Relintai/pmlpp
1
0
Fork 0
mirror of https://github.com/Relintai/pmlpp.git synced 2024-11-13 13:57:19 +01:00
Code Issues Packages Projects Releases Wiki Activity

Now MLPPMANN uses engine classes.

Browse Source
This commit is contained in:
Relintai 2023-02-17 18:46:27 +01:00
parent 1224116e12
commit f30e3a887d
5 changed files with 149 additions and 112 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

199
mlpp/mann/mann.cpp
View File

@ -5,14 +5,15 @@
// //
#include "mann.h" #include "mann.h"
#include "core/log/logger.h"
#include "../activation/activation.h" #include "../activation/activation.h"
#include "../cost/cost.h" #include "../cost/cost.h"
#include "../lin_alg/lin_alg.h" #include "../lin_alg/lin_alg.h"
#include "../regularization/reg.h" #include "../regularization/reg.h"
#include "../utilities/utilities.h" #include "../utilities/utilities.h"
#include <iostream>
/* /*
Ref<MLPPMatrix> MLPPMANN::get_input_set() { Ref<MLPPMatrix> MLPPMANN::get_input_set() {
return input_set; return input_set;
@ -33,40 +34,54 @@ void MLPPMANN::set_output_set(const Ref<MLPPMatrix> &val) {
} }
*/ */
std::vector<std::vector<real_t>> MLPPMANN::model_set_test(std::vector<std::vector<real_t>> X) { Ref<MLPPMatrix> MLPPMANN::model_set_test(const Ref<MLPPMatrix> &X) {
ERR_FAIL_COND_V(!_initialized, std::vector<std::vector<real_t>>()); ERR_FAIL_COND_V(!_initialized, Ref<MLPPMatrix>());
if (!_network.empty()) { if (!_network.empty()) {
_network[0].input = X; Ref<MLPPHiddenLayer> layer = _network[0];
_network[0].forwardPass();
for (uint32_t i = 1; i < _network.size(); i++) { layer->set_input(X);
_network[i].input = _network[i - 1].a; layer->forward_pass();
_network[i].forwardPass();
for (int i = 1; i < _network.size(); i++) {
layer = _network[i];
Ref<MLPPHiddenLayer> prev_layer = _network[i - 1];
layer->set_input(prev_layer->get_a());
layer->forward_pass();
} }
_output_layer->input = _network[_network.size() - 1].a;
_output_layer->set_input(_network.write[_network.size() - 1]->get_a());
} else { } else {
_output_layer->input = X; _output_layer->set_input(X);
} }
_output_layer->forwardPass(); _output_layer->forward_pass();
return _output_layer->a; return _output_layer->get_a();
} }
std::vector<real_t> MLPPMANN::model_test(std::vector<real_t> x) { Ref<MLPPVector> MLPPMANN::model_test(const Ref<MLPPVector> &x) {
ERR_FAIL_COND_V(!_initialized, std::vector<real_t>()); ERR_FAIL_COND_V(!_initialized, Ref<MLPPVector>());
if (!_network.empty()) { if (!_network.empty()) {
_network[0].Test(x); Ref<MLPPHiddenLayer> layer = _network[0];
for (uint32_t i = 1; i < _network.size(); i++) {
_network[i].Test(_network[i - 1].a_test); layer->test(x);
for (int i = 1; i < _network.size(); i++) {
layer = _network[i];
Ref<MLPPHiddenLayer> prev_layer = _network[i - 1];
layer->test(prev_layer->get_a_test());
} }
_output_layer->Test(_network[_network.size() - 1].a_test);
_output_layer->test(_network.write[_network.size() - 1]->get_a_test());
} else { } else {
_output_layer->Test(x); _output_layer->test(x);
} }
return _output_layer->a_test;
return _output_layer->get_a_test();
} }
void MLPPMANN::gradient_descent(real_t learning_rate, int max_epoch, bool ui) { void MLPPMANN::gradient_descent(real_t learning_rate, int max_epoch, bool ui) {
@ -85,50 +100,57 @@ void MLPPMANN::gradient_descent(real_t learning_rate, int max_epoch, bool ui) {
while (true) { while (true) {
cost_prev = cost(_y_hat, _output_set); cost_prev = cost(_y_hat, _output_set);
if (_output_layer->activation == "Softmax") { if (_output_layer->get_activation() == MLPPActivation::ACTIVATION_FUNCTION_SOFTMAX) {
_output_layer->delta = alg.subtraction(_y_hat, _output_set); _output_layer->set_delta(alg.subtractionm(_y_hat, _output_set));
} else { } else {
auto costDeriv = _output_layer->costDeriv_map[_output_layer->cost]; _output_layer->set_delta(alg.hadamard_productm(mlpp_cost.run_cost_deriv_matrix(_output_layer->get_cost(), _y_hat, _output_set), avn.run_activation_deriv_matrix(_output_layer->get_activation(), _output_layer->get_z())));
auto outputAvn = _output_layer->activation_map[_output_layer->activation];
_output_layer->delta = alg.hadamard_product((mlpp_cost.*costDeriv)(_y_hat, _output_set), (avn.*outputAvn)(_output_layer->z, 1));
} }
std::vector<std::vector<real_t>> outputWGrad = alg.matmult(alg.transpose(_output_layer->input), _output_layer->delta); Ref<MLPPMatrix> output_w_grad = alg.matmultm(alg.transposem(_output_layer->get_input()), _output_layer->get_delta());
_output_layer->weights = alg.subtraction(_output_layer->weights, alg.scalarMultiply(learning_rate / _n, outputWGrad)); _output_layer->set_weights(alg.subtractionm(_output_layer->get_weights(), alg.scalar_multiplym(learning_rate / _n, output_w_grad)));
_output_layer->weights = regularization.regWeights(_output_layer->weights, _output_layer->lambda, _output_layer->alpha, _output_layer->reg); _output_layer->set_weights(regularization.reg_weightsm(_output_layer->get_weights(), _output_layer->get_lambda(), _output_layer->get_alpha(), _output_layer->get_reg()));
_output_layer->bias = alg.subtractMatrixRows(_output_layer->bias, alg.scalarMultiply(learning_rate / _n, _output_layer->delta)); _output_layer->set_bias(alg.subtract_matrix_rows(_output_layer->get_bias(), alg.scalar_multiplym(learning_rate / _n, _output_layer->get_delta())));
if (!_network.empty()) { if (!_network.empty()) {
auto hiddenLayerAvn = _network[_network.size() - 1].activation_map[_network[_network.size() - 1].activation]; Ref<MLPPHiddenLayer> layer = _network[_network.size() - 1];
_network[_network.size() - 1].delta = alg.hadamard_product(alg.matmult(_output_layer->delta, alg.transpose(_output_layer->weights)), (avn.*hiddenLayerAvn)(_network[_network.size() - 1].z, true));
std::vector<std::vector<real_t>> hiddenLayerWGrad = alg.matmult(alg.transpose(_network[_network.size() - 1].input), _network[_network.size() - 1].delta);
_network[_network.size() - 1].weights = alg.subtraction(_network[_network.size() - 1].weights, alg.scalarMultiply(learning_rate / _n, hiddenLayerWGrad)); //auto hiddenLayerAvn = layer.activation_map[layer.activation];
_network[_network.size() - 1].weights = regularization.regWeights(_network[_network.size() - 1].weights, _network[_network.size() - 1].lambda, _network[_network.size() - 1].alpha, _network[_network.size() - 1].reg);
_network[_network.size() - 1].bias = alg.subtractMatrixRows(_network[_network.size() - 1].bias, alg.scalarMultiply(learning_rate / _n, _network[_network.size() - 1].delta)); layer->set_delta(alg.hadamard_productm(alg.matmultm(_output_layer->get_delta(), alg.transposem(_output_layer->get_weights())), avn.run_activation_deriv_matrix(layer->get_activation(), layer->get_z())));
Ref<MLPPMatrix> hidden_layer_w_grad = alg.matmultm(alg.transposem(layer->get_input()), layer->get_delta());
layer->set_weights(alg.subtractionm(layer->get_weights(), alg.scalar_multiplym(learning_rate / _n, hidden_layer_w_grad)));
layer->set_weights(regularization.reg_weightsm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg()));
layer->set_bias(alg.subtract_matrix_rows(layer->get_bias(), alg.scalar_multiplym(learning_rate / _n, layer->get_delta())));
for (int i = _network.size() - 2; i >= 0; i--) { for (int i = _network.size() - 2; i >= 0; i--) {
hiddenLayerAvn = _network[i].activation_map[_network[i].activation]; layer = _network[i];
_network[i].delta = alg.hadamard_product(alg.matmult(_network[i + 1].delta, _network[i + 1].weights), (avn.*hiddenLayerAvn)(_network[i].z, true)); Ref<MLPPHiddenLayer> next_layer = _network[i + 1];
hiddenLayerWGrad = alg.matmult(alg.transpose(_network[i].input), _network[i].delta);
_network[i].weights = alg.subtraction(_network[i].weights, alg.scalarMultiply(learning_rate / _n, hiddenLayerWGrad)); //hiddenLayerAvn = layer.activation_map[layer.activation];
_network[i].weights = regularization.regWeights(_network[i].weights, _network[i].lambda, _network[i].alpha, _network[i].reg); layer->set_delta(alg.hadamard_productm(alg.matmultm(next_layer->get_delta(), next_layer->get_weights()), avn.run_activation_deriv_matrix(layer->get_activation(), layer->get_z())));
_network[i].bias = alg.subtractMatrixRows(_network[i].bias, alg.scalarMultiply(learning_rate / _n, _network[i].delta)); hidden_layer_w_grad = alg.matmultm(alg.transposem(layer->get_input()), layer->get_delta());
layer->set_weights(alg.subtractionm(layer->get_weights(), alg.scalar_multiplym(learning_rate / _n, hidden_layer_w_grad)));
layer->set_weights(regularization.reg_weightsm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg()));
layer->set_bias(alg.subtract_matrix_rows(layer->get_bias(), alg.scalar_multiplym(learning_rate / _n, layer->get_delta())));
} }
} }
forward_pass(); forward_pass();
if (ui) { if (ui) {
MLPPUtilities::CostInfo(epoch, cost_prev, cost(_y_hat, _output_set)); MLPPUtilities::cost_info(epoch, cost_prev, cost(_y_hat, _output_set));
std::cout << "Layer " << _network.size() + 1 << ": " << std::endl; PLOG_MSG("Layer " + itos(_network.size() + 1) + ": ");
MLPPUtilities::UI(_output_layer->weights, _output_layer->bias); MLPPUtilities::print_ui_mb(_output_layer->get_weights(), _output_layer->get_bias());
if (!_network.empty()) { if (!_network.empty()) {
std::cout << "Layer " << _network.size() << ": " << std::endl;
for (int i = _network.size() - 1; i >= 0; i--) { for (int i = _network.size() - 1; i >= 0; i--) {
std::cout << "Layer " << i + 1 << ": " << std::endl; PLOG_MSG("Layer " + itos(i + 1) + ": ");
MLPPUtilities::UI(_network[i].weights, _network[i].bias);
Ref<MLPPHiddenLayer> layer = _network[i];
MLPPUtilities::print_ui_mb(layer->get_weights(), layer->get_bias());
} }
} }
} }
@ -148,39 +170,41 @@ real_t MLPPMANN::score() {
forward_pass(); forward_pass();
return util.performance(_y_hat, _output_set); return util.performance_mat(_y_hat, _output_set);
} }
void MLPPMANN::save(std::string fileName) { void MLPPMANN::save(const String &file_name) {
ERR_FAIL_COND(!_initialized); ERR_FAIL_COND(!_initialized);
/*
MLPPUtilities util; MLPPUtilities util;
if (!_network.empty()) { if (!_network.empty()) {
util.saveParameters(fileName, _network[0].weights, _network[0].bias, false, 1); util.saveParameters(file_name, _network[0].weights, _network[0].bias, false, 1);
for (uint32_t i = 1; i < _network.size(); i++) { for (uint32_t i = 1; i < _network.size(); i++) {
util.saveParameters(fileName, _network[i].weights, _network[i].bias, true, i + 1); util.saveParameters(file_name, _network[i].weights, _network[i].bias, true, i + 1);
} }
util.saveParameters(fileName, _output_layer->weights, _output_layer->bias, true, _network.size() + 1); util.saveParameters(file_name, _output_layer->weights, _output_layer->bias, true, _network.size() + 1);
} else { } else {
util.saveParameters(fileName, _output_layer->weights, _output_layer->bias, false, _network.size() + 1); util.saveParameters(file_name, _output_layer->weights, _output_layer->bias, false, _network.size() + 1);
} }
*/
} }
void MLPPMANN::add_layer(int n_hidden, std::string activation, std::string weightInit, std::string reg, real_t lambda, real_t alpha) { void MLPPMANN::add_layer(int n_hidden, MLPPActivation::ActivationFunction activation, MLPPUtilities::WeightDistributionType weight_init, MLPPReg::RegularizationType reg, real_t lambda, real_t alpha) {
if (_network.empty()) { if (_network.empty()) {
_network.push_back(MLPPOldHiddenLayer(n_hidden, activation, _input_set, weightInit, reg, lambda, alpha)); _network.push_back(Ref<MLPPHiddenLayer>(memnew(MLPPHiddenLayer(n_hidden, activation, _input_set, weight_init, reg, lambda, alpha))));
_network[0].forwardPass(); _network.write[0]->forward_pass();
} else { } else {
_network.push_back(MLPPOldHiddenLayer(n_hidden, activation, _network[_network.size() - 1].a, weightInit, reg, lambda, alpha)); _network.push_back(Ref<MLPPHiddenLayer>(memnew(MLPPHiddenLayer(n_hidden, activation, _network.write[_network.size() - 1]->get_a(), weight_init, reg, lambda, alpha))));
_network[_network.size() - 1].forwardPass(); _network.write[_network.size() - 1]->forward_pass();
} }
} }
void MLPPMANN::add_output_layer(std::string activation, std::string loss, std::string weightInit, std::string reg, real_t lambda, real_t alpha) { void MLPPMANN::add_output_layer(MLPPActivation::ActivationFunction activation, MLPPCost::CostTypes loss, MLPPUtilities::WeightDistributionType weight_init, MLPPReg::RegularizationType reg, real_t lambda, real_t alpha) {
if (!_network.empty()) { if (!_network.empty()) {
_output_layer = new MLPPOldMultiOutputLayer(_n_output, _network[0].n_hidden, activation, loss, _network[_network.size() - 1].a, weightInit, reg, lambda, alpha); _output_layer = Ref<MLPPMultiOutputLayer>(memnew(MLPPMultiOutputLayer(_n_output, _network.write[_network.size() - 1]->get_n_hidden(), activation, loss, _network.write[_network.size() - 1]->get_a(), weight_init, reg, lambda, alpha)));
} else { } else {
_output_layer = new MLPPOldMultiOutputLayer(_n_output, _k, activation, loss, _input_set, weightInit, reg, lambda, alpha); _output_layer = Ref<MLPPMultiOutputLayer>(memnew(MLPPMultiOutputLayer(_n_output, _k, activation, loss, _input_set, weight_init, reg, lambda, alpha)));
} }
} }
@ -198,12 +222,12 @@ void MLPPMANN::initialize() {
_initialized = true; _initialized = true;
} }
MLPPMANN::MLPPMANN(std::vector<std::vector<real_t>> p_input_set, std::vector<std::vector<real_t>> p_output_set) { MLPPMANN::MLPPMANN(const Ref<MLPPMatrix> &p_input_set, const Ref<MLPPMatrix> &p_output_set) {
_input_set = p_input_set; _input_set = p_input_set;
_output_set = p_output_set; _output_set = p_output_set;
_n = _input_set.size(); _n = _input_set->size().y;
_k = _input_set[0].size(); _k = _input_set->size().x;
_n_output = _output_set[0].size(); _n_output = _output_set->size().x;
_initialized = true; _initialized = true;
} }
@ -213,39 +237,48 @@ MLPPMANN::MLPPMANN() {
} }
MLPPMANN::~MLPPMANN() { MLPPMANN::~MLPPMANN() {
delete _output_layer;
} }
real_t MLPPMANN::cost(std::vector<std::vector<real_t>> y_hat, std::vector<std::vector<real_t>> y) { real_t MLPPMANN::cost(const Ref<MLPPMatrix> &y_hat, const Ref<MLPPMatrix> &y) {
MLPPReg regularization; MLPPReg regularization;
class MLPPCost cost; MLPPCost mlpp_cost;
real_t totalRegTerm = 0;
real_t total_reg_term = 0;
auto cost_function = _output_layer->cost_map[_output_layer->cost];
if (!_network.empty()) { if (!_network.empty()) {
for (uint32_t i = 0; i < _network.size() - 1; i++) { for (int i = 0; i < _network.size() - 1; i++) {
totalRegTerm += regularization.regTerm(_network[i].weights, _network[i].lambda, _network[i].alpha, _network[i].reg); Ref<MLPPHiddenLayer> layer = _network[i];
total_reg_term += regularization.reg_termm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg());
} }
} }
return (cost.*cost_function)(y_hat, y) + totalRegTerm + regularization.regTerm(_output_layer->weights, _output_layer->lambda, _output_layer->alpha, _output_layer->reg);
return mlpp_cost.run_cost_norm_matrix(_output_layer->get_cost(), y_hat, y) + total_reg_term + regularization.reg_termm(_output_layer->get_weights(), _output_layer->get_lambda(), _output_layer->get_alpha(), _output_layer->get_reg());
} }
void MLPPMANN::forward_pass() { void MLPPMANN::forward_pass() {
if (!_network.empty()) { if (!_network.empty()) {
_network[0].input = _input_set; Ref<MLPPHiddenLayer> layer = _network[0];
_network[0].forwardPass();
for (uint32_t i = 1; i < _network.size(); i++) { layer->set_input(_input_set);
_network[i].input = _network[i - 1].a; layer->forward_pass();
_network[i].forwardPass();
for (int i = 1; i < _network.size(); i++) {
layer = _network[i];
Ref<MLPPHiddenLayer> prev_layer = _network[i - 1];
layer->set_input(prev_layer->get_a());
layer->forward_pass();
} }
_output_layer->input = _network[_network.size() - 1].a;
_output_layer->set_input(_network.write[_network.size() - 1]->get_a());
} else { } else {
_output_layer->input = _input_set; _output_layer->set_input(_input_set);
} }
_output_layer->forwardPass(); _output_layer->forward_pass();
_y_hat = _output_layer->a;
_y_hat = _output_layer->get_a();
} }
void MLPPMANN::_bind_methods() { void MLPPMANN::_bind_methods() {

30
mlpp/mann/mann.h
View File

@ -20,12 +20,6 @@
#include "../hidden_layer/hidden_layer.h" #include "../hidden_layer/hidden_layer.h"
#include "../multi_output_layer/multi_output_layer.h" #include "../multi_output_layer/multi_output_layer.h"
#include "../hidden_layer/hidden_layer_old.h"
#include "../multi_output_layer/multi_output_layer_old.h"
#include <string>
#include <vector>
class MLPPMANN : public Reference { class MLPPMANN : public Reference {
GDCLASS(MLPPMANN, Reference); GDCLASS(MLPPMANN, Reference);
@ -38,38 +32,38 @@ public:
void set_output_set(const Ref<MLPPMatrix> &val); void set_output_set(const Ref<MLPPMatrix> &val);
*/ */
std::vector<std::vector<real_t>> model_set_test(std::vector<std::vector<real_t>> X); Ref<MLPPMatrix> model_set_test(const Ref<MLPPMatrix> &X);
std::vector<real_t> model_test(std::vector<real_t> x); Ref<MLPPVector> model_test(const Ref<MLPPVector> &x);
void gradient_descent(real_t learning_rate, int max_epoch, bool ui = false); void gradient_descent(real_t learning_rate, int max_epoch, bool ui = false);
real_t score(); real_t score();
void save(std::string file_name); void save(const String &file_name);
void add_layer(int n_hidden, std::string activation, std::string weight_init = "Default", std::string reg = "None", real_t lambda = 0.5, real_t alpha = 0.5); void add_layer(int n_hidden, MLPPActivation::ActivationFunction activation, MLPPUtilities::WeightDistributionType weight_init = MLPPUtilities::WEIGHT_DISTRIBUTION_TYPE_DEFAULT, MLPPReg::RegularizationType reg = MLPPReg::REGULARIZATION_TYPE_NONE, real_t lambda = 0.5, real_t alpha = 0.5);
void add_output_layer(std::string activation, std::string loss, std::string weight_init = "Default", std::string reg = "None", real_t lambda = 0.5, real_t alpha = 0.5); void add_output_layer(MLPPActivation::ActivationFunction activation, MLPPCost::CostTypes loss, MLPPUtilities::WeightDistributionType weight_init = MLPPUtilities::WEIGHT_DISTRIBUTION_TYPE_DEFAULT, MLPPReg::RegularizationType reg = MLPPReg::REGULARIZATION_TYPE_NONE, real_t lambda = 0.5, real_t alpha = 0.5);
bool is_initialized(); bool is_initialized();
void initialize(); void initialize();
MLPPMANN(std::vector<std::vector<real_t>> p_input_set, std::vector<std::vector<real_t>> p_output_set); MLPPMANN(const Ref<MLPPMatrix> &p_input_set, const Ref<MLPPMatrix> &p_output_set);
MLPPMANN(); MLPPMANN();
~MLPPMANN(); ~MLPPMANN();
private: private:
real_t cost(std::vector<std::vector<real_t>> y_hat, std::vector<std::vector<real_t>> y); real_t cost(const Ref<MLPPMatrix> &y_hat, const Ref<MLPPMatrix> &y);
void forward_pass(); void forward_pass();
static void _bind_methods(); static void _bind_methods();
std::vector<std::vector<real_t>> _input_set; Ref<MLPPMatrix> _input_set;
std::vector<std::vector<real_t>> _output_set; Ref<MLPPMatrix> _output_set;
std::vector<std::vector<real_t>> _y_hat; Ref<MLPPMatrix> _y_hat;
std::vector<MLPPOldHiddenLayer> _network; Vector<Ref<MLPPHiddenLayer>> _network;
MLPPOldMultiOutputLayer *_output_layer; Ref<MLPPMultiOutputLayer> _output_layer;
int _n; int _n;
int _k; int _k;

4
mlpp/multi_output_layer/multi_output_layer.cpp
View File

@ -136,9 +136,11 @@ void MLPPMultiOutputLayer::test(const Ref<MLPPVector> &x) {
_a_test = avn.run_activation_norm_vector(_activation, _z_test); _a_test = avn.run_activation_norm_vector(_activation, _z_test);
} }
MLPPMultiOutputLayer::MLPPMultiOutputLayer(int p_n_hidden, MLPPActivation::ActivationFunction p_activation, Ref<MLPPMatrix> p_input, MLPPUtilities::WeightDistributionType p_weight_init, MLPPReg::RegularizationType p_reg, real_t p_lambda, real_t p_alpha) { MLPPMultiOutputLayer::MLPPMultiOutputLayer(int n_output, int p_n_hidden, MLPPActivation::ActivationFunction p_activation, MLPPCost::CostTypes cost, Ref<MLPPMatrix> p_input, MLPPUtilities::WeightDistributionType p_weight_init, MLPPReg::RegularizationType p_reg, real_t p_lambda, real_t p_alpha) {
_n_output = n_output;
_n_hidden = p_n_hidden; _n_hidden = p_n_hidden;
_activation = p_activation; _activation = p_activation;
_cost = cost;
_input = p_input; _input = p_input;

2
mlpp/multi_output_layer/multi_output_layer.h
View File

@ -76,7 +76,7 @@ public:
void forward_pass(); void forward_pass();
void test(const Ref<MLPPVector> &x); void test(const Ref<MLPPVector> &x);
MLPPMultiOutputLayer(int p_n_hidden, MLPPActivation::ActivationFunction p_activation, Ref<MLPPMatrix> p_input, MLPPUtilities::WeightDistributionType p_weight_init, MLPPReg::RegularizationType p_reg, real_t p_lambda, real_t p_alpha); MLPPMultiOutputLayer(int n_output, int p_n_hidden, MLPPActivation::ActivationFunction p_activation, MLPPCost::CostTypes cost, Ref<MLPPMatrix> p_input, MLPPUtilities::WeightDistributionType p_weight_init, MLPPReg::RegularizationType p_reg, real_t p_lambda, real_t p_alpha);
MLPPMultiOutputLayer(); MLPPMultiOutputLayer();
~MLPPMultiOutputLayer(); ~MLPPMultiOutputLayer();

26
test/mlpp_tests.cpp
View File

@ -745,11 +745,19 @@ void MLPPTests::test_dynamically_sized_mann(bool ui) {
alg.printMatrix(mann_old.modelSetTest(inputSet)); alg.printMatrix(mann_old.modelSetTest(inputSet));
std::cout << "ACCURACY (old): " << 100 * mann_old.score() << "%" << std::endl; std::cout << "ACCURACY (old): " << 100 * mann_old.score() << "%" << std::endl;
MLPPMANN mann(inputSet, outputSet); Ref<MLPPMatrix> input_set;
mann.add_output_layer("Linear", "MSE"); input_set.instance();
input_set->set_from_std_vectors(inputSet);
Ref<MLPPMatrix> output_set;
output_set.instance();
output_set->set_from_std_vectors(outputSet);
MLPPMANN mann(input_set, output_set);
mann.add_output_layer(MLPPActivation::ACTIVATION_FUNCTION_LINEAR, MLPPCost::COST_TYPE_MSE);
mann.gradient_descent(0.001, 80000, false); mann.gradient_descent(0.001, 80000, false);
alg.printMatrix(mann.model_set_test(inputSet)); PLOG_MSG(mann.model_set_test(input_set)->to_string());
std::cout << "ACCURACY: " << 100 * mann.score() << "%" << std::endl; PLOG_MSG("ACCURACY: " + String::num(100 * mann.score()) + "%");
} }
void MLPPTests::test_train_test_split_mann(bool ui) { void MLPPTests::test_train_test_split_mann(bool ui) {
MLPPLinAlg alg; MLPPLinAlg alg;
@ -787,12 +795,12 @@ void MLPPTests::test_train_test_split_mann(bool ui) {
alg.printMatrix(mann_old.modelSetTest(split_data.test->get_input()->to_std_vector())); alg.printMatrix(mann_old.modelSetTest(split_data.test->get_input()->to_std_vector()));
std::cout << "ACCURACY (old): " << 100 * mann_old.score() << "%" << std::endl; std::cout << "ACCURACY (old): " << 100 * mann_old.score() << "%" << std::endl;
MLPPMANN mann(split_data.train->get_input()->to_std_vector(), split_data.train->get_output()->to_std_vector()); MLPPMANN mann(split_data.train->get_input(), split_data.train->get_output());
mann.add_layer(100, "RELU", "XavierNormal"); mann.add_layer(100, MLPPActivation::ACTIVATION_FUNCTION_RELU, MLPPUtilities::WEIGHT_DISTRIBUTION_TYPE_XAVIER_NORMAL);
mann.add_output_layer("Softmax", "CrossEntropy", "XavierNormal"); mann.add_output_layer(MLPPActivation::ACTIVATION_FUNCTION_SOFTMAX, MLPPCost::COST_TYPE_CROSS_ENTROPY, MLPPUtilities::WEIGHT_DISTRIBUTION_TYPE_XAVIER_NORMAL);
mann.gradient_descent(0.1, 80000, ui); mann.gradient_descent(0.1, 80000, ui);
alg.printMatrix(mann.model_set_test(split_data.test->get_input()->to_std_vector())); PLOG_MSG(mann.model_set_test(split_data.test->get_input())->to_string());
std::cout << "ACCURACY: " << 100 * mann.score() << "%" << std::endl; PLOG_MSG("ACCURACY: " + String::num(100 * mann.score()) + "%");
} }
void MLPPTests::test_naive_bayes() { void MLPPTests::test_naive_bayes() {