SensoryMotorIntegration/Network.h at master · daniel-navarro/SensoryMotorIntegration · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
/*
 *  Network.h
 *
 * Copyright 2018 OFTNAI. All rights reserved.
 *
 */

#ifndef NETWORK_H
#define NETWORK_H

// Forward declarations
class HiddenNeuron;
class BinaryWrite;
class HiddenRegion;
class InputRegion;

// Includes
#include "InputRegion.h"
#include "Param.h"
#include <vector>
#include "Utilities.h"
#include <gsl/gsl_cdf.h>
#include <gsl/gsl_randist.h>

using std::vector;

class Network {

    private:

        // Output FILE types
        enum OUTPUT_FILE {
            OF_REGIONAL = 0, //
            OF_REGION_NEURONAL = 1, //
            OF_REGION_SYNAPTIC = 2, //
            OF_SINGLE_CELLS = 3 //
        };

        // Outputing
        void outputHistory(const char * outputDirectory, bool isTraining);
        void openHistoryFile(BinaryWrite & file, const char * outputDirectory, const char * filename, bool isTraining, OUTPUT_FILE fileType);
        void outputRegionHistory(const char * outputDirectory, bool isTraining);
        void outputNeuronHistoryData(const char * outputDirectory, bool isTraining, DATA data);
        void outputSingleUnits(const char * outputDirectory);
        void outputSynapticHistory(const char * outputDirectory);

        // Utility functions
        void buildESPathway();
        void setupAfferentSynapsesV2();
		void setupAfferentSynapsesForV3AndAbove(u_short esPathwayIndex);
		void normalize(HiddenNeuron * n);

		bool verbose;
		Param p;

    public:
    	vector<HiddenRegion> ESPathway;
    	InputRegion area7a;
        gsl_rng * rngController;

    // dnavarro2016 convergence
    vector<vector<Synapse>> previous_epoch_synapses;

		// Build new network based on these parameters
		Network(const char * parameterFile, bool verbose);

    	// Load network from weight file
    	Network(const char * dataFile, const char * parameterFile, bool verbose, const char * inputWeightFile, bool isTraining);

    	// Destructor, frees ESPathway and rngController
    	~Network();

    	// Run based on parameter file supplied in constructor, argument is file list
		// isTraining = true means that we apply learning rule in param file and run the number of epochs in the param file
		// isTraining = false means that we have no learning and only run ONE epoch
    	void run(const char * outputDirectory, bool isTraining, int numberOfThreads, bool xgrid);
		u_short runContinous(const char * outputDirectory, bool isTraining, bool xgrid);

    	// Save final weights of network
        // dnavarro2016 convergence
        int outputConvergence();

        void outputFinalNetwork(const char * outputWeightFile);
};

#endif // NETWORK_H