For intelligent applications NuPIC is great as a starting point of your development. However, a thing to keep in mind is that this field of computer science is new, HTM is fairly new. Or in the words from Jeff Hawkins: "This stuff is not easy. I can assure you that once you understand it, you will see a beauty in it. But most people take months to deeply understand the CLA. The tasks of creating hierarchies of CLAs and adding in motor capabilities are very difficult. Even just using the CLA in its current form is not trivial due to the learning required."
When you like to run NuPic on Oracle Linux a number of steps might be a bit different from the installation on a MacBook. Also a couple of dependencies are in place before you can install NuPic on Oracle Linux which are: Python 2.7, Python development headers, pip, wheel, numpy and C++ compiler like gcc or clang.
Python development headers
Next to Python, which most likely will be shipping with your Oracle Linux installation you will have to make sure that you have the Python development headers. You can check if this is installed by executing the below command. In my case Python development headers was already installed.
[root@localhost ~]# rpm -qa | grep python-devel python-devel-2.7.5-18.0.1.el7_1.1.x86_64 [root@localhost ~]#
In case you do not get a result you will have to install the Python development headers by executing a yum install command as shown below:
[root@localhost ~]# yum install python-devel
pip
one of the requirements to be able to install NupiC is to install pip. pip is a package management system used to install and manage software packages written in Python. Many packages can be found in the Python Package Index (PyPI). If you have installed the Python setuptools which I describe in this blogpost the installation of pip can be done by using the easy_install command which is part of the setuptools distribution.
[root@localhost ~]# easy_install pip Searching for pip Best match: pip 6.1.1 Adding pip 6.1.1 to easy-install.pth file Installing pip script to /usr/bin Installing pip3.4 script to /usr/bin Installing pip3 script to /usr/bin Using /usr/lib/python2.7/site-packages Processing dependencies for pip Finished processing dependencies for pip [root@localhost ~]#
wheel
wheel is required as a dependency. Wheel(s) are a built-package format for Python. A wheel is a ZIP-format archive with a specially formatted filename and the .whl extension. It is designed to contain all the files for a PEP 376 compatible install in a way that is very close to the on-disk format. Many packages will be properly installed with only the “Unpack” step (simply extracting the file onto sys.path), and the unpacked archive preserves enough information to “Spread” (copy data and scripts to their final locations) at any later time. You can install wheel with the just installed pip by executing the below command. Which resulted in my case in some warnings which you can (should) resolve however are not blocking the installation.
[root@localhost ~]# pip install wheel
/usr/lib/python2.7/site-packages/pip/_vendor/requests/packages/urllib3/util/ssl_.py:79: InsecurePlatformWarning: A true SSLContext object is not available. This prevents urllib3 from configuring SSL appropriately and may cause certain SSL connections to fail. For more information, see https://urllib3.readthedocs.org/en/latest/security.html#insecureplatformwarning.
InsecurePlatformWarning
You are using pip version 6.1.1, however version 7.1.2 is available.
You should consider upgrading via the 'pip install --upgrade pip' command.
Collecting wheel
/usr/lib/python2.7/site-packages/pip/_vendor/requests/packages/urllib3/util/ssl_.py:79: InsecurePlatformWarning: A true SSLContext object is not available. This prevents urllib3 from configuring SSL appropriately and may cause certain SSL connections to fail. For more information, see https://urllib3.readthedocs.org/en/latest/security.html#insecureplatformwarning.
InsecurePlatformWarning
100% |ââââââââââââââââââââââââââââââââ| 65kB 1.8MB/s
Installing collected packages: wheel
Successfully installed wheel-0.26.0
[root@localhost ~]#
NumPy
it will not come as a surprise that NumPy is required to be installed on the system. NumPy is is the fundamental package for scientific computing with Python. It contains among other things; a powerful N-dimensional array object, sophisticated (broadcasting) functions, tools for integrating C/C++ and Fortran code, useful linear algebra, Fourier transform, and random number capabilities. Besides its obvious scientific uses, NumPy can also be used as an efficient multi-dimensional container of generic data. Arbitrary data-types can be defined. This allows NumPy to seamlessly and speedily integrate with a wide variety of databases
Numpy can be installed by executing the command:
pip install numpy
Compiler
As the core of Nupic is written in C++ you will need a C++ compiler. The obvious choice in this is GCC which most likely is already installed on your system. You can check the availability with the below command, which shows in my example that it is installed.
[root@localhost ~]# rpm -qa | grep gcc gcc-4.8.3-9.el7.x86_64 libgcc-4.8.3-9.el7.x86_64 [root@localhost ~]#
In case it is not installed you can execute a yum install command to install gcc on your Oracle Linux machine. One small, however important note, gcc should be GCC 4.8.
Installing NuPic
After you ensured all dependencies are done you can install NuPic. The installation of NuPic on Oracle Linux is a bit different than the installation on for example a Mac. Reason for this is that The nupic.bindings binary distribution is not stored on PyPi along with the OS X distribution. NuPIC uses the wheel binary format, and PyPi does not support hosting Linux wheel files. This forces you to download the wheel file directly from Numenta and not from PyPi.
pip install https://s3-us-west-2.amazonaws.com/artifacts.numenta.org/numenta/nupic.core/releases/nupic.bindings/nupic.bindings-0.2.1-cp27-none-linux_x86_64.whl
pip install nupic
If all is ok the "pip install nupic" command should work like a charm. However, in case you run into a compiler error like the one shown below it might be that you are missing some additional prerequisites.
cc -c /tmp/tmphmvPkY/vers.cpp -o tmp/tmphmvPkY/vers.o --std=c++11
cc: error trying to exec 'cc1plus': execvp: No such file or directory
*WARNING* no libcapnp detected. Will download and build it from source now. If you have C++ Cap'n Proto installed, it may be out of date or is not being detected. Downloading and building libcapnp may take a while.
fetching https://capnproto.org/capnproto-c++-0.5.1.2.tar.gz into /tmp/pip-build-PHQZgs/pycapnp/bundled
configure: error: *** A compiler with support for C++11 language features is required.
To resolve this issue you will need to do a additional install of gcc-c++ by executing
[root@localhost ~]# yum install gcc-c++
Testing NuPic
to ensure your installation of Nupic was successful you can run a test with the test units provided in the github repository. Execute the py.test against test/unit/ which can be found in the github repository. This should look like the example below.
[root@localhost nupic-master]# py.test tests/unit/ === test session starts === platform linux2 -- Python 2.7.5 -- pytest-2.5.1 plugins: cov, xdist collected 844 items / 2 skipped tests/unit/nupic/utils_test.py ...... tests/unit/nupic/algorithms/anomaly_likelihood_jeff_test.py ...ss.. tests/unit/nupic/algorithms/anomaly_likelihood_test.py .................... tests/unit/nupic/algorithms/anomaly_test.py .............. tests/unit/nupic/algorithms/cells4_test.py . tests/unit/nupic/algorithms/cla_classifier_diff_test.py ................... tests/unit/nupic/algorithms/cla_classifier_test.py ................... tests/unit/nupic/algorithms/fast_cla_classifier_test.py ................... tests/unit/nupic/algorithms/knn_classifier_test.py .....s tests/unit/nupic/algorithms/nab_detector_test.py .. tests/unit/nupic/algorithms/sp_overlap_test.py .s.s tests/unit/nupic/algorithms/svm_test.py ..s tests/unit/nupic/algorithms/tp10x2_test.py . tests/unit/nupic/data/aggregator_test.py . tests/unit/nupic/data/dictutils_test.py ...... tests/unit/nupic/data/fieldmeta_test.py ..... tests/unit/nupic/data/file_record_stream_test.py ...... tests/unit/nupic/data/filters_test.py s tests/unit/nupic/data/functionsource_test.py ...... tests/unit/nupic/data/inference_shifter_test.py ........ tests/unit/nupic/data/record_stream_test.py ....... tests/unit/nupic/data/utils_test.py ....... tests/unit/nupic/data/generators/anomalyzer_test.py ........... tests/unit/nupic/data/generators/pattern_machine_test.py ......... tests/unit/nupic/data/generators/sequence_machine_test.py ..... tests/unit/nupic/encoders/adaptivescalar_test.py ....... tests/unit/nupic/encoders/category_test.py .. tests/unit/nupic/encoders/coordinate_test.py ................ tests/unit/nupic/encoders/date_test.py ........ tests/unit/nupic/encoders/delta_test.py ..... tests/unit/nupic/encoders/geospatial_coordinate_test.py ........... tests/unit/nupic/encoders/logenc_test.py ...... tests/unit/nupic/encoders/multi_test.py .. tests/unit/nupic/encoders/pass_through_encoder_test.py .... tests/unit/nupic/encoders/random_distributed_scalar_test.py ............... tests/unit/nupic/encoders/scalar_test.py ............. tests/unit/nupic/encoders/scalarspace_test.py . tests/unit/nupic/encoders/sdrcategory_test.py ... tests/unit/nupic/encoders/sparse_pass_through_encoder_test.py .... tests/unit/nupic/engine/network_test.py ......... tests/unit/nupic/engine/syntactic_sugar_test.py ..... tests/unit/nupic/engine/unified_py_parameter_test.py .. tests/unit/nupic/frameworks/opf/clamodel_classifier_helper_test.py ...................... tests/unit/nupic/frameworks/opf/clamodel_test.py ...... tests/unit/nupic/frameworks/opf/opf_metrics_test.py ............................... tests/unit/nupic/frameworks/opf/previous_value_model_test.py ...... tests/unit/nupic/frameworks/opf/safe_interpreter_test.py ........ tests/unit/nupic/frameworks/opf/two_gram_model_test.py ..... tests/unit/nupic/frameworks/opf/common_models/cluster_params_test.py . tests/unit/nupic/math/array_algorithms_test.py ... tests/unit/nupic/math/cast_mode_test.py s tests/unit/nupic/math/lgamma_test.py . tests/unit/nupic/math/nupic_random_test.py ............. tests/unit/nupic/math/sparse_binary_matrix_test.py ............s............ tests/unit/nupic/math/sparse_matrix_test.py ...s............................... tests/unit/nupic/regions/anomaly_region_test.py . tests/unit/nupic/regions/knn_anomaly_classifier_region_test.py .................... tests/unit/nupic/regions/pyregion_test.py .... tests/unit/nupic/regions/record_sensor_region_test.py . tests/unit/nupic/regions/regions_spec_test.py s...s...... tests/unit/nupic/research/connections_test.py ............. tests/unit/nupic/research/inhibition_object_test.py s tests/unit/nupic/research/sp_learn_inference_test.py s tests/unit/nupic/research/spatial_pooler_boost_test.py .. tests/unit/nupic/research/spatial_pooler_compatability_test.py ....ss.. tests/unit/nupic/research/spatial_pooler_compute_test.py .. tests/unit/nupic/research/spatial_pooler_cpp_api_test.py .............................. tests/unit/nupic/research/spatial_pooler_py_api_test.py .............................. tests/unit/nupic/research/spatial_pooler_unit_test.py s................................. tests/unit/nupic/research/temporal_memory_test.py ........................... tests/unit/nupic/research/tp10x2_test.py .... tests/unit/nupic/research/tp_constant_test.py ... tests/unit/nupic/research/tp_test.py .... tests/unit/nupic/research/monitor_mixin/metric_test.py .. tests/unit/nupic/research/monitor_mixin/trace_test.py .. tests/unit/nupic/support/configuration_test.py ............s.................... tests/unit/nupic/support/custom_configuration_test.py .........s.............. tests/unit/nupic/support/decorators_test.py .... tests/unit/nupic/support/object_json_test.py ............... tests/unit/nupic/support/consoleprinter_test/consoleprinter_test.py . === 825 passed, 21 skipped in 100.95 seconds === [root@localhost nupic-master]#
This should enable you to start with exploring NuPic
No comments:
Post a Comment