pred_nlms.c

Go to the documentation of this file.

/*
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
 
#include "pred_nlms.h"
#include "blas.h"
#include "sam.h"
#include "utils.h"
 
#define N_MU (1) 
 
static const int MU_TYPE[N_MU] = {
    SAM_LOG_NORMAL 
};
 
void
pred_nlms_init(const struct XCSF *xcsf, struct Cl *c)
{
    struct PredNLMS *pred = malloc(sizeof(struct PredNLMS));
    c->pred = pred;
    // set the length of weights per predicted variable
    if (xcsf->pred->type == PRED_TYPE_NLMS_QUADRATIC) {
        // offset(1) + n linear + n quadratic + n*(n-1)/2 mixed terms
        pred->n = 1 + 2 * xcsf->x_dim + xcsf->x_dim * (xcsf->x_dim - 1) / 2;
    } else {
        pred->n = xcsf->x_dim + 1;
    }
    // initialise weights
    pred->n_weights = pred->n * xcsf->y_dim;
    pred->weights = calloc(pred->n_weights, sizeof(double));
    blas_fill(xcsf->y_dim, xcsf->pred->x0, pred->weights, pred->n);
    // initialise learning rate
    pred->mu = malloc(sizeof(double) * N_MU);
    if (xcsf->pred->evolve_eta) {
        sam_init(pred->mu, N_MU, MU_TYPE);
        pred->eta = rand_uniform(xcsf->pred->eta_min, xcsf->pred->eta);
    } else {
        memset(pred->mu, 0, sizeof(double) * N_MU);
        pred->eta = xcsf->pred->eta;
    }
    // initialise temporary storage for weight updating
    pred->tmp_input = malloc(sizeof(double) * pred->n);
}
 
void
pred_nlms_copy(const struct XCSF *xcsf, struct Cl *dest, const struct Cl *src)
{
    pred_nlms_init(xcsf, dest);
    struct PredNLMS *dest_pred = dest->pred;
    const struct PredNLMS *src_pred = src->pred;
    memcpy(dest_pred->weights, src_pred->weights,
           sizeof(double) * src_pred->n_weights);
    memcpy(dest_pred->mu, src_pred->mu, sizeof(double) * N_MU);
    dest_pred->eta = src_pred->eta;
}
 
void
pred_nlms_free(const struct XCSF *xcsf, const struct Cl *c)
{
    (void) xcsf;
    struct PredNLMS *pred = c->pred;
    free(pred->weights);
    free(pred->tmp_input);
    free(pred->mu);
    free(pred);
}
 
void
pred_nlms_update(const struct XCSF *xcsf, const struct Cl *c, const double *x,
                 const double *y)
{
    const struct PredNLMS *pred = c->pred;
    // normalise update
    const int n = pred->n;
    const double X0 = xcsf->pred->x0;
    const double norm = X0 * X0 + blas_dot(xcsf->x_dim, x, 1, x, 1);
    // update weights using the error
    for (int i = 0; i < xcsf->y_dim; ++i) {
        const double error = y[i] - c->prediction[i];
        const double correction = (pred->eta * error) / norm;
        blas_axpy(n, correction, pred->tmp_input, 1, &pred->weights[i * n], 1);
    }
}
 
void
pred_nlms_compute(const struct XCSF *xcsf, const struct Cl *c, const double *x)
{
    const struct PredNLMS *pred = c->pred;
    const int n = pred->n;
    pred_transform_input(xcsf, x, xcsf->pred->x0, pred->tmp_input);
    for (int i = 0; i < xcsf->y_dim; ++i) {
        c->prediction[i] =
            blas_dot(n, &pred->weights[i * n], 1, pred->tmp_input, 1);
    }
}
 
void
pred_nlms_print(const struct XCSF *xcsf, const struct Cl *c)
{
    char *json_str = pred_nlms_json_export(xcsf, c);
    printf("%s\n", json_str);
    free(json_str);
}
 
bool
pred_nlms_crossover(const struct XCSF *xcsf, const struct Cl *c1,
                    const struct Cl *c2)
{
    (void) xcsf;
    (void) c1;
    (void) c2;
    return false;
}
 
bool
pred_nlms_mutate(const struct XCSF *xcsf, const struct Cl *c)
{
    if (xcsf->pred->evolve_eta) {
        struct PredNLMS *pred = c->pred;
        sam_adapt(pred->mu, N_MU, MU_TYPE);
        const double orig = pred->eta;
        pred->eta += rand_normal(0, pred->mu[0]);
        pred->eta = clamp(pred->eta, xcsf->pred->eta_min, xcsf->pred->eta);
        if (orig != pred->eta) {
            return true;
        }
    }
    return false;
}
 
double
pred_nlms_size(const struct XCSF *xcsf, const struct Cl *c)
{
    (void) xcsf;
    const struct PredNLMS *pred = c->pred;
    return pred->n_weights;
}
 
size_t
pred_nlms_save(const struct XCSF *xcsf, const struct Cl *c, FILE *fp)
{
    (void) xcsf;
    const struct PredNLMS *pred = c->pred;
    size_t s = 0;
    s += fwrite(&pred->n, sizeof(int), 1, fp);
    s += fwrite(&pred->n_weights, sizeof(int), 1, fp);
    s += fwrite(pred->weights, sizeof(double), pred->n_weights, fp);
    s += fwrite(pred->mu, sizeof(double), N_MU, fp);
    s += fwrite(&pred->eta, sizeof(double), 1, fp);
    return s;
}
 
size_t
pred_nlms_load(const struct XCSF *xcsf, struct Cl *c, FILE *fp)
{
    pred_nlms_init(xcsf, c);
    struct PredNLMS *pred = c->pred;
    size_t s = 0;
    s += fread(&pred->n, sizeof(int), 1, fp);
    s += fread(&pred->n_weights, sizeof(int), 1, fp);
    s += fread(pred->weights, sizeof(double), pred->n_weights, fp);
    s += fread(pred->mu, sizeof(double), N_MU, fp);
    s += fread(&pred->eta, sizeof(double), 1, fp);
    return s;
}
 
char *
pred_nlms_json_export(const struct XCSF *xcsf, const struct Cl *c)
{
    const struct PredNLMS *pred = c->pred;
    cJSON *json = cJSON_CreateObject();
    if (xcsf->pred->type == PRED_TYPE_NLMS_QUADRATIC) {
        cJSON_AddStringToObject(json, "type", "nlms_quadratic");
    } else {
        cJSON_AddStringToObject(json, "type", "nlms_linear");
    }
    cJSON *weights = cJSON_CreateDoubleArray(pred->weights, pred->n_weights);
    cJSON_AddItemToObject(json, "weights", weights);
    cJSON_AddNumberToObject(json, "eta", pred->eta);
    cJSON *mutation = cJSON_CreateDoubleArray(pred->mu, N_MU);
    cJSON_AddItemToObject(json, "mutation", mutation);
    char *string = cJSON_Print(json);
    cJSON_Delete(json);
    return string;
}
 
void
pred_nlms_json_import(const struct XCSF *xcsf, struct Cl *c, const cJSON *json)
{
    (void) xcsf;
    struct PredNLMS *pred = c->pred;
    const cJSON *item = cJSON_GetObjectItem(json, "weights");
    if (item != NULL && cJSON_IsArray(item)) {
        if (cJSON_GetArraySize(item) == pred->n_weights) {
            for (int i = 0; i < pred->n_weights; ++i) {
                const cJSON *item_i = cJSON_GetArrayItem(item, i);
                pred->weights[i] = item_i->valuedouble;
            }
        } else {
            printf("Import error: weight length mismatch\n");
            exit(EXIT_FAILURE);
        }
    }
    item = cJSON_GetObjectItem(json, "eta");
    if (item != NULL && cJSON_IsNumber(item)) {
        pred->eta = item->valuedouble;
    }
    sam_json_import(pred->mu, N_MU, json);
}
 
char *
pred_nlms_param_json_export(const struct XCSF *xcsf)
{
    const struct ArgsPred *pred = xcsf->pred;
    cJSON *json = cJSON_CreateObject();
    cJSON_AddNumberToObject(json, "x0", pred->x0);
    cJSON_AddNumberToObject(json, "eta", pred->eta);
    cJSON_AddBoolToObject(json, "evolve_eta", pred->evolve_eta);
    if (pred->evolve_eta) {
        cJSON_AddNumberToObject(json, "eta_min", pred->eta_min);
    }
    char *string = cJSON_Print(json);
    cJSON_Delete(json);
    return string;
}
 
char *
pred_nlms_param_json_import(struct XCSF *xcsf, cJSON *json)
{
    for (cJSON *iter = json; iter != NULL; iter = iter->next) {
        if (strncmp(iter->string, "x0\0", 3) == 0 && cJSON_IsNumber(iter)) {
            pred_param_set_x0(xcsf, iter->valuedouble);
        } else if (strncmp(iter->string, "eta\0", 4) == 0 &&
                   cJSON_IsNumber(iter)) {
            pred_param_set_eta(xcsf, iter->valuedouble);
        } else if (strncmp(iter->string, "evolve_eta\0", 11) == 0 &&
                   cJSON_IsBool(iter)) {
            const bool evolve = true ? iter->type == cJSON_True : false;
            pred_param_set_evolve_eta(xcsf, evolve);
        } else if (strncmp(iter->string, "eta_min\0", 8) == 0 &&
                   cJSON_IsNumber(iter)) {
            pred_param_set_eta_min(xcsf, iter->valuedouble);
        } else {
            return iter->string;
        }
    }
    return NULL;
}