{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# -*- coding: utf-8 -*-\n\"\"\"Untitled0.ipynb\n\nAutomatically generated by Colaboratory.\n\nOriginal file is located at\n    https://colab.research.google.com/drive/1yQ-wrRfeQwg7IcdmApJdmASptqeL1Qx0\n\"\"\"\n\n# from google.colab import drive\n# drive.mount('/content/gdrive')\n\n# Commented out IPython magic to ensure Python compatibility.\n# %tensorflow_version 1.x\n\nimport tensorflow\ntensorflow.__version__\n\nimport tensorflow.keras as keras\nkeras.__version__\n\n# !ls '/content/gdrive/My Drive/Datasets/kannada-mnist'\n\n# !cp -r '/content/gdrive/My Drive/Datasets/kannada-mnist' ./\n\n# !ls ./kannada-mnist/\n\nimport pandas as pd\nfrom keras.layers import *\nfrom keras.models import *\nfrom keras.callbacks import CSVLogger, ModelCheckpoint\nfrom keras import losses\nfrom keras import optimizers\nfrom sklearn.model_selection import train_test_split\n\ntrain = pd.read_csv('../input/Kannada-MNIST/train.csv')\ntest = pd.read_csv('../input/Kannada-MNIST/test.csv')\nDig_MNIST = pd.read_csv('../input/Kannada-MNIST/Dig-MNIST.csv')\n\ntest.tail()\n\nimport numpy as np\n\nx = train.iloc[:, 1:].values\ny = np.array([ [ 1 if (i == label) else 0 for i in range(10)] for label in train.iloc[:, :1].values ])\nx_train, x_valid, y_train, y_valid = train_test_split(x, y, test_size=10000, random_state=1997)\n\ninputs = Input(shape=(784,))\nnet = BatchNormalization()(inputs)\nnet = Dense(20, activation=None)(net)\nnet = Activation('elu')(net)\nnet = Dropout(rate=0.3)(net)\nnet = Dense(20, activation=None)(net)\nnet = Activation('elu')(net)\nprediction = Dense(10, activation='softmax')(net)\n\nmodel = Model(inputs=[inputs], output=[prediction])\nmodel.compile('RMSprop', losses.categorical_crossentropy, metrics=['accuracy'])\nmodel.summary()\n\n(x_train.shape, y_train.shape)\n\nmodel.fit(\n    x_train, y_train,\n    batch_size=16,\n    epochs=5,\n    verbose=2,\n    validation_data=(x_valid, y_valid),\n    callbacks=[\n      CSVLogger('log.csv'),\n      ModelCheckpoint('model.h5', save_best_only=True),\n    ]\n)\n\nx_test=test.drop('id', axis=1).iloc[:,:].values\nx_test.shape\n\npred_probas = model.predict(x_test - 0.5, batch_size=16)\nprediction = pred_probas.argmax(axis=1)\n\nsubmission = pd.read_csv('../input/Kannada-MNIST/sample_submission.csv')\nsubmission['label'] = prediction\nsubmission.to_csv(\"submission.csv\", index=False)\nsubmission.tail()","execution_count":null,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"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":1}