train_utils.py 18.3 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
# @author lucasmiranda42
# encoding: utf-8
# module deepof

"""

Simple utility functions used in deepof example scripts. These are not part of the main package

"""

11
from datetime import date, datetime
12
from kerastuner import BayesianOptimization, Hyperband
13
from kerastuner import HyperParameters
14
from kerastuner_tensorboard_logger import TensorBoardLogger
lucas_miranda's avatar
lucas_miranda committed
15
from sklearn.metrics import roc_auc_score
16
from tensorboard.plugins.hparams import api as hp
17
from typing import Tuple, Union, Any, List
18
19
20
21
22
23
24
import deepof.hypermodels
import deepof.model_utils
import numpy as np
import os
import pickle
import tensorflow as tf

25
26
27
28
29
# Ignore warning with no downstream effect
tf.get_logger().setLevel("ERROR")
tf.autograph.set_verbosity(0)


30
class CustomStopper(tf.keras.callbacks.EarlyStopping):
31
    """ Custom early stopping callback. Prevents the model from stopping before warmup is over """
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

    def __init__(self, start_epoch, *args, **kwargs):
        super(CustomStopper, self).__init__(*args, **kwargs)
        self.start_epoch = start_epoch

    def get_config(self):  # pragma: no cover
        """Updates callback metadata"""

        config = super().get_config().copy()
        config.update({"start_epoch": self.start_epoch})
        return config

    def on_epoch_end(self, epoch, logs=None):
        if epoch > self.start_epoch:
            super().on_epoch_end(epoch, logs)


49
50
51
52
53
54
55
def load_treatments(train_path):
    """Loads a dictionary containing the treatments per individual,
    to be loaded as metadata in the coordinates class"""
    try:
        with open(
            os.path.join(
                train_path,
56
                [i for i in os.listdir(train_path) if i.endswith(".pkl")][0],
57
58
59
60
61
62
63
64
65
66
67
            ),
            "rb",
        ) as handle:
            treatment_dict = pickle.load(handle)
    except IndexError:
        treatment_dict = None

    return treatment_dict


def get_callbacks(
lucas_miranda's avatar
lucas_miranda committed
68
69
70
    X_train: np.array,
    batch_size: int,
    variational: bool,
71
    phenotype_class: float,
lucas_miranda's avatar
lucas_miranda committed
72
73
    predictor: float,
    loss: str,
lucas_miranda's avatar
lucas_miranda committed
74
    X_val: np.array = None,
75
76
77
    cp: bool = False,
    reg_cat_clusters: bool = False,
    reg_cluster_variance: bool = False,
lucas_miranda's avatar
lucas_miranda committed
78
79
    knn_samples: int = 10000,
    knn_neighbors: int = 100,
80
    logparam: dict = None,
81
    outpath: str = ".",
82
) -> List[Union[Any]]:
83
    """Generates callbacks for model training, including:
84
85
86
87
    - run_ID: run name, with coarse parameter details;
    - tensorboard_callback: for real-time visualization;
    - cp_callback: for checkpoint saving,
    - onecycle: for learning rate scheduling"""
88

89
90
91
92
93
94
95
96
    latreg = "none"
    if reg_cat_clusters and not reg_cluster_variance:
        latreg = "categorical"
    elif reg_cluster_variance and not reg_cat_clusters:
        latreg = "variance"
    elif reg_cat_clusters and reg_cluster_variance:
        latreg = "categorical+variance"

97
    run_ID = "{}{}{}{}{}{}{}_{}".format(
98
        ("GMVAE" if variational else "AE"),
lucas_miranda's avatar
lucas_miranda committed
99
        ("_Pred={}".format(predictor) if predictor > 0 and variational else ""),
100
        ("_Pheno={}".format(phenotype_class) if phenotype_class > 0 else ""),
101
        ("_loss={}".format(loss) if variational else ""),
102
103
        ("_encoding={}".format(logparam["encoding"]) if logparam is not None else ""),
        ("_k={}".format(logparam["k"]) if logparam is not None else ""),
104
        ("_latreg={}".format(latreg)),
105
        (datetime.now().strftime("%Y%m%d-%H%M%S")),
106
107
    )

108
    log_dir = os.path.abspath(os.path.join(outpath, "fit", run_ID))
109
    tensorboard_callback = tf.keras.callbacks.TensorBoard(
110
111
112
        log_dir=log_dir,
        histogram_freq=1,
        profile_batch=2,
113
114
    )

lucas_miranda's avatar
lucas_miranda committed
115
116
117
    knn = deepof.model_utils.knn_cluster_purity(
        k=knn_neighbors,
        samples=knn_samples,
lucas_miranda's avatar
lucas_miranda committed
118
        validation_data=X_val,
119
        log_dir=os.path.join(outpath, "metrics"),
lucas_miranda's avatar
lucas_miranda committed
120
121
    )

122
    onecycle = deepof.model_utils.one_cycle_scheduler(
123
124
        X_train.shape[0] // batch_size * 250,
        max_rate=0.005,
lucas_miranda's avatar
lucas_miranda committed
125
        log_dir=os.path.join(outpath, "metrics"),
126
127
    )

lucas_miranda's avatar
lucas_miranda committed
128
    callbacks = [run_ID, tensorboard_callback, knn, onecycle]
129
130
131

    if cp:
        cp_callback = tf.keras.callbacks.ModelCheckpoint(
132
            os.path.join(outpath, "checkpoints", run_ID + "/cp-{epoch:04d}.ckpt"),
133
134
135
136
137
138
139
140
            verbose=1,
            save_best_only=False,
            save_weights_only=True,
            save_freq="epoch",
        )
        callbacks.append(cp_callback)

    return callbacks
141
142


lucas_miranda's avatar
lucas_miranda committed
143
def log_hyperparameters(phenotype_class: float, rec: str):
lucas_miranda's avatar
lucas_miranda committed
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
    """Blueprint for hyperparameter and metric logging in tensorboard during hyperparameter tuning"""

    logparams = [
        hp.HParam(
            "encoding",
            hp.Discrete([2, 4, 6, 8, 12, 16]),
            display_name="encoding",
            description="encoding size dimensionality",
        ),
        hp.HParam(
            "k",
            hp.IntInterval(min_value=1, max_value=25),
            display_name="k",
            description="cluster_number",
        ),
        hp.HParam(
            "loss",
            hp.Discrete(["ELBO", "MMD", "ELBO+MMD"]),
            display_name="loss function",
            description="loss function",
        ),
    ]

    metrics = [
        hp.Metric("val_{}mae".format(rec), display_name="val_{}mae".format(rec)),
        hp.Metric("val_{}mse".format(rec), display_name="val_{}mse".format(rec)),
    ]
    if phenotype_class:
        logparams.append(
            hp.HParam(
                "pheno_weight",
                hp.RealInterval(min_value=0.0, max_value=1000.0),
                display_name="pheno weight",
                description="weight applied to phenotypic classifier from the latent space",
            )
        )
        metrics += [
            hp.Metric(
                "phenotype_prediction_accuracy",
                display_name="phenotype_prediction_accuracy",
            ),
            hp.Metric(
                "phenotype_prediction_auc",
                display_name="phenotype_prediction_auc",
            ),
        ]

    return logparams, metrics


# noinspection PyUnboundLocalVariable
lucas_miranda's avatar
lucas_miranda committed
195
196
197
198
199
200
201
202
203
204
def tensorboard_metric_logging(
    run_dir: str,
    hpms: Any,
    ae: Any,
    X_val: np.ndarray,
    y_val: np.ndarray,
    phenotype_class: float,
    predictor: float,
    rec: str,
):
lucas_miranda's avatar
lucas_miranda committed
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
    """Autoencoder metric logging in tensorboard"""

    output = ae.predict(X_val)
    if phenotype_class or predictor:
        reconstruction = output[0]
        prediction = output[1]
        pheno = output[-1]
    else:
        reconstruction = output

    with tf.summary.create_file_writer(run_dir).as_default():
        hp.hparams(hpms)  # record the values used in this trial
        val_mae = tf.reduce_mean(
            tf.keras.metrics.mean_absolute_error(X_val, reconstruction)
        )
        val_mse = tf.reduce_mean(
            tf.keras.metrics.mean_squared_error(X_val, reconstruction)
        )
        tf.summary.scalar("val_{}mae".format(rec), val_mae, step=1)
        tf.summary.scalar("val_{}mse".format(rec), val_mse, step=1)

        if predictor:
            pred_mae = tf.reduce_mean(
                tf.keras.metrics.mean_absolute_error(X_val, prediction)
            )
            pred_mse = tf.reduce_mean(
                tf.keras.metrics.mean_squared_error(X_val, prediction)
            )
            tf.summary.scalar("val_prediction_mae".format(rec), pred_mae, step=1)
            tf.summary.scalar("val_prediction_mse".format(rec), pred_mse, step=1)

        if phenotype_class:
            pheno_acc = tf.keras.metrics.binary_accuracy(y_val, tf.squeeze(pheno))
            pheno_auc = roc_auc_score(y_val, pheno)

            tf.summary.scalar("phenotype_prediction_accuracy", pheno_acc, step=1)
            tf.summary.scalar("phenotype_prediction_auc", pheno_auc, step=1)


244
def autoencoder_fitting(
245
    preprocessed_object: Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray],
246
247
    batch_size: int,
    encoding_size: int,
248
    epochs: int,
249
250
251
252
253
    hparams: dict,
    kl_warmup: int,
    log_history: bool,
    log_hparams: bool,
    loss: str,
254
255
256
257
258
    mmd_warmup: int,
    montecarlo_kl: int,
    n_components: int,
    output_path: str,
    phenotype_class: float,
259
260
261
    predictor: float,
    pretrained: str,
    save_checkpoints: bool,
262
    save_weights: bool,
263
    variational: bool,
264
265
    reg_cat_clusters: bool,
    reg_cluster_variance: bool,
lucas_miranda's avatar
lucas_miranda committed
266
267
    knn_neighbors: int,
    knn_samples: int,
268
):
269
270
    """Implementation function for deepof.data.coordinates.deep_unsupervised_embedding"""

271
    # Load data
272
273
274
275
276
    X_train, y_train, X_val, y_val = preprocessed_object

    # To avoid stability issues
    tf.keras.backend.clear_session()

277
    # Defines what to log on tensorboard (useful for trying out different models)
278
279
    logparam = {
        "encoding": encoding_size,
280
        "k": n_components,
281
282
283
284
285
        "loss": loss,
    }
    if phenotype_class:
        logparam["pheno_weight"] = phenotype_class

286
    # Load callbacks
287
    run_ID, *cbacks = get_callbacks(
288
        X_train=X_train,
lucas_miranda's avatar
lucas_miranda committed
289
        X_val=(X_val if X_val.shape != (0,) else None),
290
291
292
        batch_size=batch_size,
        cp=save_checkpoints,
        variational=variational,
293
        phenotype_class=phenotype_class,
294
295
        predictor=predictor,
        loss=loss,
lucas_miranda's avatar
lucas_miranda committed
296
297
        knn_neighbors=knn_neighbors,
        knn_samples=knn_samples,
298
        reg_cat_clusters=reg_cat_clusters,
299
        reg_cluster_variance=reg_cluster_variance,
300
301
302
        logparam=logparam,
        outpath=output_path,
    )
303
304
    if not log_history:
        cbacks = cbacks[1:]
305

306
    # Logs hyperparameters to tensorboard
lucas_miranda's avatar
lucas_miranda committed
307
    rec = "reconstruction_" if phenotype_class else ""
308
    if log_hparams:
lucas_miranda's avatar
lucas_miranda committed
309
        logparams, metrics = log_hyperparameters(phenotype_class, rec)
310
311
312
313
314
315
316
317

        with tf.summary.create_file_writer(
            os.path.join(output_path, "hparams", run_ID)
        ).as_default():
            hp.hparams_config(
                hparams=logparams,
                metrics=metrics,
            )
318

319
    # Build models
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
    if not variational:
        encoder, decoder, ae = deepof.models.SEQ_2_SEQ_AE(
            ({} if hparams is None else hparams)
        ).build(X_train.shape)
        return_list = (encoder, decoder, ae)

    else:
        (
            encoder,
            generator,
            grouper,
            ae,
            kl_warmup_callback,
            mmd_warmup_callback,
        ) = deepof.models.SEQ_2_SEQ_GMVAE(
            architecture_hparams=({} if hparams is None else hparams),
            batch_size=batch_size,
            compile_model=True,
            encoding=encoding_size,
            kl_warmup_epochs=kl_warmup,
            loss=loss,
            mmd_warmup_epochs=mmd_warmup,
            montecarlo_kl=montecarlo_kl,
            neuron_control=False,
            number_of_components=n_components,
            overlap_loss=False,
            phenotype_prediction=phenotype_class,
            predictor=predictor,
348
349
            reg_cat_clusters=reg_cat_clusters,
            reg_cluster_variance=reg_cluster_variance,
350
351
352
353
354
355
        ).build(
            X_train.shape
        )
        return_list = (encoder, generator, grouper, ae)

    if pretrained:
356
        # If pretrained models are specified, load weights and return
357
358
359
360
361
362
363
364
365
        ae.load_weights(pretrained)
        return return_list

    else:
        if not variational:

            ae.fit(
                x=X_train,
                y=X_train,
366
                epochs=epochs,
367
368
369
                batch_size=batch_size,
                verbose=1,
                validation_data=(X_val, X_val),
370
371
                callbacks=cbacks
                + [
372
373
374
375
                    CustomStopper(
                        monitor="val_loss",
                        patience=5,
                        restore_best_weights=True,
376
                        start_epoch=max(kl_warmup, mmd_warmup),
377
378
379
380
                    ),
                ],
            )

381
382
383
            if save_weights:
                ae.save_weights("{}_final_weights.h5".format(run_ID))

384
385
        else:

386
            callbacks_ = cbacks + [
387
388
389
390
                CustomStopper(
                    monitor="val_loss",
                    patience=5,
                    restore_best_weights=True,
391
                    start_epoch=max(kl_warmup, mmd_warmup),
392
393
394
                ),
            ]

395
            if "ELBO" in loss and kl_warmup > 0:
396
397
                # noinspection PyUnboundLocalVariable
                callbacks_.append(kl_warmup_callback)
398
            if "MMD" in loss and mmd_warmup > 0:
399
400
401
                # noinspection PyUnboundLocalVariable
                callbacks_.append(mmd_warmup_callback)

402
403
404
405
406
407
408
            Xs, ys = [X_train], [X_train]
            Xvals, yvals = [X_val], [X_val]

            if predictor > 0.0:
                Xs, ys = X_train[:-1], [X_train[:-1], X_train[1:]]
                Xvals, yvals = X_val[:-1], [X_val[:-1], X_val[1:]]

409
            if phenotype_class > 0.0:
410
411
412
                ys += [y_train]
                yvals += [y_val]

413
            ae.fit(
414
415
                x=Xs,
                y=ys,
416
                epochs=epochs,
417
418
419
420
421
422
423
424
425
                batch_size=batch_size,
                verbose=1,
                validation_data=(
                    Xvals,
                    yvals,
                ),
                callbacks=callbacks_,
            )

426
            if not os.path.exists(os.path.join(output_path, "trained_weights")):
427
428
                os.makedirs("trained_weights")

429
            if save_weights:
430
431
                ae.save_weights(
                    os.path.join(
432
433
434
                        "{}".format(output_path),
                        "trained_weights",
                        "{}_final_weights.h5".format(run_ID),
435
436
                    )
                )
437

438
439
440
441
442
            if log_hparams:
                # Logparams to tensorboard
                tensorboard_metric_logging(
                    os.path.join(output_path, "hparams", run_ID),
                    logparam,
lucas_miranda's avatar
lucas_miranda committed
443
                    ae,
lucas_miranda's avatar
lucas_miranda committed
444
445
                    Xvals,
                    yvals[-1],
lucas_miranda's avatar
lucas_miranda committed
446
447
448
                    phenotype_class,
                    predictor,
                    rec,
449
                )
450

451
452
453
    return return_list


454
def tune_search(
455
    data: List[np.array],
456
    encoding_size: int,
457
458
    hypertun_trials: int,
    hpt_type: str,
459
460
    hypermodel: str,
    k: int,
461
    kl_warmup_epochs: int,
462
    loss: str,
463
    mmd_warmup_epochs: int,
464
    overlap_loss: float,
465
    phenotype_class: float,
466
467
    predictor: float,
    project_name: str,
468
    callbacks: List,
469
    n_epochs: int = 30,
470
    n_replicas: int = 1,
471
    outpath: str = ".",
472
) -> Union[bool, Tuple[Any, Any]]:
473
474
    """Define the search space using keras-tuner and bayesian optimization

475
476
477
478
479
480
481
482
483
484
485
    Parameters:
        - train (np.array): dataset to train the model on
        - test (np.array): dataset to validate the model on
        - hypertun_trials (int): number of Bayesian optimization iterations to run
        - hpt_type (str): specify one of Bayesian Optimization (bayopt) and Hyperband (hyperband)
        - hypermodel (str): hypermodel to load. Must be one of S2SAE (plain autoencoder)
        or S2SGMVAE (Gaussian Mixture Variational autoencoder).
        - k (int) number of components of the Gaussian Mixture
        - loss (str): one of [ELBO, MMD, ELBO+MMD]
        - overlap_loss (float): assigns as weight to an extra loss term which
        penalizes overlap between GM components
486
        - phenotype_class (float): adds an extra regularizing neural network to the model,
487
488
489
490
491
492
493
494
495
496
497
498
        which tries to predict the phenotype of the animal from which the sequence comes
        - predictor (float): adds an extra regularizing neural network to the model,
        which tries to predict the next frame from the current one
        - project_name (str): ID of the current run
        - callbacks (list): list of callbacks for the training loop
        - n_epochs (int): optional. Number of epochs to train each run for
        - n_replicas (int): optional. Number of replicas per parameter set. Higher values
         will yield more robust results, but will affect performance severely

    Returns:
        - best_hparams (dict): dictionary with the best retrieved hyperparameters
        - best_run (tf.keras.Model): trained instance of the best model found
499
500
501

    """

502
503
    X_train, y_train, X_val, y_val = data

504
505
506
507
    assert hpt_type in ["bayopt", "hyperband"], (
        "Invalid hyperparameter tuning framework. " "Select one of bayopt and hyperband"
    )

lucas_miranda's avatar
lucas_miranda committed
508
    if hypermodel == "S2SAE":  # pragma: no cover
509
        assert (
510
            predictor == 0.0 and phenotype_class == 0.0
511
        ), "Prediction branches are only available for variational models. See documentation for more details"
512
        hypermodel = deepof.hypermodels.SEQ_2_SEQ_AE(input_shape=X_train.shape)
513
514
515

    elif hypermodel == "S2SGMVAE":
        hypermodel = deepof.hypermodels.SEQ_2_SEQ_GMVAE(
516
            input_shape=X_train.shape,
517
            encoding=encoding_size,
518
            kl_warmup_epochs=kl_warmup_epochs,
519
            loss=loss,
520
            mmd_warmup_epochs=mmd_warmup_epochs,
521
            number_of_components=k,
522
            overlap_loss=overlap_loss,
523
            phenotype_predictor=phenotype_class,
524
            predictor=predictor,
525
        )
lucas_miranda's avatar
lucas_miranda committed
526

527
528
529
    else:
        return False

530
531
532
    hpt_params = {
        "hypermodel": hypermodel,
        "executions_per_trial": n_replicas,
533
534
535
        "logger": TensorBoardLogger(
            metrics=["val_mae"], logdir=os.path.join(outpath, "logged_hparams")
        ),
536
537
538
539
540
541
542
543
        "objective": "val_mae",
        "project_name": project_name,
        "seed": 42,
        "tune_new_entries": True,
    }

    if hpt_type == "hyperband":
        tuner = Hyperband(
544
545
546
            directory=os.path.join(
                outpath, "HyperBandx_{}_{}".format(loss, str(date.today()))
            ),
547
548
            max_epochs=35,
            hyperband_iterations=hypertun_trials,
549
            factor=3,
550
551
552
553
            **hpt_params
        )
    else:
        tuner = BayesianOptimization(
554
555
556
            directory=os.path.join(
                outpath, "BayOpt_{}_{}".format(loss, str(date.today()))
            ),
557
558
559
            max_trials=hypertun_trials,
            **hpt_params
        )
560
561
562

    print(tuner.search_space_summary())

563
564
565
566
567
568
569
    Xs, ys = [X_train], [X_train]
    Xvals, yvals = [X_val], [X_val]

    if predictor > 0.0:
        Xs, ys = X_train[:-1], [X_train[:-1], X_train[1:]]
        Xvals, yvals = X_val[:-1], [X_val[:-1], X_val[1:]]

570
    if phenotype_class > 0.0:
571
572
573
        ys += [y_train]
        yvals += [y_val]

574
    tuner.search(
575
576
        Xs,
        ys,
577
        epochs=n_epochs,
578
        validation_data=(Xvals, yvals),
579
        verbose=1,
580
        batch_size=64,
lucas_miranda's avatar
lucas_miranda committed
581
        callbacks=callbacks,
582
583
584
585
586
    )

    best_hparams = tuner.get_best_hyperparameters(num_trials=1)[0]
    best_run = tuner.hypermodel.build(best_hparams)

lucas_miranda's avatar
lucas_miranda committed
587
588
    print(tuner.results_summary())

589
    return best_hparams, best_run