{"cells":[{"metadata":{},"cell_type":"markdown","source":"# Overview\n\nThis notebook performs training more more than one model with TPU, performs inference for each of them and creates a baseline ensemble out of them.","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"!pip install -q efficientnet ","execution_count":null,"outputs":[]},{"metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","trusted":true},"cell_type":"code","source":"import os\nimport re\n\nimport numpy as np\nimport pandas as pd\nimport math, random\n\nfrom matplotlib import pyplot as plt\n\nfrom sklearn import metrics\nfrom sklearn.model_selection import train_test_split\n\nimport tensorflow as tf\nimport tensorflow.keras.layers as L\n\nimport tensorflow.keras.backend as K\n\nimport efficientnet.tfkeras as efn\n\nfrom kaggle_datasets import KaggleDatasets","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# TPU Strategy and other configs ","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# Detect hardware, return appropriate distribution strategy\ntry:\n    # TPU detection. No parameters necessary if TPU_NAME environment variable is\n    # set: this is always the case on Kaggle.\n    tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n    print('Running on TPU ', tpu.master())\nexcept ValueError:\n    tpu = None\n\nif tpu:\n    tf.config.experimental_connect_to_cluster(tpu)\n    tf.tpu.experimental.initialize_tpu_system(tpu)\n    strategy = tf.distribute.experimental.TPUStrategy(tpu)\nelse:\n    # Default distribution strategy in Tensorflow. Works on CPU and single GPU.\n    strategy = tf.distribute.get_strategy()\n\nprint(\"REPLICAS: \", strategy.num_replicas_in_sync)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# For tf.dataset\nAUTO = tf.data.experimental.AUTOTUNE\n\n# Data access\nGCS_PATH = KaggleDatasets().get_gcs_path('siim-isic-melanoma-classification')\n\n# Configuration\nEPOCHS = 10\nBATCH_SIZE = 8 * strategy.num_replicas_in_sync\nIMAGE_SIZE = [1024, 1024]","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Load Labels and Paths","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"def append_path(pre):\n    return np.vectorize(lambda file: os.path.join(GCS_DS_PATH, pre, file))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sub = pd.read_csv('/kaggle/input/siim-isic-melanoma-classification/sample_submission.csv')\nTRAINING_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/tfrecords/train*.tfrec')\nTEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/tfrecords/test*.tfrec')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(len(TRAINING_FILENAMES))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print(TRAINING_FILENAMES)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"VALIDATION_FILENAMES = TRAINING_FILENAMES[int(0.8*len(TRAINING_FILENAMES)):]\nTRAINING_FILENAMES = TRAINING_FILENAMES[:int(0.8*len(TRAINING_FILENAMES))]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(TRAINING_FILENAMES)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"len(VALIDATION_FILENAMES)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_mat(rotation, shear, height_zoom, width_zoom, height_shift, width_shift):\n    # returns 3x3 transformmatrix which transforms indicies\n        \n    # CONVERT DEGREES TO RADIANS\n    rotation = math.pi * rotation / 180.\n    shear = math.pi * shear / 180.\n    \n    # ROTATION MATRIX\n    c1 = tf.math.cos(rotation)\n    s1 = tf.math.sin(rotation)\n    one = tf.constant([1],dtype='float32')\n    zero = tf.constant([0],dtype='float32')\n    rotation_matrix = tf.reshape( tf.concat([c1,s1,zero, -s1,c1,zero, zero,zero,one],axis=0),[3,3] )\n        \n    # SHEAR MATRIX\n    c2 = tf.math.cos(shear)\n    s2 = tf.math.sin(shear)\n    shear_matrix = tf.reshape( tf.concat([one,s2,zero, zero,c2,zero, zero,zero,one],axis=0),[3,3] )    \n    \n    # ZOOM MATRIX\n    zoom_matrix = tf.reshape( tf.concat([one/height_zoom,zero,zero, zero,one/width_zoom,zero, zero,zero,one],axis=0),[3,3] )\n    \n    # SHIFT MATRIX\n    shift_matrix = tf.reshape( tf.concat([one,zero,height_shift, zero,one,width_shift, zero,zero,one],axis=0),[3,3] )\n    \n    return K.dot(K.dot(rotation_matrix, shear_matrix), K.dot(zoom_matrix, shift_matrix))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def transform(image,label):\n    # input image - is one image of size [dim,dim,3] not a batch of [b,dim,dim,3]\n    # output - image randomly rotated, sheared, zoomed, and shifted\n    DIM = IMAGE_SIZE[0]\n    XDIM = DIM%2 #fix for size 331\n    \n    rot = 15. * tf.random.normal([1],dtype='float32')\n    shr = 5. * tf.random.normal([1],dtype='float32') \n    h_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.\n    w_zoom = 1.0 + tf.random.normal([1],dtype='float32')/10.\n    h_shift = 16. * tf.random.normal([1],dtype='float32') \n    w_shift = 16. * tf.random.normal([1],dtype='float32') \n  \n    # GET TRANSFORMATION MATRIX\n    m = get_mat(rot,shr,h_zoom,w_zoom,h_shift,w_shift) \n\n    # LIST DESTINATION PIXEL INDICES\n    x = tf.repeat( tf.range(DIM//2,-DIM//2,-1), DIM )\n    y = tf.tile( tf.range(-DIM//2,DIM//2),[DIM] )\n    z = tf.ones([DIM*DIM],dtype='int32')\n    idx = tf.stack( [x,y,z] )\n    \n    # ROTATE DESTINATION PIXELS ONTO ORIGIN PIXELS\n    idx2 = K.dot(m,tf.cast(idx,dtype='float32'))\n    idx2 = K.cast(idx2,dtype='int32')\n    idx2 = K.clip(idx2,-DIM//2+XDIM+1,DIM//2)\n    \n    # FIND ORIGIN PIXEL VALUES           \n    idx3 = tf.stack( [DIM//2-idx2[0,], DIM//2-1+idx2[1,]] )\n    d = tf.gather_nd(image,tf.transpose(idx3))\n        \n    return tf.reshape(d,[DIM,DIM,3]),label","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def decode_image(image_data):\n    image = tf.image.decode_jpeg(image_data, channels=3)\n    image = tf.cast(image, tf.float32) / 255.0  # convert image to floats in [0, 1] range\n    image = tf.reshape(image, [*IMAGE_SIZE, 3]) # explicit size needed for TPU\n    return image\n\ndef read_labeled_tfrecord(example):\n    LABELED_TFREC_FORMAT = {\n        \"image\": tf.io.FixedLenFeature([], tf.string), # tf.string means bytestring\n        #\"class\": tf.io.FixedLenFeature([], tf.int64),  # shape [] means single element\n        \"target\": tf.io.FixedLenFeature([], tf.int64),  # shape [] means single element\n    }\n    example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)\n    image = decode_image(example['image'])\n    #label = tf.cast(example['class'], tf.int32)\n    label = tf.cast(example['target'], tf.int32)\n    return image, label # returns a dataset of (image, label) pairs\n\ndef read_unlabeled_tfrecord(example):\n    UNLABELED_TFREC_FORMAT = {\n        \"image\": tf.io.FixedLenFeature([], tf.string), # tf.string means bytestring\n        \"image_name\": tf.io.FixedLenFeature([], tf.string),  # shape [] means single element\n        # class is missing, this competitions's challenge is to predict flower classes for the test dataset\n    }\n    example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)\n    image = decode_image(example['image'])\n    idnum = example['image_name']\n    return image, idnum # returns a dataset of image(s)\n\ndef load_dataset(filenames, labeled=True, ordered=False):\n    # Read from TFRecords. For optimal performance, reading from multiple files at once and\n    # disregarding data order. Order does not matter since we will be shuffling the data anyway.\n\n    ignore_order = tf.data.Options()\n    if not ordered:\n        ignore_order.experimental_deterministic = False # disable order, increase speed\n\n    dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTO) # automatically interleaves reads from multiple files\n    dataset = dataset.with_options(ignore_order) # uses data as soon as it streams in, rather than in its original order\n    dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls=AUTO)\n    # returns a dataset of (image, label) pairs if labeled=True or (image, id) pairs if labeled=False\n    return dataset\n\ndef data_augment(image, label):\n    # data augmentation. Thanks to the dataset.prefetch(AUTO) statement in the next function (below),\n    # this happens essentially for free on TPU. Data pipeline code is executed on the \"CPU\" part\n    # of the TPU while the TPU itself is computing gradients.\n    image = tf.image.random_flip_left_right(image)\n    #image = tf.image.random_saturation(image, 0, 2)\n    return image, label   \n\ndef get_training_dataset():\n    dataset = load_dataset(TRAINING_FILENAMES, labeled=True)\n    dataset = dataset.map(data_augment, num_parallel_calls=AUTO)\n    dataset = dataset.map(transform, num_parallel_calls=AUTO)\n    dataset = dataset.repeat() # the training dataset must repeat for several epochs\n    dataset = dataset.shuffle(2048)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO) # prefetch next batch while training (autotune prefetch buffer size)\n    return dataset\n\ndef get_validation_dataset(ordered=False):\n    dataset = load_dataset(VALIDATION_FILENAMES, labeled=True, ordered=ordered)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.cache()\n    dataset = dataset.prefetch(AUTO) # prefetch next batch while training (autotune prefetch buffer size)\n    return dataset\n\ndef get_test_dataset(ordered=False):\n    dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO) # prefetch next batch while training (autotune prefetch buffer size)\n    return dataset\n\ndef count_data_items(filenames):\n    # the number of data items is written in the name of the .tfrec files, i.e. flowers00-230.tfrec = 230 data items\n    n = [int(re.compile(r\"-([0-9]*)\\.\").search(filename).group(1)) for filename in filenames]\n    return np.sum(n)\n\nNUM_TRAINING_IMAGES = count_data_items(TRAINING_FILENAMES)\nNUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)\nSTEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE\nprint('Dataset: {} training images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_TEST_IMAGES))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"LR_START = 0.00001\nLR_MAX = 0.00005 * strategy.num_replicas_in_sync\nLR_MIN = 0.00001\nLR_RAMPUP_EPOCHS = 5\nLR_SUSTAIN_EPOCHS = 0\nLR_EXP_DECAY = .8\n\ndef lrfn(epoch):\n    if epoch < LR_RAMPUP_EPOCHS:\n        lr = (LR_MAX - LR_START) / LR_RAMPUP_EPOCHS * epoch + LR_START\n    elif epoch < LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS:\n        lr = LR_MAX\n    else:\n        lr = (LR_MAX - LR_MIN) * LR_EXP_DECAY**(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS) + LR_MIN\n    return lr\n    \nlr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose=True)\n\nrng = [i for i in range(EPOCHS)]\ny = [lrfn(x) for x in rng]\nplt.plot(rng, y)\nprint(\"Learning rate schedule: {:.3g} to {:.3g} to {:.3g}\".format(y[0], max(y), y[-1]))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Models","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# with strategy.scope():\n#     model = tf.keras.Sequential([\n#         efn.EfficientNetB7(\n#             input_shape=(*IMAGE_SIZE, 3),\n#             weights='noisy-student',\n#             include_top=False\n#         ),\n#         L.GlobalAveragePooling2D(),\n#         L.Dense(1, activation='sigmoid')\n#     ])\n#     model.compile(\n#         optimizer='adam',\n#         loss = 'binary_crossentropy',\n#         metrics=['accuracy']\n#     )\n#     model.summary()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# model = tf.keras.models.load_model('/kaggle/input/siim-isic-enet-model-weights/efficientnetb7.h5')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"with strategy.scope():\n    model2 = tf.keras.Sequential([\n        efn.EfficientNetB7(\n            input_shape=(*IMAGE_SIZE, 3),\n            weights='imagenet',\n            include_top=False\n        ),\n        L.GlobalAveragePooling2D(),\n        L.Dense(1, activation='sigmoid')\n    ])\n    model2.compile(\n        optimizer='adam',\n        loss = 'binary_crossentropy',\n        metrics=['accuracy']\n    )\n    model2.summary()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Training","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"train_dataset = get_training_dataset()\nvalid_dataset = get_validation_dataset()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# history = model.fit(\n#     train_dataset, \n#     epochs=EPOCHS, \n#     callbacks=[lr_callback],\n#     steps_per_epoch=STEPS_PER_EPOCH,\n#     validation_data=valid_dataset\n# )","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"history2 = model2.fit(\n    train_dataset, \n    epochs=EPOCHS, \n    callbacks=[lr_callback],\n    steps_per_epoch=STEPS_PER_EPOCH,\n    validation_data=valid_dataset\n)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# model.save(\"efficientnetb7.h5\")","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model2.save(\"efficientnetb7.h5\")","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Training Evaluation","execution_count":null},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"def display_training_curves(training, validation, title, subplot):\n    \"\"\"\n    Source: https://www.kaggle.com/mgornergoogle/getting-started-with-100-flowers-on-tpu\n    \"\"\"\n    if subplot%10==1: # set up the subplots on the first call\n        plt.subplots(figsize=(10,10), facecolor='#F0F0F0')\n        plt.tight_layout()\n    ax = plt.subplot(subplot)\n    ax.set_facecolor('#F8F8F8')\n    ax.plot(training)\n    ax.plot(validation)\n    ax.set_title('model '+ title)\n    ax.set_ylabel(title)\n    #ax.set_ylim(0.28,1.05)\n    ax.set_xlabel('epoch')\n    ax.legend(['train', 'valid.'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_kg_hide-input":true},"cell_type":"code","source":"# display_training_curves(\n#     history.history['loss'], \n#     history.history['val_loss'], \n#     'model 1 loss', 211)\n# display_training_curves(\n#     history.history['accuracy'], \n#     history.history['val_accuracy'], \n#     'model 1 accuracy', 212)","execution_count":null,"outputs":[]},{"metadata":{"_kg_hide-input":true,"trusted":true},"cell_type":"code","source":"display_training_curves(\n    history2.history['loss'], \n    history2.history['val_loss'], \n    'model 2 loss', 211)\ndisplay_training_curves(\n    history2.history['accuracy'], \n    history2.history['val_accuracy'], \n    'model 2 accuracy', 212)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Prediction & Submission","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"# create copies for each model if you want to\n# sub1 = sub.copy()\nsub2 = sub.copy()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"%%time\n\ntest_ds = get_test_dataset(ordered=True)\n\nprint('Computing predictions...')\ntest_images_ds = test_ds.map(lambda image, idnum: image)\n# probabilities = model.predict(test_images_ds)\nprobabilities2 = model2.predict(test_images_ds)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print('Generating submission.csv file...')\ntest_ids_ds = test_ds.map(lambda image, idnum: idnum).unbatch()\ntest_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES))).numpy().astype('U') # all in one batch","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# pred_df = pd.DataFrame({'image_name': test_ids, 'target': np.concatenate(probabilities)})\n# pred_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"pred_df2 = pd.DataFrame({'image_name': test_ids, 'target': np.concatenate(probabilities2)})\npred_df2.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# del sub1['target']\n# sub1 = sub1.merge(pred_df, on='image_name')\n# sub1.to_csv('submission_efficientnetb7_noisystudent.csv', index=False)\n# sub1.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"del sub2['target']\nsub2 = sub2.merge(pred_df2, on='image_name')\nsub2.to_csv('submission_enetb7_imagenet.csv', index=False)\nsub2.head()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Ensemble","execution_count":null},{"metadata":{"trusted":true},"cell_type":"code","source":"sub1 = pd.read_csv('/kaggle/input/siim-isic-enet-model-weights/submission_efficientnetb7.csv')\nsub1.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sub1.to_csv('submission_enetb7_noisystudent.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sub_es = sub1[['image_name']]\nsub_es['target'] = 0.5*sub1['target'] + 0.5*sub2['target']\nsub_es.to_csv('submission.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# References\n\n- https://www.kaggle.com/mgornergoogle/getting-started-with-100-flowers-on-tpu\n- https://www.kaggle.com/xhlulu/alaska2-efficientnet-on-tpus \n- https://www.kaggle.com/yasufuminakama/siim-isic-efficientnetb3-inference","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"# Future Work & Tips\n\n- As validation data was not provided explicitly, we have to derive them from the training data. Currently a 80-20 split is performed with fixed start point. Next step is to impose K-fold cross-validation on the training set and ensemble each model's fold's predictions.","execution_count":null},{"metadata":{},"cell_type":"markdown","source":"## This notebook will be periodically updated with better performing models to reflect the progress in the competition. Augmentation techniques will be introduced in later versions, so stay tuned!","execution_count":null}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}