101 Curie Nueral Network - whichversion of CuriePME.h and .cpp should I be using

Official Hardware Retired 101
1

I've got 2 versions of the neural network files, CuriePME.h and CuriePME.cpp. They have no version which is ridiculous, so which should I be using ?

and why do they have the same name as the general vision files, and they are almost identical ?

The headers are different, in that one has a

static const uint32_t noMatch = 0x7fff;

And one doesnt. Also is this really the correct value that comes back when there is not match from a classify ?

The code in one function looks like this :
uint16_t Intel_PMT::learn(uint8_t *pattern_vector, int32_t vector_length, uint16_t category)
{
if( vector_length > MaxVectorSize )
vector_length = MaxVectorSize;

for( int i = 0; i < vector_length -1; i++ )
{
regWrite16( COMP , pattern_vector );

  • }*
  • regWrite16( LCOMP, pattern_vector[ vector_length - 1 ] );*
    _ /* Mask off the 15th bit-- valid categories range from 1-32766,_
    _ * and bit 15 is used to indicate if a firing neuron has degenerated */_
  • regWrite16(CAT, (regRead16(CAT) & ~CAT_CATEGORY) | (category & CAT_CATEGORY));*
  • return regRead16( FORGET_NCOUNT );*
    }
    So is this the correct cpp ?It seems to be the one you pointed me at recently. The other file I've got looks like this :
    uint16_t Intel_PMT::learn(uint8_t *pattern_vector, int32_t vector_length, uint8_t category)
    {
  • if( vector_length > MaxVectorSize )*
  • vector_length = MaxVectorSize;*
  • for( int i = 0; i < vector_length -1; i++ )*
  • {*
    regWrite16( COMP , pattern_vector );
    * }*
    * regWrite16( LCOMP, pattern_vector[ vector_length - 1 ] );
    _     regWrite16( CAT, category );_
    return regRead16( FORGET_NCOUNT );
    _    }*_
2

The official Intel libraries are here: