train_utils.py 18.4 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"),
120
        variational=variational,
lucas_miranda's avatar
lucas_miranda committed
121
122
    )

123
    onecycle = deepof.model_utils.one_cycle_scheduler(
124
125
        X_train.shape[0] // batch_size * 250,
        max_rate=0.005,
lucas_miranda's avatar
lucas_miranda committed
126
        log_dir=os.path.join(outpath, "metrics"),
127
128
    )

lucas_miranda's avatar
lucas_miranda committed
129
    callbacks = [run_ID, tensorboard_callback, knn, onecycle]
130
131
132

    if cp:
        cp_callback = tf.keras.callbacks.ModelCheckpoint(
133
            os.path.join(outpath, "checkpoints", run_ID + "/cp-{epoch:04d}.ckpt"),
134
135
136
137
138
139
140
141
            verbose=1,
            save_best_only=False,
            save_weights_only=True,
            save_freq="epoch",
        )
        callbacks.append(cp_callback)

    return callbacks
142
143


lucas_miranda's avatar
lucas_miranda committed
144
def log_hyperparameters(phenotype_class: float, rec: str):
lucas_miranda's avatar
lucas_miranda committed
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
    """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
196
197
198
199
200
201
202
203
204
205
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
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
244
    """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)


245
def autoencoder_fitting(
246
    preprocessed_object: Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray],
247
248
    batch_size: int,
    encoding_size: int,
249
    epochs: int,
250
251
252
253
254
    hparams: dict,
    kl_warmup: int,
    log_history: bool,
    log_hparams: bool,
    loss: str,
255
256
257
258
259
    mmd_warmup: int,
    montecarlo_kl: int,
    n_components: int,
    output_path: str,
    phenotype_class: float,
260
261
262
    predictor: float,
    pretrained: str,
    save_checkpoints: bool,
263
    save_weights: bool,
264
    variational: bool,
265
266
    reg_cat_clusters: bool,
    reg_cluster_variance: bool,
lucas_miranda's avatar
lucas_miranda committed
267
268
    knn_neighbors: int,
    knn_samples: int,
269
):
270
271
    """Implementation function for deepof.data.coordinates.deep_unsupervised_embedding"""

272
    # Load data
273
274
275
276
277
    X_train, y_train, X_val, y_val = preprocessed_object

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

278
    # Defines what to log on tensorboard (useful for trying out different models)
279
280
    logparam = {
        "encoding": encoding_size,
281
        "k": n_components,
282
283
284
285
286
        "loss": loss,
    }
    if phenotype_class:
        logparam["pheno_weight"] = phenotype_class

287
    # Load callbacks
288
    run_ID, *cbacks = get_callbacks(
289
        X_train=X_train,
lucas_miranda's avatar
lucas_miranda committed
290
        X_val=(X_val if X_val.shape != (0,) else None),
291
292
293
        batch_size=batch_size,
        cp=save_checkpoints,
        variational=variational,
294
        phenotype_class=phenotype_class,
295
296
        predictor=predictor,
        loss=loss,
lucas_miranda's avatar
lucas_miranda committed
297
298
        knn_neighbors=knn_neighbors,
        knn_samples=knn_samples,
299
        reg_cat_clusters=reg_cat_clusters,
300
        reg_cluster_variance=reg_cluster_variance,
301
302
303
        logparam=logparam,
        outpath=output_path,
    )
304
305
    if not log_history:
        cbacks = cbacks[1:]
306

307
    # Logs hyperparameters to tensorboard
lucas_miranda's avatar
lucas_miranda committed
308
    rec = "reconstruction_" if phenotype_class else ""
309
    if log_hparams:
lucas_miranda's avatar
lucas_miranda committed
310
        logparams, metrics = log_hyperparameters(phenotype_class, rec)
311
312
313
314
315
316
317
318

        with tf.summary.create_file_writer(
            os.path.join(output_path, "hparams", run_ID)
        ).as_default():
            hp.hparams_config(
                hparams=logparams,
                metrics=metrics,
            )
319

320
    # Build models
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
348
    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,
349
350
            reg_cat_clusters=reg_cat_clusters,
            reg_cluster_variance=reg_cluster_variance,
351
352
353
354
355
356
        ).build(
            X_train.shape
        )
        return_list = (encoder, generator, grouper, ae)

    if pretrained:
357
        # If pretrained models are specified, load weights and return
358
359
360
361
362
363
364
365
366
        ae.load_weights(pretrained)
        return return_list

    else:
        if not variational:

            ae.fit(
                x=X_train,
                y=X_train,
367
                epochs=epochs,
368
369
370
                batch_size=batch_size,
                verbose=1,
                validation_data=(X_val, X_val),
371
372
                callbacks=cbacks
                + [
373
374
375
376
                    CustomStopper(
                        monitor="val_loss",
                        patience=5,
                        restore_best_weights=True,
377
                        start_epoch=max(kl_warmup, mmd_warmup),
378
379
380
381
                    ),
                ],
            )

382
383
384
            if save_weights:
                ae.save_weights("{}_final_weights.h5".format(run_ID))

385
386
        else:

387
            callbacks_ = cbacks + [
388
389
390
391
                CustomStopper(
                    monitor="val_loss",
                    patience=5,
                    restore_best_weights=True,
392
                    start_epoch=max(kl_warmup, mmd_warmup),
393
394
395
                ),
            ]

396
            if "ELBO" in loss and kl_warmup > 0:
397
398
                # noinspection PyUnboundLocalVariable
                callbacks_.append(kl_warmup_callback)
399
            if "MMD" in loss and mmd_warmup > 0:
400
401
402
                # noinspection PyUnboundLocalVariable
                callbacks_.append(mmd_warmup_callback)

403
404
405
406
407
408
409
            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:]]

410
            if phenotype_class > 0.0:
411
412
413
                ys += [y_train]
                yvals += [y_val]

414
            ae.fit(
415
416
                x=Xs,
                y=ys,
417
                epochs=epochs,
418
419
420
421
422
423
424
425
426
                batch_size=batch_size,
                verbose=1,
                validation_data=(
                    Xvals,
                    yvals,
                ),
                callbacks=callbacks_,
            )

427
            if not os.path.exists(os.path.join(output_path, "trained_weights")):
428
429
                os.makedirs("trained_weights")

430
            if save_weights:
431
432
                ae.save_weights(
                    os.path.join(
433
434
435
                        "{}".format(output_path),
                        "trained_weights",
                        "{}_final_weights.h5".format(run_ID),
436
437
                    )
                )
438

439
440
441
442
443
            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
444
                    ae,
lucas_miranda's avatar
lucas_miranda committed
445
446
                    Xvals,
                    yvals[-1],
lucas_miranda's avatar
lucas_miranda committed
447
448
449
                    phenotype_class,
                    predictor,
                    rec,
450
                )
451

452
453
454
    return return_list


455
def tune_search(
456
    data: List[np.array],
457
    encoding_size: int,
458
459
    hypertun_trials: int,
    hpt_type: str,
460
461
    hypermodel: str,
    k: int,
462
    kl_warmup_epochs: int,
463
    loss: str,
464
    mmd_warmup_epochs: int,
465
    overlap_loss: float,
466
    phenotype_class: float,
467
468
    predictor: float,
    project_name: str,
469
    callbacks: List,
470
    n_epochs: int = 30,
471
    n_replicas: int = 1,
472
    outpath: str = ".",
473
) -> Union[bool, Tuple[Any, Any]]:
474
475
    """Define the search space using keras-tuner and bayesian optimization

476
477
478
479
480
481
482
483
484
485
486
    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
487
        - phenotype_class (float): adds an extra regularizing neural network to the model,
488
489
490
491
492
493
494
495
496
497
498
499
        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
500
501
502

    """

503
504
    X_train, y_train, X_val, y_val = data

505
506
507
508
    assert hpt_type in ["bayopt", "hyperband"], (
        "Invalid hyperparameter tuning framework. " "Select one of bayopt and hyperband"
    )

lucas_miranda's avatar
lucas_miranda committed
509
    if hypermodel == "S2SAE":  # pragma: no cover
510
        assert (
511
            predictor == 0.0 and phenotype_class == 0.0
512
        ), "Prediction branches are only available for variational models. See documentation for more details"
513
        hypermodel = deepof.hypermodels.SEQ_2_SEQ_AE(input_shape=X_train.shape)
514
515
516

    elif hypermodel == "S2SGMVAE":
        hypermodel = deepof.hypermodels.SEQ_2_SEQ_GMVAE(
517
            input_shape=X_train.shape,
518
            encoding=encoding_size,
519
            kl_warmup_epochs=kl_warmup_epochs,
520
            loss=loss,
521
            mmd_warmup_epochs=mmd_warmup_epochs,
522
            number_of_components=k,
523
            overlap_loss=overlap_loss,
524
            phenotype_predictor=phenotype_class,
525
            predictor=predictor,
526
        )
lucas_miranda's avatar
lucas_miranda committed
527

528
529
530
    else:
        return False

531
532
533
    hpt_params = {
        "hypermodel": hypermodel,
        "executions_per_trial": n_replicas,
534
535
536
        "logger": TensorBoardLogger(
            metrics=["val_mae"], logdir=os.path.join(outpath, "logged_hparams")
        ),
537
538
539
540
541
542
543
544
        "objective": "val_mae",
        "project_name": project_name,
        "seed": 42,
        "tune_new_entries": True,
    }

    if hpt_type == "hyperband":
        tuner = Hyperband(
545
546
547
            directory=os.path.join(
                outpath, "HyperBandx_{}_{}".format(loss, str(date.today()))
            ),
548
549
            max_epochs=35,
            hyperband_iterations=hypertun_trials,
550
            factor=3,
551
552
553
554
            **hpt_params
        )
    else:
        tuner = BayesianOptimization(
555
556
557
            directory=os.path.join(
                outpath, "BayOpt_{}_{}".format(loss, str(date.today()))
            ),
558
559
560
            max_trials=hypertun_trials,
            **hpt_params
        )
561
562
563

    print(tuner.search_space_summary())

564
565
566
567
568
569
570
    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:]]

571
    if phenotype_class > 0.0:
572
573
574
        ys += [y_train]
        yvals += [y_val]

575
    tuner.search(
576
577
        Xs,
        ys,
578
        epochs=n_epochs,
579
        validation_data=(Xvals, yvals),
580
        verbose=1,
581
        batch_size=64,
lucas_miranda's avatar
lucas_miranda committed
582
        callbacks=callbacks,
583
584
585
586
587
    )

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

lucas_miranda's avatar
lucas_miranda committed
588
589
    print(tuner.results_summary())

590
    return best_hparams, best_run