AppAIfES()

AppAIfES::AppAIfES ( )

inline

Definition at line 21 of file AppAIfES.h.

              :AppBaseClass(){
        strcpy(name,"AI");
        x_test = NULL;
        y_test = NULL;
        // The layer structures for F32 data type and their configurations
        input_layer_shape[0] = 1;
        input_layer_shape[1] = 3;
        input_layer      = AILAYER_INPUT_F32_A(2, input_layer_shape);
        dense_layer_1    = AILAYER_DENSE_F32_A(3);
        leaky_relu_layer = AILAYER_LEAKY_RELU_F32_A(0.01f);
        dense_layer_2    = AILAYER_DENSE_F32_A(2);
        sigmoid_layer    = AILAYER_SIGMOID_F32_A();
 
        model.input_layer = ailayer_input_f32_default(&input_layer);
        x_layer = ailayer_dense_f32_default(&dense_layer_1, model.input_layer);
        x_layer = ailayer_leaky_relu_f32_default(&leaky_relu_layer, x_layer);
        x_layer = ailayer_dense_f32_default(&dense_layer_2, x_layer);
        x_layer = ailayer_sigmoid_f32_default(&sigmoid_layer, x_layer);
        model.output_layer = x_layer;
        
        // Finish the model creation by checking the connections and setting some parameters for further processing
        aialgo_compile_model(&model);
 
        uint32_t parameter_memory_size = aialgo_sizeof_parameter_memory(&model);
        parameter_memory = extmem_malloc(parameter_memory_size);
 
        // Distribute the memory to the trainable parameters of the model
        aialgo_distribute_parameter_memory(&model, parameter_memory, parameter_memory_size);
        //aialgo_print_model_structure(&model);
                
        model.loss = ailoss_crossentropy_f32_default(&crossentropy_loss, model.output_layer);
        //aialgo_print_loss_specs(model.loss);
 
        //config the optimizer
        adam_opti = AIOPTI_ADAM_F32(0.01f, 0.9f, 0.999f, 1e-7f);
        optimizer = aiopti_adam_f32_default(&adam_opti);
        //aialgo_print_optimizer_specs(optimizer);
 
        // Set the seed for your configured random function for example with
        srand(time(0));
        
        aimath_f32_default_init_he_uniform(&dense_layer_1.weights);
        aimath_f32_default_init_zeros(&dense_layer_1.bias);
        
        aimath_f32_default_init_glorot_uniform(&dense_layer_2.weights);
        aimath_f32_default_init_zeros(&dense_layer_2.bias);
 
        uint32_t training_memory_size = aialgo_sizeof_training_memory(&model, optimizer);
        void *training_memory = extmem_malloc(training_memory_size);
        
        // Schedule the memory to the model
        aialgo_schedule_training_memory(&model, optimizer, training_memory, training_memory_size);
 
        
        // One epoch of training. Iterates through the whole data once
        //float loss;
        //int batch_size = 3;
        //aialgo_train_model(&model, &x_train, &y_train, optimizer, batch_size);
        //aialgo_calc_loss_model_f32(&model, x_test, y_test, &loss);
            
        // Print the loss to the console
        //aiprint("M VM Epoch "); aiprint_int("%5d", 0);
        //aiprint(": test loss: "); aiprint_float("%f", loss);
        //aiprint("\n");
  
 
        //extmem_free(parameter_memory);
        //extmem_free(training_memory); 
    }; 

References adam_opti, crossentropy_loss, dense_layer_1, dense_layer_2, input_layer, input_layer_shape, leaky_relu_layer, model, AppBaseClass::name, optimizer, parameter_memory, sigmoid_layer, x_layer, x_test, and y_test.