIgnoreIndices(gluon.loss.Loss):
    def __init__(self, axis=-1, ignore_indices=[], sparse_label=True, from_logits=False, weight=None, batch_axis=0, **kwargs):
        super(SoftmaxCrossEntropyLossIgnoreIndices, self).__init__(weight, batch_axis, **kwargs)
        self._axis = axis
        self._ignore_indices = ignore_indices
        self._sparse_label = sparse_label
        self._from_logits = from_logits

    def hybrid_forward(self, F, pred, label, sample_weight=None):
        log_softmax = F.log_softmax
        pick = F.pick
        if not self._from_logits:
            pred = log_softmax(pred, self._axis)
        if self._sparse_label:
            loss = -pick(pred, label, axis=self._axis, keepdims=True)
        else:
            label = _reshape_like(F, label, pred)
            loss = -(pred * label).sum(axis=self._axis, keepdims=True)
        #loss = _apply_weighting(F, loss, self._weight, sample_weight)
        # ignore some indices for loss, e.g. <pad> tokens in NLP applications
        for i in self._ignore_indices:
            loss = loss * mx.nd.logical_not(mx.nd.equal(mx.nd.argmax(pred, axis=1), mx.nd.ones_like(mx.nd.argmax(pred, axis=1))*i))
        return loss.mean(axis=self._batch_axis, exclude=True)

Sebastian Nickels's avatar
Sebastian Nickels committed
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
@mx.metric.register
class BLEU(mx.metric.EvalMetric):
    N = 4

    def __init__(self, exclude=None, name='bleu', output_names=None, label_names=None):
        super(BLEU, self).__init__(name=name, output_names=output_names, label_names=label_names)

        self._exclude = exclude or []

        self._match_counts = [0 for _ in range(self.N)]
        self._counts = [0 for _ in range(self.N)]

        self._size_ref = 0
        self._size_hyp = 0

    def update(self, labels, preds):
        labels, preds = mx.metric.check_label_shapes(labels, preds, True)

        new_labels = self._convert(labels)
        new_preds = self._convert(preds)

        for label, pred in zip(new_labels, new_preds):
            reference = [word for word in label if word not in self._exclude]
            hypothesis = [word for word in pred if word not in self._exclude]

            self._size_ref += len(reference)
            self._size_hyp += len(hypothesis)

            for n in range(self.N):
                reference_ngrams = self._get_ngrams(reference, n + 1)
                hypothesis_ngrams = self._get_ngrams(hypothesis, n + 1)

                match_count = 0

                for ngram in hypothesis_ngrams:
                    if ngram in reference_ngrams:
                        reference_ngrams.remove(ngram)

                        match_count += 1

                self._match_counts[n] += match_count
                self._counts[n] += len(hypothesis_ngrams)

    def get(self):
        precisions = [sys.float_info.min for n in range(self.N)]

        i = 1

        for n in range(self.N):
            match_counts = self._match_counts[n]
            counts = self._counts[n]

            if counts != 0:
                if match_counts == 0:
                    i *= 2
                    match_counts = 1 / i

                precisions[n] = match_counts / counts

        bleu = self._get_brevity_penalty() * math.exp(sum(map(math.log, precisions)) / self.N)

        return (self.name, bleu)

    def calculate(self):
        precisions = [sys.float_info.min for n in range(self.N)]

        i = 1

        for n in range(self.N):
            match_counts = self._match_counts[n]
            counts = self._counts[n]

            if counts != 0:
                if match_counts == 0:
                    i *= 2
                    match_counts = 1 / i

                precisions[n] = match_counts / counts

        return self._get_brevity_penalty() * math.exp(sum(map(math.log, precisions)) / self.N)

    def _get_brevity_penalty(self):
        if self._size_hyp >= self._size_ref:
            return 1
        else:
            return math.exp(1 - (self._size_ref / self._size_hyp))

    @staticmethod
    def _get_ngrams(sentence, n):
        ngrams = []

        if len(sentence) >= n:
            for i in range(len(sentence) - n + 1):
                ngrams.append(sentence[i:i+n])

        return ngrams

    @staticmethod
    def _convert(nd_list):
        if len(nd_list) == 0:
            return []

        new_list = [[] for _ in range(nd_list[0].shape[0])]

        for element in nd_list:
            for i in range(element.shape[0]):
                new_list[i].append(element[i].asscalar())

        return new_list
Christian Fuß's avatar
Christian Fuß committed
170

Sebastian Nickels's avatar
Sebastian Nickels committed
171
172


173
class CNNSupervisedTrainer_CifarClassifierNetwork:
Christian Fuß's avatar
Christian Fuß committed
174
    def applyBeamSearch(input, length, width, maxLength, currProb, netIndex, bestOutput):
175
        bestProb = 0.0
Christian Fuß's avatar
Christian Fuß committed
176
177
        while length < maxLength:
            length += 1
178
179
180
181
182
183
184
185
            batchIndex = 0
            for batchEntry in input:
                top_k_indices = mx.nd.topk(batchEntry, axis=0, k=width)
                top_k_values = mx.nd.topk(batchEntry, ret_typ='value', axis=0, k=width)
                for index in range(top_k_indices.size):

                    #print mx.nd.array(top_k_indices[index])
                    #print top_k_values[index]
Christian Fuß's avatar
Christian Fuß committed
186
                    if length == 1:
187
                        #print mx.nd.array(top_k_indices[index])
Christian Fuß's avatar
Christian Fuß committed
188
                        result = applyBeamSearch(self._networks[netIndex](mx.nd.array(top_k_indices[index])), length, width, maxLength,
189
190
                            currProb * top_k_values[index], netIndex, self._networks[netIndex](mx.nd.array(top_k_indices[index])))
                    else:
Christian Fuß's avatar
Christian Fuß committed
191
                        result = applyBeamSearch(self._networks[netIndex](mx.nd.array(top_k_indices[index])), length, width, maxLength,
192
193
                            currProb * top_k_values[index], netIndex, bestOutput)

Christian Fuß's avatar
Christian Fuß committed
194
                    if length == maxLength:
195
196
197
198
199
200
201
202
203
204
205
206
                        #print currProb
                        if currProb > bestProb:
                            bestProb = currProb
                            bestOutput[batchIndex] = result[batchIndex]
                            #print "new bestOutput: ", bestOutput

                batchIndex += 1
        #print bestOutput
        #print bestProb
        return bestOutput


207
    def __init__(self, data_loader, net_constructor):
Nicola Gatto's avatar
Nicola Gatto committed
208
209
        self._data_loader = data_loader
        self._net_creator = net_constructor
210
        self._networks = {}
Nicola Gatto's avatar
Nicola Gatto committed
211
212
213
214

    def train(self, batch_size=64,
              num_epoch=10,
              eval_metric='acc',
Sebastian Nickels's avatar
Sebastian Nickels committed
215
              eval_metric_params={},
Eyüp Harputlu's avatar
Eyüp Harputlu committed
216
217
              loss ='softmax_cross_entropy',
              loss_params={},
Nicola Gatto's avatar
Nicola Gatto committed
218
219
220
221
222
              optimizer='adam',
              optimizer_params=(('learning_rate', 0.001),),
              load_checkpoint=True,
              context='gpu',
              checkpoint_period=5,
223
              save_attention_image=False,
Nicola Gatto's avatar
Nicola Gatto committed
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
              normalize=True):
        if context == 'gpu':
            mx_context = mx.gpu()
        elif context == 'cpu':
            mx_context = mx.cpu()
        else:
            logging.error("Context argument is '" + context + "'. Only 'cpu' and 'gpu are valid arguments'.")

        if 'weight_decay' in optimizer_params:
            optimizer_params['wd'] = optimizer_params['weight_decay']
            del optimizer_params['weight_decay']
        if 'learning_rate_decay' in optimizer_params:
            min_learning_rate = 1e-08
            if 'learning_rate_minimum' in optimizer_params:
                min_learning_rate = optimizer_params['learning_rate_minimum']
                del optimizer_params['learning_rate_minimum']
            optimizer_params['lr_scheduler'] = mx.lr_scheduler.FactorScheduler(
                                                   optimizer_params['step_size'],
                                                   factor=optimizer_params['learning_rate_decay'],
                                                   stop_factor_lr=min_learning_rate)
            del optimizer_params['step_size']
            del optimizer_params['learning_rate_decay']


248
        train_iter, test_iter, data_mean, data_std, train_images, test_images = self._data_loader.load_data(batch_size)
249
250
251
252
253

        if normalize:
            self._net_creator.construct(context=mx_context, data_mean=data_mean, data_std=data_std)
        else:
            self._net_creator.construct(context=mx_context)
Nicola Gatto's avatar
Nicola Gatto committed
254
255
256
257
258
259
260
261

        begin_epoch = 0
        if load_checkpoint:
            begin_epoch = self._net_creator.load(mx_context)
        else:
            if os.path.isdir(self._net_creator._model_dir_):
                shutil.rmtree(self._net_creator._model_dir_)

262
        self._networks = self._net_creator.networks
Nicola Gatto's avatar
Nicola Gatto committed
263
264
265
266
267
268
269

        try:
            os.makedirs(self._net_creator._model_dir_)
        except OSError:
            if not os.path.isdir(self._net_creator._model_dir_):
                raise

Sebastian Nickels's avatar
Sebastian Nickels committed
270
        trainers = [mx.gluon.Trainer(network.collect_params(), optimizer, optimizer_params) for network in self._networks.values() if len(network.collect_params().values()) != 0]
Nicola Gatto's avatar
Nicola Gatto committed
271

Eyüp Harputlu's avatar
Eyüp Harputlu committed
272
273
        margin = loss_params['margin'] if 'margin' in loss_params else 1.0
        sparseLabel = loss_params['sparse_label'] if 'sparse_label' in loss_params else True
274
275
276
        #if loss == 'softmax_cross_entropy':
        #    fromLogits = loss_params['from_logits'] if 'from_logits' in loss_params else False
        #    loss_function = mx.gluon.loss.SoftmaxCrossEntropyLoss(from_logits=fromLogits, sparse_label=sparseLabel)
Eyüp Harputlu's avatar
Eyüp Harputlu committed
277
278
        if loss == 'softmax_cross_entropy':
            fromLogits = loss_params['from_logits'] if 'from_logits' in loss_params else False
279
280
            ignore_indices = [2]
            loss_function = SoftmaxCrossEntropyLossIgnoreIndices(ignore_indices=ignore_indices, from_logits=fromLogits, sparse_label=sparseLabel)
Eyüp Harputlu's avatar
Eyüp Harputlu committed
281
        elif loss == 'sigmoid_binary_cross_entropy':
Nicola Gatto's avatar
Nicola Gatto committed
282
            loss_function = mx.gluon.loss.SigmoidBinaryCrossEntropyLoss()
Eyüp Harputlu's avatar
Eyüp Harputlu committed
283
284
285
        elif loss == 'cross_entropy':
            loss_function = CrossEntropyLoss(sparse_label=sparseLabel)
        elif loss == 'l2':
Nicola Gatto's avatar
Nicola Gatto committed
286
            loss_function = mx.gluon.loss.L2Loss()
Eyüp Harputlu's avatar
Eyüp Harputlu committed
287
        elif loss == 'l1':
Nicola Gatto's avatar
Nicola Gatto committed
288
            loss_function = mx.gluon.loss.L2Loss()
Eyüp Harputlu's avatar
Eyüp Harputlu committed
289
290
291
292
293
294
295
296
297
298
299
300
301
        elif loss == 'huber':
            rho = loss_params['rho'] if 'rho' in loss_params else 1
            loss_function = mx.gluon.loss.HuberLoss(rho=rho)
        elif loss == 'hinge':
            loss_function = mx.gluon.loss.HingeLoss(margin=margin)
        elif loss == 'squared_hinge':
            loss_function = mx.gluon.loss.SquaredHingeLoss(margin=margin)
        elif loss == 'logistic':
            labelFormat = loss_params['label_format'] if 'label_format' in loss_params else 'signed'
            loss_function = mx.gluon.loss.LogisticLoss(label_format=labelFormat)
        elif loss == 'kullback_leibler':
            fromLogits = loss_params['from_logits'] if 'from_logits' in loss_params else True
            loss_function = mx.gluon.loss.KLDivLoss(from_logits=fromLogits)
Eyüp Harputlu's avatar
Eyüp Harputlu committed
302
303
        elif loss == 'log_cosh':
            loss_function = LogCoshLoss()
Eyüp Harputlu's avatar
Eyüp Harputlu committed
304
305
        else:
            logging.error("Invalid loss parameter.")
Nicola Gatto's avatar
Nicola Gatto committed
306
307
308
309
310
311
312

        speed_period = 50
        tic = None

        for epoch in range(begin_epoch, begin_epoch + num_epoch):
            train_iter.reset()
            for batch_i, batch in enumerate(train_iter):
313
                data_ = batch.data[0].as_in_context(mx_context)
314
                softmax_label = batch.label[0].as_in_context(mx_context)
315

Christian Fuß's avatar
Christian Fuß committed
316
317
                outputs=[]

Nicola Gatto's avatar
Nicola Gatto committed
318
                with autograd.record():
319
                    softmax_ = mx.nd.zeros((batch_size, 10,), ctx=mx_context)
320

321
                    lossList = []
322
                    softmax_ = self._networks[0](data_)
323
324
325
326
327
                    lossList.append(loss_function(softmax_, softmax_label))

                    loss = 0
                    for element in lossList:
                        loss = loss + element
328

Nicola Gatto's avatar
Nicola Gatto committed
329
                loss.backward()
330
331
332

                for trainer in trainers:
                    trainer.step(batch_size)
Nicola Gatto's avatar
Nicola Gatto committed
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349

                if tic is None:
                    tic = time.time()
                else:
                    if batch_i % speed_period == 0:
                        try:
                            speed = speed_period * batch_size / (time.time() - tic)
                        except ZeroDivisionError:
                            speed = float("inf")

                        logging.info("Epoch[%d] Batch[%d] Speed: %.2f samples/sec" % (epoch, batch_i, speed))

                        tic = time.time()

            tic = None

            train_iter.reset()
Sebastian Nickels's avatar
Sebastian Nickels committed
350
            metric = mx.metric.create(eval_metric, **eval_metric_params)
Nicola Gatto's avatar
Nicola Gatto committed
351
            for batch_i, batch in enumerate(train_iter):
352
                data_ = batch.data[0].as_in_context(mx_context)
353
354
355
356
357

                labels = [
                    batch.label[0].as_in_context(mx_context)
                ]

358
                outputs=[]
359

360
                if True: 
361
                    softmax_ = mx.nd.zeros((batch_size, 10,), ctx=mx_context)
362

363
                    attentionList=[]
364
                    softmax_ = self._networks[0](data_)
365
                    outputs.append(softmax_)
366

367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397

                    if save_attention_image == "True":
                        import matplotlib.pyplot as plt
                        logging.getLogger('matplotlib').setLevel(logging.ERROR)

                        plt.clf()
                        fig = plt.figure(figsize=(10,10))
                        max_length = len(labels)-1

                        if(os.path.isfile('src/test/resources/training_data/Show_attend_tell/dict.pkl')):
                            with open('src/test/resources/training_data/Show_attend_tell/dict.pkl', 'rb') as f:
                                dict = pickle.load(f)

                        for l in range(max_length):
                            attention = attentionList[l]
                            attention = mx.nd.slice_axis(attention, axis=0, begin=0, end=1)
                            attention = mx.nd.squeeze(attention)
                            attention_resized = np.resize(attention.asnumpy(), (8, 8))
                            ax = fig.add_subplot(max_length//3, max_length//4, l+1)
                            ax.set_title(dict[int(labels[l+1][0].asscalar())])
                            img = ax.imshow(train_images[0+batch_size*(batch_i)])
                            ax.imshow(attention_resized, cmap='gray', alpha=0.6, extent=img.get_extent())


                        plt.tight_layout()
                        target_dir = 'target/attention_images'
                        if not os.path.exists(target_dir):
                                    os.makedirs(target_dir)
                        plt.savefig(target_dir + '/attention_train.png')
                        plt.close()

398
                predictions = []
399
                for output_name in outputs:
Sebastian Nickels's avatar
Sebastian Nickels committed
400
                    if mx.nd.shape_array(mx.nd.squeeze(output_name)).size > 1:
401
402
403
404
                        predictions.append(mx.nd.argmax(output_name, axis=1))
                    #ArgMax already applied
                    else:
                        predictions.append(output_name)
405

406
                metric.update(preds=predictions, labels=labels)
Nicola Gatto's avatar
Nicola Gatto committed
407
408
409
            train_metric_score = metric.get()[1]

            test_iter.reset()
Sebastian Nickels's avatar
Sebastian Nickels committed
410
            metric = mx.metric.create(eval_metric, **eval_metric_params)
Nicola Gatto's avatar
Nicola Gatto committed
411
            for batch_i, batch in enumerate(test_iter):
412
                data_ = batch.data[0].as_in_context(mx_context)
413
414
415
416
417

                labels = [
                    batch.label[0].as_in_context(mx_context)
                ]

418
419
                outputs=[]

420
                if True: 
421
                    softmax_ = mx.nd.zeros((batch_size, 10,), ctx=mx_context)
422

423
                    attentionList=[]
424
                    softmax_ = self._networks[0](data_)
425
                    outputs.append(softmax_)
426

427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448

                    if save_attention_image == "True":
                        plt.clf()
                        fig = plt.figure(figsize=(10,10))
                        max_length = len(labels)-1

                        for l in range(max_length):
                            attention = attentionList[l]
                            attention = mx.nd.slice_axis(attention, axis=2, begin=0, end=1)
                            attention = mx.nd.slice_axis(attention, axis=0, begin=0, end=1)
                            attention = mx.nd.squeeze(attention)
                            attention_resized = np.resize(attention.asnumpy(), (8, 8))
                            ax = fig.add_subplot(max_length//3, max_length//4, l+1)
                            ax.set_title(dict[int(mx.nd.slice_axis(mx.nd.argmax(outputs[l+1], axis=1), axis=0, begin=0, end=1).asscalar())])
                            img = ax.imshow(test_images[0+batch_size*(batch_i)])
                            ax.imshow(attention_resized, cmap='gray', alpha=0.6, extent=img.get_extent())


                        plt.tight_layout()
                        plt.savefig(target_dir + '/attention_test.png')
                        plt.close()

449
                predictions = []
450
                for output_name in outputs:
Sebastian Nickels's avatar
Sebastian Nickels committed
451
                    if mx.nd.shape_array(mx.nd.squeeze(output_name)).size > 1:
452
453
454
455
                        predictions.append(mx.nd.argmax(output_name, axis=1))
                    #ArgMax already applied
                    else:
                        predictions.append(output_name)
456

457
                metric.update(preds=predictions, labels=labels)
Nicola Gatto's avatar
Nicola Gatto committed
458
459
460
461
            test_metric_score = metric.get()[1]

            logging.info("Epoch[%d] Train: %f, Test: %f" % (epoch, train_metric_score, test_metric_score))

462

Nicola Gatto's avatar
Nicola Gatto committed
463
            if (epoch - begin_epoch) % checkpoint_period == 0:
464
465
                for i, network in self._networks.items():
                    network.save_parameters(self.parameter_path(i) + '-' + str(epoch).zfill(4) + '.params')
Nicola Gatto's avatar
Nicola Gatto committed
466

467
468
469
        for i, network in self._networks.items():
            network.save_parameters(self.parameter_path(i) + '-' + str(num_epoch + begin_epoch).zfill(4) + '.params')
            network.export(self.parameter_path(i) + '_newest', epoch=0)
Nicola Gatto's avatar
Nicola Gatto committed
470

471
472
    def parameter_path(self, index):
        return self._net_creator._model_dir_ + self._net_creator._model_prefix_ + '_' + str(index)