pmlpp/mann/mann.cpp

327 lines
12 KiB
C++
Raw Normal View History

2023-12-30 00:41:59 +01:00
/*************************************************************************/
/* mann.cpp */
/*************************************************************************/
/* This file is part of: */
/* PMLPP Machine Learning Library */
/* https://github.com/Relintai/pmlpp */
/*************************************************************************/
2023-12-30 00:43:39 +01:00
/* Copyright (c) 2023-present Péter Magyar. */
2023-12-30 00:41:59 +01:00
/* Copyright (c) 2022-2023 Marc Melikyan */
/* */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the */
/* "Software"), to deal in the Software without restriction, including */
/* without limitation the rights to use, copy, modify, merge, publish, */
/* distribute, sublicense, and/or sell copies of the Software, and to */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions: */
/* */
/* The above copyright notice and this permission notice shall be */
/* included in all copies or substantial portions of the Software. */
/* */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */
/*************************************************************************/
2023-01-24 18:12:23 +01:00
#include "mann.h"
2023-02-17 18:46:27 +01:00
2024-01-25 13:42:45 +01:00
#ifdef USING_SFW
#include "sfw.h"
#else
2023-02-17 18:46:27 +01:00
#include "core/log/logger.h"
2024-01-25 13:42:45 +01:00
#endif
2023-02-17 18:46:27 +01:00
2023-01-24 18:12:23 +01:00
#include "../activation/activation.h"
2023-01-24 19:00:54 +01:00
#include "../cost/cost.h"
2023-01-24 18:12:23 +01:00
#include "../regularization/reg.h"
#include "../utilities/utilities.h"
2023-02-11 09:53:16 +01:00
/*
Ref<MLPPMatrix> MLPPMANN::get_input_set() {
return input_set;
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
void MLPPMANN::set_input_set(const Ref<MLPPMatrix> &val) {
input_set = val;
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
_initialized = false;
}
Ref<MLPPMatrix> MLPPMANN::get_output_set() {
return output_set;
}
void MLPPMANN::set_output_set(const Ref<MLPPMatrix> &val) {
output_set = val;
_initialized = false;
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
*/
2023-01-24 19:00:54 +01:00
2023-02-17 18:46:27 +01:00
Ref<MLPPMatrix> MLPPMANN::model_set_test(const Ref<MLPPMatrix> &X) {
ERR_FAIL_COND_V(!_initialized, Ref<MLPPMatrix>());
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
Ref<MLPPHiddenLayer> layer = _network[0];
2023-02-11 09:53:16 +01:00
2023-02-17 18:46:27 +01:00
layer->set_input(X);
layer->forward_pass();
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();
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
_output_layer->set_input(_network.write[_network.size() - 1]->get_a());
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
_output_layer->set_input(X);
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
2023-02-17 18:46:27 +01:00
_output_layer->forward_pass();
2023-02-11 09:53:16 +01:00
2023-02-17 18:46:27 +01:00
return _output_layer->get_a();
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
Ref<MLPPVector> MLPPMANN::model_test(const Ref<MLPPVector> &x) {
ERR_FAIL_COND_V(!_initialized, Ref<MLPPVector>());
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
Ref<MLPPHiddenLayer> layer = _network[0];
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());
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
_output_layer->test(_network.write[_network.size() - 1]->get_a_test());
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
_output_layer->test(x);
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
return _output_layer->get_a_test();
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
void MLPPMANN::gradient_descent(real_t learning_rate, int max_epoch, bool ui) {
ERR_FAIL_COND(!_initialized);
MLPPCost mlpp_cost;
2023-01-24 19:23:30 +01:00
MLPPActivation avn;
2023-01-25 00:54:50 +01:00
MLPPReg regularization;
2023-01-24 19:00:54 +01:00
2023-01-27 13:01:16 +01:00
real_t cost_prev = 0;
2023-01-24 19:00:54 +01:00
int epoch = 1;
2023-02-11 09:53:16 +01:00
forward_pass();
2023-01-24 19:00:54 +01:00
while (true) {
2023-02-11 09:53:16 +01:00
cost_prev = cost(_y_hat, _output_set);
2023-01-24 19:00:54 +01:00
2023-02-17 18:46:27 +01:00
if (_output_layer->get_activation() == MLPPActivation::ACTIVATION_FUNCTION_SOFTMAX) {
_output_layer->set_delta(_y_hat->subn(_output_set));
2023-01-24 19:00:54 +01:00
} else {
2023-12-28 11:17:24 +01:00
Ref<MLPPMatrix> r1 = mlpp_cost.run_cost_deriv_matrix(_output_layer->get_cost(), _y_hat, _output_set);
Ref<MLPPMatrix> r2 = avn.run_activation_deriv_matrix(_output_layer->get_activation(), _output_layer->get_z());
_output_layer->set_delta(r1->hadamard_productn(r2));
2023-01-24 19:00:54 +01:00
}
Ref<MLPPMatrix> output_w_grad = _output_layer->get_input()->transposen()->multn(_output_layer->get_delta());
2023-01-24 19:00:54 +01:00
_output_layer->set_weights(_output_layer->get_weights()->subn(output_w_grad->scalar_multiplyn(learning_rate / _n)));
2023-12-28 11:17:24 +01:00
_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->set_bias(_output_layer->get_bias()->subtract_matrix_rowsn(_output_layer->get_delta()->scalar_multiplyn(learning_rate / _n)));
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
Ref<MLPPHiddenLayer> layer = _network[_network.size() - 1];
layer->set_delta(_output_layer->get_delta()->multn(_output_layer->get_weights()->transposen())->hadamard_productn(avn.run_activation_deriv_matrix(layer->get_activation(), layer->get_z())));
2023-01-24 19:00:54 +01:00
Ref<MLPPMatrix> hidden_layer_w_grad = layer->get_input()->transposen()->multn(layer->get_delta());
layer->set_weights(layer->get_weights()->subn(hidden_layer_w_grad->scalar_multiplyn(learning_rate / _n)));
2023-02-17 18:46:27 +01:00
layer->set_weights(regularization.reg_weightsm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg()));
layer->set_bias(layer->get_bias()->subtract_matrix_rowsn(layer->get_delta()->scalar_multiplyn(learning_rate / _n)));
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
for (int i = _network.size() - 2; i >= 0; i--) {
2023-02-17 18:46:27 +01:00
layer = _network[i];
Ref<MLPPHiddenLayer> next_layer = _network[i + 1];
layer->set_delta(next_layer->get_delta()->multn(next_layer->get_weights())->hadamard_productn(avn.run_activation_deriv_matrix(layer->get_activation(), layer->get_z())));
hidden_layer_w_grad = layer->get_input()->transposen()->multn(layer->get_delta());
layer->set_weights(layer->get_weights()->subn(hidden_layer_w_grad->scalar_multiplyn(learning_rate / _n)));
2023-02-17 18:46:27 +01:00
layer->set_weights(regularization.reg_weightsm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg()));
layer->set_bias(layer->get_bias()->subtract_matrix_rowsn(layer->get_delta()->scalar_multiplyn(learning_rate / _n)));
2023-01-24 19:00:54 +01:00
}
}
2023-02-11 09:53:16 +01:00
forward_pass();
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
if (ui) {
2023-02-17 18:46:27 +01:00
MLPPUtilities::cost_info(epoch, cost_prev, cost(_y_hat, _output_set));
PLOG_MSG("Layer " + itos(_network.size() + 1) + ": ");
MLPPUtilities::print_ui_mb(_output_layer->get_weights(), _output_layer->get_bias());
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
for (int i = _network.size() - 1; i >= 0; i--) {
2023-02-17 18:46:27 +01:00
PLOG_MSG("Layer " + itos(i + 1) + ": ");
Ref<MLPPHiddenLayer> layer = _network[i];
MLPPUtilities::print_ui_mb(layer->get_weights(), layer->get_bias());
2023-01-24 19:00:54 +01:00
}
}
}
epoch++;
2023-02-11 09:53:16 +01:00
2023-01-24 19:00:54 +01:00
if (epoch > max_epoch) {
break;
}
}
}
2023-01-27 13:01:16 +01:00
real_t MLPPMANN::score() {
2023-02-11 09:53:16 +01:00
ERR_FAIL_COND_V(!_initialized, 0);
2023-02-10 21:41:05 +01:00
MLPPUtilities util;
2023-02-11 09:53:16 +01:00
forward_pass();
2023-02-17 18:46:27 +01:00
return util.performance_mat(_y_hat, _output_set);
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
void MLPPMANN::save(const String &file_name) {
2023-02-11 09:53:16 +01:00
ERR_FAIL_COND(!_initialized);
2023-02-17 18:46:27 +01:00
/*
2023-02-10 21:41:05 +01:00
MLPPUtilities util;
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
util.saveParameters(file_name, _network[0].weights, _network[0].bias, false, 1);
2023-02-11 09:53:16 +01:00
for (uint32_t i = 1; i < _network.size(); i++) {
2023-02-17 18:46:27 +01:00
util.saveParameters(file_name, _network[i].weights, _network[i].bias, true, i + 1);
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
util.saveParameters(file_name, _output_layer->weights, _output_layer->bias, true, _network.size() + 1);
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
util.saveParameters(file_name, _output_layer->weights, _output_layer->bias, false, _network.size() + 1);
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
*/
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
void MLPPMANN::add_layer(int n_hidden, MLPPActivation::ActivationFunction activation, MLPPUtilities::WeightDistributionType weight_init, MLPPReg::RegularizationType reg, real_t lambda, real_t alpha) {
2023-02-11 09:53:16 +01:00
if (_network.empty()) {
2023-02-17 18:46:27 +01:00
_network.push_back(Ref<MLPPHiddenLayer>(memnew(MLPPHiddenLayer(n_hidden, activation, _input_set, weight_init, reg, lambda, alpha))));
_network.write[0]->forward_pass();
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
_network.push_back(Ref<MLPPHiddenLayer>(memnew(MLPPHiddenLayer(n_hidden, activation, _network.write[_network.size() - 1]->get_a(), weight_init, reg, lambda, alpha))));
_network.write[_network.size() - 1]->forward_pass();
2023-01-24 19:00:54 +01:00
}
}
2023-02-17 18:46:27 +01:00
void MLPPMANN::add_output_layer(MLPPActivation::ActivationFunction activation, MLPPCost::CostTypes loss, MLPPUtilities::WeightDistributionType weight_init, MLPPReg::RegularizationType reg, real_t lambda, real_t alpha) {
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
_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)));
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
_output_layer = Ref<MLPPMultiOutputLayer>(memnew(MLPPMultiOutputLayer(_n_output, _k, activation, loss, _input_set, weight_init, reg, lambda, alpha)));
2023-02-11 09:53:16 +01:00
}
}
bool MLPPMANN::is_initialized() {
return _initialized;
}
void MLPPMANN::initialize() {
if (_initialized) {
return;
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
//ERR_FAIL_COND(!input_set.is_valid() || !output_set.is_valid() || n_hidden == 0);
_initialized = true;
}
2023-02-17 18:46:27 +01:00
MLPPMANN::MLPPMANN(const Ref<MLPPMatrix> &p_input_set, const Ref<MLPPMatrix> &p_output_set) {
2023-02-11 09:53:16 +01:00
_input_set = p_input_set;
_output_set = p_output_set;
2023-02-17 18:46:27 +01:00
_n = _input_set->size().y;
_k = _input_set->size().x;
_n_output = _output_set->size().x;
2023-02-11 09:53:16 +01:00
_initialized = true;
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
MLPPMANN::MLPPMANN() {
_initialized = false;
}
MLPPMANN::~MLPPMANN() {
}
2023-02-17 18:46:27 +01:00
real_t MLPPMANN::cost(const Ref<MLPPMatrix> &y_hat, const Ref<MLPPMatrix> &y) {
2023-01-25 00:54:50 +01:00
MLPPReg regularization;
2023-02-17 18:46:27 +01:00
MLPPCost mlpp_cost;
real_t total_reg_term = 0;
2023-01-24 19:00:54 +01:00
2023-02-11 09:53:16 +01:00
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
for (int i = 0; i < _network.size() - 1; i++) {
Ref<MLPPHiddenLayer> layer = _network[i];
total_reg_term += regularization.reg_termm(layer->get_weights(), layer->get_lambda(), layer->get_alpha(), layer->get_reg());
2023-01-24 19:00:54 +01:00
}
}
2023-02-17 18:46:27 +01:00
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());
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
void MLPPMANN::forward_pass() {
if (!_network.empty()) {
2023-02-17 18:46:27 +01:00
Ref<MLPPHiddenLayer> layer = _network[0];
2023-01-24 19:00:54 +01:00
2023-02-17 18:46:27 +01:00
layer->set_input(_input_set);
layer->forward_pass();
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();
2023-01-24 19:00:54 +01:00
}
2023-02-17 18:46:27 +01:00
_output_layer->set_input(_network.write[_network.size() - 1]->get_a());
2023-01-24 19:00:54 +01:00
} else {
2023-02-17 18:46:27 +01:00
_output_layer->set_input(_input_set);
2023-01-24 19:00:54 +01:00
}
2023-02-11 09:53:16 +01:00
2023-02-17 18:46:27 +01:00
_output_layer->forward_pass();
_y_hat = _output_layer->get_a();
2023-02-11 09:53:16 +01:00
}
void MLPPMANN::_bind_methods() {
/*
ClassDB::bind_method(D_METHOD("get_input_set"), &MLPPMANN::get_input_set);
ClassDB::bind_method(D_METHOD("set_input_set", "val"), &MLPPMANN::set_input_set);
ADD_PROPERTY(PropertyInfo(Variant::OBJECT, "input_set", PROPERTY_HINT_RESOURCE_TYPE, "MLPPMatrix"), "set_input_set", "get_input_set");
ClassDB::bind_method(D_METHOD("get_output_set"), &MLPPMANN::get_output_set);
ClassDB::bind_method(D_METHOD("set_output_set", "val"), &MLPPMANN::set_output_set);
ADD_PROPERTY(PropertyInfo(Variant::OBJECT, "output_set", PROPERTY_HINT_RESOURCE_TYPE, "MLPPMatrix"), "set_output_set", "get_output_set");
*/
2023-01-24 19:00:54 +01:00
}