Hi,
I’ve compared LSTM to HTM (results are listed below, ignore the HTM MAPE column, lower is better across all statistics) and I got consistently better results using LSTM. How can I improve the performance across each increasing timesteps? Keep in mind that my LSTM code trained a separate network specifically for each step while HTM just used the built in multiStepBestPredictions feature.
My LSTM code is here: https://github.com/JonnoFTW/traffic-prediction/blob/master/main.py
My HTM evaluation code is here: https://github.com/JonnoFTW/htm-models-adelaide/blob/master/engine/evaluate.py
My Model Params are as follows:
{ 'aggregationInfo': { 'days': 0,
'fields': [],
'hours': 0,
'microseconds': 0,
'milliseconds': 0,
'minutes': 0,
'months': 0,
'seconds': 0,
'weeks': 0,
'years': 0},
'model': 'CLA',
'modelParams': { 'anomalyParams': { u'anomalyCacheRecords': None,
u'autoDetectThreshold': None,
u'autoDetectWaitRecords': None},
'clEnable': True,
'clParams': { 'alpha': 0.050050000000000004,
'clVerbosity': 0,
'regionName': 'CLAClassifierRegion',
'steps': '1,3,6,9,12'},
'inferenceType': 'TemporalAnomaly',
'sensorParams': { 'encoders': { 'downstream': { 'fieldname': 'downstream',
'name': 'downstream',
'resolution': 0.8,
'type': 'RandomDistributedScalarEncoder',
'w': 21},
'timestamp_dayOfWeek': { 'fieldname': 'timestamp',
'name': 'timestamp_dayOfWeek',
'timeOfDay': ( 51,
9.49),
'type': 'DateEncoder'},
'timestamp_timeOfDay': { 'fieldname': 'timestamp',
'name': 'timestamp_timeOfDay',
'timeOfDay': ( 51,
9.49),
'type': 'DateEncoder'},
'timestamp_weekend': { 'fieldname': 'timestamp',
'name': 'timestamp_weekend',
'type': 'DateEncoder',
'weekend': ( 51,
9)}},
'sensorAutoReset': None,
'verbosity': 0},
'spEnable': True,
'spParams': { 'columnCount': 2048,
'globalInhibition': 1,
'inputWidth': 0,
'maxBoost': 2.0,
'numActiveColumnsPerInhArea': 40,
'potentialPct': 0.8,
'seed': 1956,
'spVerbosity': 0,
'spatialImp': 'cpp',
'synPermActiveInc': 0.05,
'synPermConnected': 0.1,
'synPermInactiveDec': 0.05015},
'tpEnable': True,
'tpParams': { 'activationThreshold': 14,
'cellsPerColumn': 32,
'columnCount': 2048,
'globalDecay': 0.0,
'initialPerm': 0.21,
'inputWidth': 2048,
'maxAge': 0,
'maxSegmentsPerCell': 128,
'maxSynapsesPerSegment': 32,
'minThreshold': 11,
'newSynapseCount': 20,
'outputType': 'normal',
'pamLength': 3,
'permanenceDec': 0.1,
'permanenceInc': 0.1,
'seed': 1960,
'temporalImp': 'cpp',
'verbosity': 0},
'trainSPNetOnlyIfRequested': False},
'predictAheadTime': None,
'version': 1}