0.4.1

• Manualy update of nupic.bindings dev version after botched release attempt

0.4.0

• Updated hello_tm.py to use accessors
• Updated TP_shim.py to use accessors Updated columnForCell and _validateCell in FastTemporalMemory to conform to their docstrings, which is needed for the change to TP_shim.py
• Updated temporal memory monitor mixin to use accessors
• Updated temporal_memory_test.py to use accessor methods.
• Added accessors to temporal_memory.py
• Change temporalImp to tm_py for both networks and add comment about it being a temporary value until C++ TM is implemented
• Refactored to remove common code between network_checkpoint_test.py and temporal_memory_compatibility_test.py
• Use named constants from nupic.data.fieldmeta in aggregator module instead of naked constants.
• Fix AttributeError: 'TPShim' object has no attribute 'topDownCompute'
• Support more parameters in TPShim
• Serialize remaining fields in CLAModel using capnproto
• Enforce pyproj==1.9.3 in requirements.txt
• Use FastCLAClassifier read class method instead of instance method
• Have CLAClassifierFactory.read take just the proto object
• Add capnp serialization to CLAClassifierRegion
• Add capnp serialization to SPRegion

0.3.6

• Windows support
• Serialization work
• Moved SWIG out into nupic.core
• Major build changes

0.3.5

• Raise explicit exception if user passes non-str path
• SP: simplify local inhibition
• SP: adapt tests, sort winning columns output
• SP: simplify active columns assignment
• SP: simplify global inhibition
• file Rename as hello_tm.py and modifications in comments

0.3.4

• Added src/nupic/frameworks/opf/common_models/cluster_params.py and supporting files from numenta-apps htmengine. A separate numenta-apps PR will remove this code from htmengine.
• fixes #2592
• fix for #2265
• fix for bug #2265
• Fixup Dockerfile to install nupic.bindings, and other cleanup
• Adding C++ compiler requirement to README.
• Fix for test failure
• Fixed stream definition reference error.
• Reduce default reestimation period.
• Remove greedy reestimation of distribution
• Pointing README to proper bindings version.
• Continuing work on 0.3.4.dev0.
• removing a test that depends on nupic.vision
• PCA_Node test: some fixes, WIP
• formatting
• test for PCANode region
• remove Pillow from requirements.txt as it was used for vision only
• fix merge mistake in csv file
• move test from PCANode to nupic.vision unittest

0.3.3

• Include additional file types in MANIFEST.in, consistent with setup.py
• Pattern and Sequence machines using nupic::Random
• Wrap sparse matrix implementations with cortical column-centric semantics as a way to abstract away the underlying implementation
• Re-enable testHotgymRegression

0.3.2

• Update to nupic.bindings version with fix for platform differences
• Rename nupic directory to src/nupic
• Updated S3 URL to nupic.bindings for Linux install
• Fix paths for data files in an integration test
• Fix issue with storing temporary file in wrong location in integration test

0.3.1

• Specify nupic.bindings version to match commit sha (0.2).
• Use logging.debug for emitting the message about not being able to import matplotlib; we log it at debug level to avoid polluting the logs of apps and services that don't care about plotting.
• Add Dockerfile ready to perform swarming.
• Removes PCANode
• Updated Linux binary install instructions.

0.3.0

• Updated comment about greedy stats refresh when likelihood > 0.99

0.2.12

• Implemented unit tests for the new features in AnomalyLikelihood class.
• Convert AnomalyLikelihood._historicalScores to a user-configurable sliding window, instead of accumulating all of the incoming data points. This improved performance a ton! Added AnomalyLikelihood.forceModelRefresh() method.
• Update nupic.core to include backwards compatibility fix for RandomImpl.
• Uninstall pycapnp to avoid running tests that utilize the functionality and currently fail with Duplicate ID error.
• Makes pycapnp and corresponding serialization optional. If pycapnp is not installed then the corresponding serialization tests will be skipped.
• Add Multiple Prediction Test for NegLL Metric
• Add test for NegLL Error Metric
• Fix Orphan Decay Bug in temporal memory test
• Change decreasing overlaps test for coordinate encoder to not require a strict decrease (staying the same is ok).
• Allow specifying MonitoredTemporalMemory as TM implementation through OPF
• include bucket likelihood and classifier input in clamodel
• update metrics managers to pass model results to metrics
• introducting a computeFlag to prevent double-computation. * The flag is used to prevent double computation in the event that customCompute() is called at the same time as compute()
• Added numRecords param for consitency with the newly added infer method in FastCLACLassifier
• checking if classifier has a maxCategoryCount attribute. If not, set it to solve backward compatibilities issues
• renaming numCategories to maxCategoryCount to be constistent between KNN and CLA classifier
• made new experimentutils file containing InferenceElement, InferenceType, and ModelResult duplicates which we will want to change in the future