kopt - Hyper-parameter optimization for Keras

kopt is a hyper-parameter optimization library for Keras. It is based on hyperopt.

Installation

# Install hyperopt from the master branch
pip install git+https://github.com/hyperopt/hyperopt.git

# Install kopt
pip install kopt

Alternatively, kopt can be installed using Conda (most easily obtained via the Miniconda Python distribution):

conda install -c bioconda kopt

Installation issues

• Reported by gokceneraslan - 2018-03-11

  • hyperopt on pypi doesn't work with latest networkx 2, there are several issues. Maybe it would have been better to wait for the upcoming hyperopt release and then pin required hyperopt to new version.

    • possible solution to networkx 2 issue: pip install networkx==1.11 before installing hyperopt

Getting started

Here is an example of hyper-parameter optimization for the Keras IMDB example model.

from keras.datasets import imdb
from keras.preprocessing import sequence
from keras.models import Sequential
import keras.layers as kl
from keras.optimizers import Adam
# kopt and hyoperot imports
from kopt import CompileFN, KMongoTrials, test_fn
from hyperopt import fmin, tpe, hp, STATUS_OK, Trials


# 1. define the data function returning training, (validation, test) data
def data(max_features=5000, maxlen=80):
    (x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=max_features)
    x_train = sequence.pad_sequences(x_train, maxlen=maxlen)
    x_test = sequence.pad_sequences(x_test, maxlen=maxlen)
    return (x_train[:100], y_train[:100], max_features), (x_test, y_test)


# 2. Define the model function returning a compiled Keras model
def model(train_data, lr=0.001,
          embedding_dims=128, rnn_units=64,
          dropout=0.2):
	# extract data dimensions
    max_features = train_data[2]

    model = Sequential()
    model.add(kl.Embedding(max_features, embedding_dims))
    model.add(kl.LSTM(rnn_units, dropout=dropout, recurrent_dropout=dropout))
    model.add(kl.Dense(1, activation='sigmoid'))

    model.compile(loss='binary_crossentropy',
                  optimizer=Adam(lr=lr),
                  metrics=['accuracy'])
    return model

# Specify the optimization metrics
db_name="imdb"
exp_name="myexp1"
objective = CompileFN(db_name, exp_name,
                      data_fn=data,
                      model_fn=model,
                      loss_metric="acc", # which metric to optimize for
                      loss_metric_mode="max",  # try to maximize the metric
                      valid_split=.2, # use 20% of the training data for the validation set
                      save_model='best', # checkpoint the best model
                      save_results=True, # save the results as .json (in addition to mongoDB)
                      save_dir="./saved_models/")  # place to store the models

# define the hyper-parameter ranges
# see https://github.com/hyperopt/hyperopt/wiki/FMin for more info
hyper_params = {
	"data": {
	    "max_features": 100,
		"maxlen": 80,
	},
	"model": {
     	"lr": hp.loguniform("m_lr", np.log(1e-4), np.log(1e-2)), # 0.0001 - 0.01
	    "embedding_dims": hp.choice("m_emb", (64, 128)),
	    "rnn_units": 64,
		"dropout": hp.uniform("m_do", 0, 0.5),
	},
	"fit": {
	    "epochs": 20
	}
}

# test model training, on a small subset for one epoch
test_fn(objective, hyper_params)

# run hyper-parameter optimization sequentially (without any database)
trials = Trials()
best = fmin(objective, hyper_params, trials=trials, algo=tpe.suggest, max_evals=2)

# run hyper-parameter optimization in parallel (saving the results to MonogoDB)
# Follow the hyperopt guide:
# https://github.com/hyperopt/hyperopt/wiki/Parallelizing-Evaluations-During-Search-via-MongoDB
# KMongoTrials extends hyperopt.MongoTrials with convenience methods
trials = KMongoTrials(db_name, exp_name,
                      ip="localhost",
	              port=22334)
best = fmin(objective, hyper_params, trials=trials, algo=tpe.suggest, max_evals=2)

See also

• nbs/imdb_example.ipynb

The documentation of concise.hyopt (kopt was ported from concise.hyopt):

• Tutorial
• API documentation
• Jupyter notebook