Skip to main content

Browse the archive

https://github.com/GPflow/GPflow
21 November 2022, 03:07:34 UTC
Revision bb08f22e337d1487b8d9ab9944d8b9f7fff853ff authored by Vincent Dutordoir on 18 June 2018, 17:04:06 UTC, committed by Artem Artemev on 18 June 2018, 17:04:06 UTC 
Multi-output conditionals (#724)
 
* Introduction of MultiOutputFeatures (Mof) and MultiOutputKernels (Mok).
These are used to specify a particular setup of multi-output correlation.

* Multiple-dispatch for conditional. This allows GPflow to select the most efficient conditional code depending on your choice of Mof and Mok.

* Multiple-dispatch for Kuu and Kuf. Previously Kuu(.) and Kuf(.) were member functions of the feature class. This became cumbersome as the calculation of Kuu and Kuf also depends on the kernel used. In line with conditional we now also use multiple-dispatch to calculate Kuu and Kuf for a particular combination of Mok and Mof.

* The actual maths to efficiently calculate the output-correlated conditional (credits to @markvdw )

* sample_conditional function that makes sure that the most efficient code is used to get a sample from the conditional distribution.

* Minor: we updated a couple of models to use the new multi-output conditional.
1 parent 6baeb43
Raw File
Tip revision: bb08f22e337d1487b8d9ab9944d8b9f7fff853ff authored by Vincent Dutordoir on 18 June 2018, 17:04:06 UTC
Multi-output conditionals (#724)
Tip revision: bb08f22
setup.py 
#!/usr/bin/env python
# -*- coding: utf-8 -*-

# pylint: skip-file

from setuptools import setup
from setuptools import find_packages

import re
import os
import sys
from pkg_resources import parse_version

# load version form _version.py
exec(open("gpflow/_version.py").read())

# Dependencies of GPflow
requirements = [
    'numpy>=1.10.0',
    'scipy>=0.18.0',
    'pandas>=0.18.1',
    'multipledispatch>=0.4.9',
    'pytest>=3.5.0',
    'h5py>=2.7.0',
]

min_tf_version = '1.5.0'
tf_cpu = 'tensorflow>={}'.format(min_tf_version)
tf_gpu = 'tensorflow-gpu>={}'.format(min_tf_version)

# Only detect TF if not installed or outdated. If not, do not do not list as
# requirement to avoid installing over e.g. tensorflow-gpu
# To avoid this, rely on importing rather than the package name (like pip).

try:
    # If tf not installed, import raises ImportError
    import tensorflow as tf
    if parse_version(tf.VERSION) < parse_version(min_tf_version):
        # TF pre-installed, but below the minimum required version
        raise DeprecationWarning("TensorFlow version below minimum requirement")
except (ImportError, DeprecationWarning) as e:
    # Add TensorFlow to dependencies to trigger installation/update
    requirements.append(tf_cpu)

packages = find_packages('.')
package_data={'gpflow': ['gpflow/gpflowrc']}

setup(name='gpflow',
      version=__version__,
      author="James Hensman, Alex Matthews",
      author_email="james.hensman@gmail.com",
      description=("Gaussian process methods in tensorflow"),
      license="Apache License 2.0",
      keywords="machine-learning gaussian-processes kernels tensorflow",
      url="http://github.com/GPflow/GPflow",
      packages=packages,
      install_requires=requirements,
      package_data=package_data,
      include_package_data=True,
      test_suite='tests',
      extras_require={'Tensorflow with GPU': [tf_gpu]},
      classifiers=[
          'License :: OSI Approved :: Apache Software License',
          'Natural Language :: English',
          'Operating System :: MacOS :: MacOS X',
          'Operating System :: Microsoft :: Windows',
          'Operating System :: POSIX :: Linux',
          'Programming Language :: Python :: 3.5',
          'Programming Language :: Python :: 3.6',
          'Topic :: Scientific/Engineering :: Artificial Intelligence'
      ])
back to top