Classes | |
class | dcnnasgn::loss_data |
Loss data class. More... | |
class | dcnnsol::image_data< SP, permutation_policy > |
Forward-propagated data between layers. More... | |
class | dcnnsol::feature_data< CSP, permutation_policy > |
Output of the final Aggregate layer. More... | |
class | dcnnsol::conv_weights< KSP, CSPI, CSPO, permutation_policy > |
Convolution or Aggregate layer model. More... | |
class | dcnnsol::feature_weights< CSPI, CSPO, permutation_policy > |
Fully connected layer model. More... | |
class | dcnnsol::feature_multiplier< CSP, permutation_policy > |
Multiplier layer model. More... | |
class | dcnnsol::feature_bias< CSP, permutation_policy > |
Shift layer model. More... | |
These classes form the interfaces between layers
Each of the classes contains one or more instances of tagged::tensor_class
Additional data members may be inserted into these classes if required for optimization
Note: The additional data members will typically be redundant versions of the data already contained, e.g. their transposed versions
Temporary data shall be inserted as local variables into the respective functions