我们从Python开源项目中，提取了以下33个代码示例，用于说明如何使用keras.constraints.maxnorm()。
def train(img_shape): classes = ['ALB', 'BET', 'DOL', 'LAG', 'NoF', 'OTHER', 'SHARK', 'YFT'] # Model model = Sequential() model.add(Convolution2D( 32, 3, 3, input_shape=img_shape, activation='relu', W_constraint=maxnorm(3))) model.add(Dropout(0.2)) model.add(Convolution2D(32, 3, 3, activation='relu', W_constraint=maxnorm(3))) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Flatten()) model.add(Dense(512, activation='relu', W_constraint=maxnorm(3))) model.add(Dropout(0.5)) model.add(Dense(len(classes), activation='softmax')) features, labels = get_featurs_labels(img_shape) model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy']) model.fit(features, labels, nb_epoch=10, batch_size=32, validation_split=0.2, verbose=1) return model
def buildConvolution(self, name): filters = self.params.get('filters') nb_filter = self.params.get('nb_filter') assert filters assert nb_filter convs = [] for fsz in filters: layer_name = '%s-conv-%d' % (name, fsz) conv = Convolution2D( nb_filter=nb_filter, nb_row=fsz, nb_col=self.wdim, border_mode='valid', init='glorot_uniform', W_constraint=maxnorm(self.params.get('w_maxnorm')), b_constraint=maxnorm(self.params.get('b_maxnorm')), name=layer_name ) convs.append(conv) self.layers['%s-convolution' % name] = convs
def load_model(data): ''' Load keras model. ''' model = Sequential() model.add(Dense(data.shape[1], activation='relu', input_dim=data.shape[1], kernel_constraint=maxnorm(3))) model.add(Dropout(0.25)) #model.add(Dense(16, activation='relu', kernel_constraint=maxnorm(3))) #model.add(Dropout(0.25)) model.add(Dense(1, activation='sigmoid')) model.compile(optimizer='RMSprop', loss='binary_crossentropy', metrics=['accuracy']) return model
def create_model(learning_rate=0.1, momentum=0.9): model = Sequential() model.add(Convolution2D(20, 9, 9, border_mode='same', input_shape=(3, SIZE, SIZE))) model.add(Activation("relu")) model.add(MaxPooling2D(pool_size=(2,2), strides=(2,2))) model.add(Convolution2D(50, 5, 5, activation = "relu")) model.add(Activation("relu")) model.add(MaxPooling2D(pool_size=(2,2), strides=(2,2))) model.add(Flatten()) model.add(Dense(768, input_dim=3072, init='uniform', activation = 'relu')) model.add(Dropout(0.1)) model.add(Dense(384, init = 'uniform', activation = 'relu', W_constraint=maxnorm(3))) model.add(Dense(4)) model.add(Activation("softmax")) sgd = SGD(lr=learning_rate, momentum=momentum, nesterov=True, decay=1e-6) model.compile(optimizer=sgd, loss='categorical_crossentropy', metrics=["accuracy"]) return model
def CNNWithKeywordLayer(embed_matrix, embed_input, sequence_length, keywords_length, filter_sizes, num_filters, dropout_prob, hidden_dims, model_variation, embedding_dim=300): ''' 2-way input model: left is cnn for sentence embedding while right is keywords ''' embed1 = Embedding(embed_input, embedding_dim,input_length=sequence_length, weights=[embed_matrix]) # 1. question model part question_branch = Sequential() cnn_model = TextCNN(sequence_length, embedding_dim, filter_sizes, num_filters) question_branch.add(embed1) question_branch.add(cnn_model) # 2. keyword model part #keyword_branch = KeywordLayer(keywords_length, embed_input, embedding_dim, embed_matrix) keyword_branch = LSTMLayer(embed_matrix, embed_input, keywords_length, dropout_prob, hidden_dims, embedding_dim) # 3. merge layer merged = Merge([question_branch, keyword_branch], mode='concat') final_model = Sequential() final_model.add(merged) final_model.add(Dense(hidden_dims, W_constraint = maxnorm(3))) final_model.add(Dropout(0.5)) final_model.add(Activation('relu')) final_model.add(Dense(1)) final_model.add(Activation('sigmoid')) #sgd = SGD(lr=0.01, momentum=0.9) final_model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['accuracy']) return final_model
def QuestionWithAnswersModel(embed_matrix, embed_input, sequence_length, ans_cnt, keywords_length, filter_sizes, num_filters, dropout_prob, hidden_dims, embedding_dim=300): ''' path1: question embedding (CNN model) path2: answer embeddin(Hierachical RNN model) merge ''' # path 1 embed1 = Embedding(embed_input, embedding_dim,input_length=sequence_length, weights=[embed_matrix]) question_branch = Sequential() cnn_model = TextCNN(sequence_length, embedding_dim, filter_sizes, num_filters) question_branch.add(embed1) question_branch.add(cnn_model) # path 2 answer_branch = HierarchicalRNN(embed_matrix, embed_input, ans_cnt, keywords_length, embedding_dim) merged = Merge([question_branch, answer_branch], mode='concat') final_model = Sequential() final_model.add(merged) final_model.add(Dense(hidden_dims, W_constraint = maxnorm(3))) final_model.add(Dropout(0.5)) final_model.add(Activation('relu')) final_model.add(Dense(1)) final_model.add(Activation('sigmoid')) final_model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['accuracy']) return final_model # vim: set expandtab ts=4 sw=4 sts=4 tw=100:
def test_maxnorm(): for m in test_values: norm_instance = constraints.maxnorm(m) normed = norm_instance(K.variable(example_array)) assert(np.all(K.eval(normed) < m)) # a more explicit example norm_instance = constraints.maxnorm(2.0) x = np.array([[0, 0, 0], [1.0, 0, 0], [3, 0, 0], [3, 3, 3]]).T x_normed_target = np.array([[0, 0, 0], [1.0, 0, 0], [2.0, 0, 0], [2. / np.sqrt(3), 2. / np.sqrt(3), 2. / np.sqrt(3)]]).T x_normed_actual = K.eval(norm_instance(K.variable(x))) assert_allclose(x_normed_actual, x_normed_target, rtol=1e-05)
def setup_model(embeddings, seq_len, vocab_size): # Add input inputs = Input(shape=(seq_len, ), dtype='int32', name='inputs') # Add word vector embeddings embedding = Embedding(input_dim=vocab_size, output_dim=embedding_size, input_length=seq_len, name='embedding', trainable=True)(inputs) h = GlobalAveragePooling1D()(embedding) # Add output layer output = Dense(units=output_size, activation='sigmoid', kernel_initializer='he_normal', # kernel_regularizer=regularizers.l2(l2_reg_lambda), # kernel_constraint=maxnorm(max_norm), # bias_constraint=maxnorm(max_norm), name='output')(h) # build the model model = Model(inputs=inputs, outputs=output) model.compile(loss={'output':'binary_crossentropy'}, optimizer=Adam(lr=base_lr, epsilon=1e-6, decay=decay_rate), metrics=["accuracy"]) return model
def train(img_shape): # Model model = Sequential() model.add( Convolution2D(32, 3, 3, input_shape=img_shape, activation='relu', W_constraint=maxnorm(3), dim_ordering='tf')) model.add(Dropout(0.2)) model.add(Convolution2D(32, 3, 3, activation='relu', W_constraint=maxnorm(3), dim_ordering='tf')) model.add(MaxPooling2D()) model.add(Convolution2D(32, 3, 3, activation='relu', W_constraint=maxnorm(3), dim_ordering='tf')) model.add(MaxPooling2D()) model.add(Convolution2D(32, 3, 3, activation='relu', W_constraint=maxnorm(3), dim_ordering='tf')) model.add(MaxPooling2D()) model.add(Flatten()) model.add(Dense(512, activation='relu', W_constraint=maxnorm(3))) model.add(Dropout(0.5)) model.add(Dense(8)) model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy']) for features, labels in feature_labels_generator(): model.fit(features, labels, nb_epoch=1) # TODO: Get generator to # samples_per_epoch = 100 # model.fit_generator(feature_labels_generator(), samples_per_epoch, nb_epoch=10) return model
def create_model(dropout_rate=0.0, weight_constraint=0): # create model model = Sequential() model.add(Dense(12, input_dim=8, init='uniform', activation='softplus', W_constraint=maxnorm(weight_constraint))) model.add(Dropout(dropout_rate)) model.add(Dense(1, init='uniform', activation='sigmoid')) # Compile model model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) return model # fix random seed for reproducibility
def create_model(neurons=1): # create model model = Sequential() model.add(Dense(neurons, input_dim=8, init='uniform', activation='softplus', W_constraint=maxnorm(4))) model.add(Dropout(0.1)) model.add(Dense(1, init='uniform', activation='sigmoid')) # Compile model model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) return model # fix random seed for reproducibility
def buildConvolution(self, name): filters = self.params.get('filters') nb_filter = self.params.get('nb_filter') assert filters assert nb_filter convs = [] for fsz in filters: layer_name = '%s-conv-%d' % (name, fsz) conv = Convolution1D( nb_filter=nb_filter, filter_length=fsz, border_mode='valid', #activation='relu', subsample_length=1, init='glorot_uniform', #init=init, #init=lambda shape, name: initializations.uniform(shape, scale=0.01, name=name), W_constraint=maxnorm(self.params.get('w_maxnorm')), b_constraint=maxnorm(self.params.get('b_maxnorm')), #W_regularizer=regularizers.l2(self.params.get('w_l2')), #b_regularizer=regularizers.l2(self.params.get('b_l2')), #input_shape=(self.q_length, self.wdim), name=layer_name ) convs.append(conv) self.layers['%s-convolution' % name] = convs
def test_maxnorm(self): from keras.constraints import maxnorm for m in self.some_values: norm_instance = maxnorm(m) normed = norm_instance(self.example_array) assert (np.all(normed.eval() < m)) # a more explicit example norm_instance = maxnorm(2.0) x = np.array([[0, 0, 0], [1.0, 0, 0], [3, 0, 0], [3, 3, 3]]).T x_normed_target = np.array([[0, 0, 0], [1.0, 0, 0], [2.0, 0, 0], [2./np.sqrt(3), 2./np.sqrt(3), 2./np.sqrt(3)]]).T x_normed_actual = norm_instance(x).eval() assert_allclose(x_normed_actual, x_normed_target)
def feed_forward_net(input, output, hidden_layers=[64, 64], activations='relu', dropout_rate=0., l2=0., constrain_norm=False): ''' Helper function for building a Keras feed forward network. input: Keras Input object appropriate for the data. e.g. input=Input(shape=(20,)) output: Function representing final layer for the network that maps from the last hidden layer to output. e.g. if output = Dense(10, activation='softmax') if we're doing 10 class classification or output = Dense(1, activation='linear') if we're doing regression. ''' state = input if isinstance(activations, str): activations = [activations] * len(hidden_layers) for h, a in zip(hidden_layers, activations): if l2 > 0.: w_reg = keras.regularizers.l2(l2) else: w_reg = None const = maxnorm(2) if constrain_norm else None state = Dense(h, activation=a, kernel_regularizer=w_reg, kernel_constraint=const)(state) if dropout_rate > 0.: state = Dropout(dropout_rate)(state) return output(state)
def convnet(input, output, dropout_rate=0., input_shape=(1, 28, 28), batch_size=100, l2_rate=0.001, nb_epoch=12, img_rows=28, img_cols=28, nb_filters=64, pool_size=(2, 2), kernel_size=(3, 3), activations='relu', constrain_norm=False): ''' Helper function for building a Keras convolutional network. input: Keras Input object appropriate for the data. e.g. input=Input(shape=(20,)) output: Function representing final layer for the network that maps from the last hidden layer to output. e.g. if output = Dense(10, activation='softmax') if we're doing 10 class classification or output = Dense(1, activation='linear') if we're doing regression. ''' const = maxnorm(2) if constrain_norm else None state = Convolution2D(nb_filters, kernel_size, padding='valid', input_shape=input_shape, activation=activations, kernel_regularizer=l2(l2_rate), kernel_constraint=const)(input) state = Convolution2D(nb_filters, kernel_size, activation=activations, kernel_regularizer=l2(l2_rate), kernel_constraint=const)(state) state = MaxPooling2D(pool_size=pool_size)(state) state = Flatten()(state) if dropout_rate > 0.: state = Dropout(dropout_rate)(state) state = Dense(128, activation=activations, kernel_regularizer=l2(l2_rate), kernel_constraint=const)(state) if dropout_rate > 0.: state = Dropout(dropout_rate)(state) return output(state)
def base_model(): model = Sequential() model.add(Conv2D(32, (3, 3), padding='same', activation='relu', input_shape=x_train.shape[1:])) model.add(Dropout(0.2)) model.add(Conv2D(32,(3,3),padding='same', activation='relu')) model.add(MaxPooling2D(pool_size=(2,2))) model.add(Conv2D(64,(3,3),padding='same',activation='relu')) model.add(Dropout(0.2)) model.add(Conv2D(64,(3,3),padding='same',activation='relu')) model.add(MaxPooling2D(pool_size=(2,2))) model.add(Conv2D(128,(3,3),padding='same',activation='relu')) model.add(Dropout(0.2)) model.add(Conv2D(128,(3,3),padding='same',activation='relu')) model.add(MaxPooling2D(pool_size=(2,2))) model.add(Flatten()) model.add(Dropout(0.2)) model.add(Dense(1024,activation='relu',kernel_constraint=maxnorm(3))) model.add(Dropout(0.2)) model.add(Dense(num_classes, activation='softmax')) sgd = SGD(lr=0.01, momentum=0.9, decay=1e-6, nesterov=False) # Train model model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) return model
def build_model(data, word_weights, tag_window=5, embed_dim=100): batch_size = 32 nb_epoch = 16 nb_class = 4 hidden_dim = 128 train_x = np.array(list(data['x'])) train_y = np.array(list(data['y'])) train_y = np_utils.to_categorical(train_y, nb_class) print(train_x.shape) print(train_y.shape) input_x = Input(shape=(tag_window, ), dtype='float32', name='input_x') embed_x = Embedding(output_dim=embed_dim, input_dim=word_weights.shape[0], input_length=tag_window, weights=[word_weights], name='embed_x')(input_x) bi_lstm = Bidirectional(LSTM(hidden_dim, return_sequences=False), merge_mode='sum')(embed_x) x_dropout = Dropout(0.5)(bi_lstm) x_output = Dense(nb_class, # kernel_regularizer=regularizers.l2(0.01), # kernel_constraint=maxnorm(3.0), # activity_regularizer=regularizers.l2(0.01), activation='softmax')(x_dropout) model = Model(input=[input_x], output=[x_output]) model.compile(optimizer='adamax', loss='categorical_crossentropy',metrics=['accuracy']) model.fit([train_x], [train_y], validation_split=0.2, batch_size=batch_size, epochs=nb_epoch, shuffle=True)
def build_cnn(kernel_size, nb_filters, embed_x): """ """ name = 'cnn_' + str(kernel_size) cnn_x = Conv1D(filters=nb_filters, kernel_size=kernel_size, padding='valid', activation='tanh')(embed_x) # kernel_regularizer=regularizers.l2(0.01), # kernel_constraint=maxnorm(3.0), # activity_regularizer=regularizers.l2(0.01))(embed_x) maxPooling_x = MaxPooling1D(kernel_size)(cnn_x) return maxPooling_x
def build_model(data, word_weights, tag_window=5, embed_dim=100): batch_size = 50 nb_epoch = 8 nb_class = 4 hidden_dim = 128 nb_filters = 100 train_x = np.array(list(data['x'])) train_y = np.array(list(data['y'])) train_y = np_utils.to_categorical(train_y, nb_class) print(train_x.shape) print(train_y.shape) input_x = Input(shape=(tag_window, ), dtype='float32', name='input_x') embed_x = Embedding(output_dim=embed_dim, input_dim=word_weights.shape[0], input_length=tag_window, weights=[word_weights], name='embed_x')(input_x) # bi_lstm = Bidirectional(LSTM(hidden_dim, return_sequences=False), merge_mode='sum')(embed_x) maxPooling_2 = build_cnn(2, nb_filters, embed_x) print('finish 2') maxPooling_3 = build_cnn(3, nb_filters, embed_x) print('finish 3') maxPooling_4 = build_cnn(4, nb_filters, embed_x) print('finish 4') maxPooling_5 = build_cnn(5, nb_filters, embed_x) maxPooling = concatenate([maxPooling_2, maxPooling_3, maxPooling_4, maxPooling_5], axis=1) x_dropout = Dropout(0.5)(maxPooling_2) x_flatten = Flatten()(x_dropout) x_output = Dense(nb_class, # kernel_regularizer=regularizers.l2(0.01), # kernel_constraint=maxnorm(3.0), # activity_regularizer=regularizers.l2(0.01), activation='softmax')(x_flatten) model = Model(input=[input_x], output=[x_output]) model.compile(optimizer='adamax', loss='categorical_crossentropy',metrics=['accuracy']) model.fit([train_x], [train_y], validation_split=0.2, batch_size=batch_size, epochs=nb_epoch, shuffle=True)
def model(X_train, Y_train, X_test, Y_test): W_maxnorm = 3 DROPOUT = {{choice([0.3,0.5,0.7])}} model = Sequential() model.add(Convolution2D(64, 1, 5, border_mode='same', input_shape=(4, 1, DATASIZE),activation='relu',W_constraint=maxnorm(W_maxnorm))) model.add(MaxPooling2D(pool_size=(1, 5),strides=(1,3))) model.add(Flatten()) model.add(Dense(32,activation='relu')) model.add(Dropout(DROPOUT)) model.add(Dense(32,activation='relu')) model.add(Dropout(DROPOUT)) model.add(Dense(2)) model.add(Activation('softmax')) myoptimizer = RMSprop(lr={{choice([0.01,0.001,0.0001])}}, rho=0.9, epsilon=1e-06) mylossfunc = 'categorical_crossentropy' model.compile(loss=mylossfunc, optimizer=myoptimizer,metrics=['accuracy']) model.fit(X_train, Y_train, batch_size=100, nb_epoch=5,validation_split=0.1) score, acc = model.evaluate(X_test,Y_test) model_arch = 'MODEL_ARCH' bestaccfile = join('TOPDIR',model_arch,model_arch+'_hyperbestacc') reportAcc(acc,score,bestaccfile) return {'loss': score, 'status': STATUS_OK,'model':(model.to_json(),myoptimizer,mylossfunc)}
def build_model(): main_input = Input(shape=(maxlen, ), dtype='int32', name='main_input') embedding = Embedding(max_features, embedding_dims, weights=[np.matrix(W)], input_length=maxlen, name='embedding')(main_input) embedding = Dropout(0.50)(embedding) conv4 = Convolution1D(nb_filter=nb_filter, filter_length=4, border_mode='valid', activation='relu', subsample_length=1, name='conv4')(embedding) maxConv4 = MaxPooling1D(pool_length=2, name='maxConv4')(conv4) conv5 = Convolution1D(nb_filter=nb_filter, filter_length=5, border_mode='valid', activation='relu', subsample_length=1, name='conv5')(embedding) maxConv5 = MaxPooling1D(pool_length=2, name='maxConv5')(conv5) x = merge([maxConv4, maxConv5], mode='concat') x = Dropout(0.15)(x) x = RNN(rnn_output_size)(x) x = Dense(hidden_dims, activation='relu', init='he_normal', W_constraint = maxnorm(3), b_constraint=maxnorm(3), name='mlp')(x) x = Dropout(0.10, name='drop')(x) output = Dense(1, init='he_normal', activation='sigmoid', name='output')(x) model = Model(input=main_input, output=output) model.compile(loss={'output':'binary_crossentropy'}, optimizer=Adadelta(lr=0.95, epsilon=1e-06), metrics=["accuracy"]) return model
def build_model(): main_input = Input(shape=(maxlen, ), dtype='int32', name='main_input') embedding = Embedding(max_features, embedding_dims, weights=[np.matrix(W)], input_length=maxlen, name='embedding')(main_input) embedding = Dropout(0.50)(embedding) conv4 = Conv1D(filters=nb_filter, kernel_size=4, padding='valid', activation='relu', strides=1, name='conv4')(embedding) maxConv4 = MaxPooling1D(pool_size=2, name='maxConv4')(conv4) conv5 = Conv1D(filters=nb_filter, kernel_size=5, padding='valid', activation='relu', strides=1, name='conv5')(embedding) maxConv5 = MaxPooling1D(pool_size=2, name='maxConv5')(conv5) # x = merge([maxConv4, maxConv5], mode='concat') x = keras.layers.concatenate([maxConv4, maxConv5]) x = Dropout(0.15)(x) x = RNN(rnn_output_size)(x) x = Dense(hidden_dims, activation='relu', kernel_initializer='he_normal', kernel_constraint = maxnorm(3), bias_constraint=maxnorm(3), name='mlp')(x) x = Dropout(0.10, name='drop')(x) output = Dense(1, kernel_initializer='he_normal', activation='sigmoid', name='output')(x) model = Model(inputs=main_input, outputs=output) model.compile(loss='binary_crossentropy', # optimizer=Adadelta(lr=0.95, epsilon=1e-06), # optimizer=Adadelta(lr=1.0, rho=0.95, epsilon=1e-08, decay=0.0), # optimizer=Adagrad(lr=0.01, epsilon=1e-08, decay=1e-4), metrics=["accuracy"]) return model
def build_model(): print('Build model...%d of %d' % (i + 1, folds)) main_input = Input(shape=(maxlen, ), dtype='int32', name='main_input') embedding = Embedding(max_features, embedding_dims, weights=[np.matrix(W)], input_length=maxlen, name='embedding')(main_input) embedding = Dropout(0.50)(embedding) conv4 = Convolution1D(nb_filter=nb_filter, filter_length=4, border_mode='valid', activation='relu', subsample_length=1, name='conv4')(embedding) maxConv4 = MaxPooling1D(pool_length=2, name='maxConv4')(conv4) conv5 = Convolution1D(nb_filter=nb_filter, filter_length=5, border_mode='valid', activation='relu', subsample_length=1, name='conv5')(embedding) maxConv5 = MaxPooling1D(pool_length=2, name='maxConv5')(conv5) x = merge([maxConv4, maxConv5], mode='concat') x = Dropout(0.15)(x) x = RNN(rnn_output_size)(x) x = Dense(hidden_dims, activation='relu', init='he_normal', W_constraint = maxnorm(3), b_constraint=maxnorm(3), name='mlp')(x) x = Dropout(0.10, name='drop')(x) output = Dense(1, init='he_normal', activation='sigmoid', name='output')(x) model = Model(input=main_input, output=output) model.compile(loss={'output':'binary_crossentropy'}, optimizer=Adadelta(lr=0.95, epsilon=1e-06), metrics=["accuracy"]) return model
def build_model(): main_input = Input(shape=(maxlen, ), dtype='int32', name='main_input') embedding = Embedding(max_features, embedding_dims, weights=[np.matrix(W)], input_length=maxlen, name='embedding')(main_input) embedding = Dropout(0.50)(embedding) conv4 = Convolution1D(nb_filter=nb_filter, filter_length=4, border_mode='valid', activation='relu', subsample_length=1, name='conv4')(embedding) maxConv4 = MaxPooling1D(pool_length=2, name='maxConv4')(conv4) conv5 = Convolution1D(nb_filter=nb_filter, filter_length=5, border_mode='valid', activation='relu', subsample_length=1, name='conv5')(embedding) maxConv5 = MaxPooling1D(pool_length=2, name='maxConv5')(conv5) x = merge([maxConv4, maxConv5], mode='concat') x = Dropout(0.15)(x) x = RNN(rnn_output_size)(x) x = Dense(hidden_dims, activation='relu', init='he_normal', W_constraint = maxnorm(3), b_constraint=maxnorm(3), name='mlp')(x) x = Dropout(0.10, name='drop')(x) output = Dense(nb_classes, init='he_normal', activation='softmax', name='output')(x) model = Model(input=main_input, output=output) model.compile(loss={'output':'categorical_crossentropy'}, optimizer=Adadelta(lr=0.95, epsilon=1e-06), metrics=["accuracy"]) return model
def model(): model = Sequential() #input layer model.add(Dense(120, input_dim=input_dims)) #, kernel_constraint=maxnorm(5) #model.add(BatchNormalization()) model.add(Activation('relu')) model.add(Dropout(0.2)) # Reduce Overfitting With Dropout Regularization # hidden layers model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) model.add(Dense(120)) model.add(Activation(act_func)) model.add(Dropout(0.2)) # output layer (y_pred) model.add(Dense(1, activation='linear')) # Use a large learning rate with decay and a large momentum. Increase your learning rate by a factor of 10 to 100 and use a high momentum value of 0.9 or 0.99 # sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) # adam = Adam(lr=0.01, beta_1=0.9, beta_2=0.999, epsilon=1e-08, decay=0.0) # compile this model model.compile(loss='mean_squared_error', # one may use 'mean_absolute_error' as alternative optimizer='adam', metrics=[r2_keras] # you can add several if needed ) # Visualize NN architecture print(model.summary()) return model # initialize input dimension
def build_model(max_length=1000, nb_filters=64, kernel_size=3, pool_size=2, regularization=0.01, weight_constraint=2., dropout_prob=0.4, clear_session=True): if clear_session: K.clear_session() model = Sequential() model.add(Embedding( embeddings.shape[0], embeddings.shape[1], input_length=max_length, trainable=False, weights=[embeddings])) model.add(Conv1D(nb_filters, kernel_size, activation='relu')) model.add(Conv1D(nb_filters, kernel_size, activation='relu')) model.add(MaxPooling1D(pool_size)) model.add(Dropout(dropout_prob)) model.add(Conv1D(nb_filters * 2, kernel_size, activation='relu')) model.add(Conv1D(nb_filters * 2, kernel_size, activation='relu')) model.add(MaxPooling1D(pool_size)) model.add(Dropout(dropout_prob)) model.add(GlobalAveragePooling1D()) model.add(Dense(1, kernel_regularizer=l2(regularization), kernel_constraint=maxnorm(weight_constraint), activation='sigmoid')) model.compile( loss='binary_crossentropy', optimizer='rmsprop', metrics=['accuracy']) return model
def create_model(data): ''' Load keras model. ''' # Entity branch entity_inputs = Input(shape=(data[0].shape[1],)) entity_x = Dense(data[0].shape[1], activation='relu', kernel_constraint=maxnorm(3))(entity_inputs) entity_x = Dropout(0.25)(entity_x) #entity_x = Dense(50, activation='relu', # kernel_constraint=maxnorm(3))(entity_x) #entity_x = Dropout(0.25)(entity_x) # Candidate branch candidate_inputs = Input(shape=(data[1].shape[1],)) candidate_x = Dense(data[1].shape[1], activation='relu', kernel_constraint=maxnorm(3))(candidate_inputs) candidate_x = Dropout(0.25)(candidate_x) #candidate_x = Dense(50, activation='relu', # kernel_constraint=maxnorm(3))(candidate_x) #candidate_x = Dropout(0.25)(candidate_x) # Cosine proximity # cos_x = dot([entity_x, candidate_x], axes=1, normalize=False) # cos_x = concatenate([entity_x, candidate_x]) # cos_output = Dense(1, activation='sigmoid')(cos_x) # Match branch match_inputs = Input(shape=(data[2].shape[1],)) match_x = Dense(data[1].shape[1], activation='relu', kernel_constraint=maxnorm(3))(match_inputs) match_x = Dropout(0.25)(match_x) # Merge x = concatenate([entity_x, candidate_x, match_x]) x = Dense(32, activation='relu', kernel_constraint=maxnorm(3))(x) x = Dropout(0.25)(x) x = Dense(16, activation='relu', kernel_constraint=maxnorm(3))(x) x = Dropout(0.25)(x) x = Dense(8, activation='relu', kernel_constraint=maxnorm(3))(x) x = Dropout(0.25)(x) predictions = Dense(1, activation='sigmoid')(x) model = Model(inputs=[entity_inputs, candidate_inputs, match_inputs], outputs=predictions) model.compile(optimizer='RMSprop', loss='binary_crossentropy', metrics=['accuracy']) return model
def build_model(data, word_weights, max_len, tag_window=5, embed_dim=100, location_dim=10): batch_size = 2048 nb_epoch = 16 nb_class = 4 hidden_dim = 128 train_x = np.array(list(data['x'])) train_l = np.array(list(data['l'])) train_y = np.array(list(data['y'])) train_y = np_utils.to_categorical(train_y, nb_class) print(train_x.shape) print(train_l.shape) print(train_y.shape) input_x = Input(shape=(tag_window, ), dtype='float32', name='input_x') input_l = Input(shape=(tag_window, ), dtype='float32', name='input_l') embed_x = Embedding(output_dim=embed_dim, input_dim=word_weights.shape[0], input_length=tag_window, weights=[word_weights], name='embed_x')(input_x) embed_l = Embedding(output_dim=location_dim, input_dim=max_len, input_length=tag_window, name='embed_l')(input_l) merge_embed = merge([embed_x, embed_l], mode='concat', concat_axis=2) bi_lstm = Bidirectional(LSTM(hidden_dim, return_sequences=False), merge_mode='sum')(merge_embed) x_dropout = Dropout(0.5)(bi_lstm) x_output = Dense(nb_class, # kernel_regularizer=regularizers.l2(0.01), # kernel_constraint=maxnorm(3.0), # activity_regularizer=regularizers.l2(0.01), activation='softmax')(x_dropout) model = Model(inputs=[input_x, input_l], outputs=[x_output]) model.compile(optimizer='adamax', loss='categorical_crossentropy',metrics=['accuracy']) print('Train...') model_path = './model/location_128hidden_2048batch' modelcheckpoint = ModelCheckpoint(model_path, verbose=1, save_best_only=True) model.fit([train_x, train_l], [train_y], validation_split=0.2, batch_size=batch_size, epochs=nb_epoch, shuffle=True)
