8f43927575
examples/ is based on KasperskyLab/knp at 55470ed6 with local changes to the mnist-learn altai/blifat models, cuda-backend CMake and model_loader. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
117 lines
3.7 KiB
C++
117 lines
3.7 KiB
C++
/**
|
||
* @file hyperparameters.h
|
||
* @brief BLIFAT neuron model hyperparameters.
|
||
* @details This header file defines all hyperparameters for the BLIFAT neuron model implementation.
|
||
* The parameters are carefully tuned for spiking neural network learning with STDP mechanisms
|
||
* and are organized into logical categories for better understanding and maintenance.
|
||
* @kaspersky_support A. Vartenkov
|
||
* @date 28.07.2025
|
||
* @license Apache 2.0
|
||
* @copyright © 2025 AO Kaspersky Lab
|
||
*
|
||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||
* you may not use this file except in compliance with the License.
|
||
* You may obtain a copy of the License at
|
||
*
|
||
* http://www.apache.org/licenses/LICENSE-2.0
|
||
*
|
||
* Unless required by applicable law or agreed to in writing, software
|
||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||
* See the License for the specific language governing permissions and
|
||
* limitations under the License.
|
||
*/
|
||
#pragma once
|
||
|
||
#include <cstdint>
|
||
|
||
// cppcheck-suppress missingInclude
|
||
#include "global_config.h"
|
||
|
||
|
||
/**
|
||
* @brief Default activation threshold for BLIFAT neurons.
|
||
*/
|
||
constexpr float default_threshold = 8.571F;
|
||
/**
|
||
* @brief Minimum synaptic weight value.
|
||
*/
|
||
constexpr float min_synaptic_weight = -0.253;
|
||
/**
|
||
* @brief Maximum synaptic weight value
|
||
*/
|
||
constexpr float max_synaptic_weight = 0.0924;
|
||
/**
|
||
* @brief Base weight value
|
||
*/
|
||
constexpr float base_weight_value = 0.000F;
|
||
/**
|
||
* @brief Dopamine plasticity period for neuron-level learning.
|
||
*/
|
||
constexpr uint32_t neuron_dopamine_period = 10;
|
||
/**
|
||
* @brief Dopamine plasticity period for synapse-level learning.
|
||
*/
|
||
constexpr uint32_t synapse_dopamine_period = 10;
|
||
/**
|
||
* @brief A time constant during which the input neuron potential tends to zero.
|
||
*/
|
||
constexpr float input_neuron_potential_decay = 1.0 - 1.0 / 3.0;
|
||
/**
|
||
* @brief Dopamine modulation parameter for learning signals.
|
||
*/
|
||
constexpr float dopamine_parameter = 0.042F;
|
||
/**
|
||
* @brief Hebbian plasticity for STDP learning mechanisms.
|
||
*/
|
||
constexpr float hebbian_plasticity = -0.177;
|
||
/**
|
||
* @brief Threshold weight coefficient for synaptic sum calculations.
|
||
*/
|
||
constexpr float threshold_weight_coeff = 0.218F;
|
||
/**
|
||
* @brief Time between spikes in the ISI period.
|
||
*/
|
||
constexpr uint32_t isi_max = 10;
|
||
/**
|
||
* @brief A value to which membrane potential tends.
|
||
*/
|
||
constexpr double min_potential = 0;
|
||
/**
|
||
* @brief The stability fluctuation value.
|
||
*/
|
||
constexpr float stability_change_parameter = 0.05F;
|
||
/**
|
||
* @brief The number of silent synapses.
|
||
*/
|
||
constexpr uint32_t resource_drain_coefficient = 27;
|
||
/**
|
||
* @brief Random number in range [0,stochastic_stimulation) that is added to the potential every tick.
|
||
*/
|
||
constexpr float stochastic_stimulation = 2.212;
|
||
|
||
|
||
/**
|
||
* @brief Number of neurons reserved per digit class in input population.
|
||
*
|
||
* @details This parameter determines the columnar organization of the input population, with 20 neurons
|
||
* allocated for each of the 10 digit classes (200 total input neurons). Each neuron in a column corresponds
|
||
* to a specific feature of the digit class.
|
||
*/
|
||
constexpr size_t neurons_per_column = 20;
|
||
|
||
/**
|
||
* @brief Total number of input neurons in the network.
|
||
*
|
||
* @details Calculated as neurons_per_column × classes_amount, providing a direct mapping from digit classes
|
||
* to input neuron populations.
|
||
*/
|
||
constexpr size_t num_input_neurons = neurons_per_column * classes_amount;
|
||
|
||
/**
|
||
* @brief Number of independent subnetworks to use.
|
||
*
|
||
* @details This parameter controls the multi-subnetwork architecture of the BLIFAT model. Setting to 1 creates
|
||
* a single network, while higher values enable distributed processing across multiple network instances.
|
||
*/
|
||
constexpr size_t num_subnetworks = 1;
|