|
DaNNet
|
Classes | |
| class | act_LReLU |
| Leaky ReLU activation class. More... | |
| class | act_ReLU |
| ReLU activation class. More... | |
| class | act_sigmoid |
| Sigmoid activation class. More... | |
| class | act_softmax |
| Softmax activation class. More... | |
| class | act_softplus |
| Softplus activation class. More... | |
| class | act_tanh |
| tanh activation class More... | |
| class | cost_CE |
| Cross Entropy cost/output layer class with linear activation. More... | |
| class | cost_CE_sigmoid |
| Cross Entropy cost/output layer class with sigmoid activation. More... | |
| class | cost_CE_softmax |
| Cross Entropy cost/output layer class with softmax activation. More... | |
| class | cost_MSE |
| Mean Square Error cost/output layer class. More... | |
| class | cost_MSE_sigmoid |
| Mean Square Error cost/output layer class with sigmoid activation. More... | |
| class | layer |
| Layer base class. More... | |
| class | layer_act |
| Activation layer base class. More... | |
| class | layer_conv |
| A convolution layer class. More... | |
| class | layer_cost |
| Cost/output layer base class. More... | |
| class | layer_dense |
| A fully connected/dense layer class. More... | |
| class | layer_drop |
| Dropout layer class. More... | |
| class | layer_input |
| Input/data layer class. More... | |
| class | layer_norm |
| Batch normalization layer class. More... | |
| class | layer_pool |
| Pooling layer base class. More... | |
| class | net |
| Neural netowork model class. More... | |
| class | opt |
| Optimizer base class. More... | |
| class | opt_adadelta |
| ADAdelta optimizer class. More... | |
| class | opt_adagrad |
| ADAgrad optimizer class. More... | |
| class | opt_adam |
| ADAM optimizer class. More... | |
| class | opt_adamax |
| ADAMax optimizer class. More... | |
| class | opt_rmsprop |
| RMSprop optimizer class. More... | |
| class | opt_SGD |
| Stochastic Gradient Descent optimizer class. More... | |
| class | opt_SGD_momentum |
| Stochastic Gradient Descent with momentum optimizer class. More... | |
| class | opt_SGD_nesterov |
| Stochastic Gradient Descent with Nesterov momentum optimizer class. More... | |
| class | pool_average |
| Average pooling layer class. More... | |
| class | pool_max |
| Max pooling layer class. More... | |
Typedefs | |
| typedef float | DNN_Dtype |
| Data type used in the network (float or double) More... | |
Enumerations | |
| enum | PARAM_FORMAT { PARAM_FORMAT::AUTO, PARAM_FORMAT::ASCII, PARAM_FORMAT::BIN, PARAM_FORMAT::HDF5 } |
| enum | INIT_W_ALG { INIT_W_ALG::LECUN, INIT_W_ALG::XAVIER, INIT_W_ALG::HE, INIT_W_ALG::SIN } |
| enum | INIT_W_DIST { INIT_W_DIST::NORMAL, INIT_W_DIST::UNIFORM, INIT_W_DIST::NONE } |
| enum | PHASE { PHASE::TRAIN, PHASE::TEST, PHASE::PRED } |
| enum | LR_ALG { LR_ALG::CONST, LR_ALG::TIME_DECAY, LR_ALG::STEP_DECAY, LR_ALG::EXP_DECAY } |
Functions | |
| std::string | version_info (void) |
| Generate DaNNet version string. More... | |
| template<typename T > | |
| void | pol2cart (const T phi, const T r, T &x, T &y) |
| Polar to cartesian convertion. More... | |
| template<typename T > | |
| void | pol2cart (const arma::Mat< T > phi, const arma::Mat< T > r, arma::Mat< T > &x, arma::Mat< T > &y) |
| Polar to cartesian convertion (matrix) More... | |
| template<typename T > | |
| void | cart2pol (const T x, const T y, T &phi, T &r) |
| Cartesian to polar convertion. More... | |
| template<typename T > | |
| void | cart2pol (const arma::Mat< T > x, const arma::Mat< T > y, arma::Mat< T > &phi, arma::Mat< T > &r) |
| Cartesian to polar convertion (matrix) More... | |
| arma_inline arma::Mat< DNN_Dtype > | row2col (const arma::Mat< DNN_Dtype > &A) |
| Flattens a matrix row-wise to a one column matrix. More... | |
| arma_inline arma::Mat< DNN_Dtype > | col2col (const arma::Mat< DNN_Dtype > &A) |
| Flattens a matrix col-wise to a one column matrix. More... | |
| arma_inline arma::Mat< DNN_Dtype > | mat2mat (const arma::Mat< DNN_Dtype > &A, const arma::uword rows, const arma::uword cols) |
| Reshapes a matrix. More... | |
| arma_inline arma::Cube< DNN_Dtype > | mat2cube (const arma::Mat< DNN_Dtype > &A, const arma::uword rows, const arma::uword cols, const arma::uword slices) |
| Converts a matrix to a cube. More... | |
| arma_inline arma::Mat< DNN_Dtype > | cube2mat (arma::Cube< DNN_Dtype > &A, const arma::uword rows, const arma::uword cols) |
| Converts a cube to a matrix. More... | |
| void | gen_spiral (arma::Mat< DNN_Dtype > &X, arma::Mat< DNN_Dtype > &T, arma::uword N, arma::uword K, const double ph_var=0.1, const double r_min=0.1, const double ph_max=7) |
| Dataset generator: a rotating spiral with K classes. More... | |
| uint32_t | swap_endian (uint32_t i) |
| Endian converter for IDX1 and IDX3 file. More... | |
| bool | read_idx3 (std::string fname, arma::Cube< DNN_Dtype > &img, const double frac=1.0) |
| Reads a IDX3 file. More... | |
| bool | read_idx1 (std::string fname, arma::ivec &label, const double frac=1.0) |
| Reads a IDX1 file. More... | |
| void | read_MNIST (std::string fname_X, std::string fname_T, arma::Mat< DNN_Dtype > &X, arma::Mat< DNN_Dtype > &T, double frac=1.0) |
| Reads a MNIST digit dataset. More... | |
| void | progress_bar (const std::string str, double p) |
| Console progress bar. More... | |
| void | progress_bar (const std::string str, arma::uword num, arma::uword maxnum) |
| Console progress bar. More... | |
| void | remove_progress_bar (void) |
| Clears progress bar. More... | |
| typedef float dnn::DNN_Dtype |
|
strong |
|
strong |
|
strong |
|
strong |
|
strong |
1.8.13