ApPredict_GPmat/Fig3.m at master · sanmitraghosh/ApPredict_GPmat · 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
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    % This script gathers actively learnt training data for
    % classifier/boundary prediction. This corresponds to the Fig3 in the paper.
    % You can control the following variables to get different
    % behaviour::
    % 1) 'CScale'--No. of initial random sampled data
    % 1) 'AScale'--Active swarm/particle size. For surface fix this to 1
    % 1) 'Rounds'--No. active learning rounds
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    % Set up
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
close all
clear all
startup
load('Alearning_4D_100k_Train.mat');%Change this with 'Alearning_4D_100k_Train.mat' for 4D

GridData=load('Alearning_2D_10k_Grid.mat');
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Pass al GP related information using the gpoptions structure
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

CScale=500;
AScale=1;
Rounds=2500;
STOPSURF=CScale + Rounds*AScale;
gpoptions.pacing=100;
gpoptions.NumInducingClass=300;
gpoptions.sparseMargin=5000;
gpoptions.NumInducingSurf=1000;
gpoptions.sparseMarginSurf=10000;
gpoptions.classHyperParams.minimize=0;
gpoptions.surfHyperParams.minimize=0;
gpoptions.covarianceKernels=@covNNone;%@covRQiso;%{'covMaterniso',5};
gpoptions.covarianceKernelsParams=[0.1;1];%[0.1;0.21;1];%[0.1;1.20];
gpoptions.likelihoodParams=0.015;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    % Load up a pool of random training data
    % from which we will draw randomely to compare with active learning
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

X_random=gk;
Y_random=APDtrue;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Use the gridded data for visualisations for the 2D case
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if size(gk,2)==2
    X_Grid=GridData.gk;
    Y_Grid=GridData.APDtrue;
    [ ~, labelGrid ] = labelFinder( X_Grid, Y_Grid);
    dlmwrite('GridLabels.txt',labelGrid);
end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Sample Initial Random Training Data from the pool loaded earlier
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
nu = fix(CScale); cu = randperm(length(X_random)); cu = cu(1:nu);
x_Initrandom = X_random(cu,:);
y_Initrandom = Y_random(cu,:); %%% Start with random CScale points
X_random(cu,:)=[]; %%% Remove these chosen points from the pool
Y_random(cu)=[];

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Learn a classifier to constrain the chosen active points in the AP
%   region
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

gpoptions.LearningMode='classifier';
outparam= learnGPhyp( x_Initrandom, y_Initrandom, gpoptions );
gpoptions.classHyperParams=outparam;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Learn hyperparameters for surface active learning
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
[ ~, label ] = labelFinder( x_Initrandom, y_Initrandom );
Cord=find(label(:,2)==1);
x_InitTrain= x_Initrandom(Cord,:);
y_InitTrain= y_Initrandom(Cord);
gpoptions.LearningMode='surface';
outparam= learnGPhyp( x_InitTrain, y_InitTrain, gpoptions );
gpoptions.surfHyperParams=outparam;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Do a few rounds of surface active learning
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
gpoptions.Batch=0;
gpoptions.method='grid'; % Flag to trigger the swarming types or grid ('grid' 'pso' 'ga')
gpoptions.Model='APD';
gpoptions.LearningMode='surface';
gpoptions.STOP=STOPSURF;
gpoptions.AScale=AScale;
gpoptions.classHyperParams.minimize=0;
gpoptions.surfAlClass=1;
gpoptions.pltDisabled=1; % Turning it on starts plotting contour plots (slow)
gpoptions.ALtime=ones(100,1);
[ X_intellitrain, Y_intellitrain ] = intelligentTraining( x_Initrandom, y_Initrandom, gpoptions );
close all;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Keep only the AP related points for the active training data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

[ ~, label ] = labelFinder( X_intellitrain, Y_intellitrain );
Cord=find(label(:,2)==1);
x_train= X_intellitrain(Cord,:);
y_train= Y_intellitrain(Cord);


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Keep only the AP related points for random training data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

[ ~, label ] = labelFinder( X_random, Y_random );
Cord=find(label(:,2)==1);
X_train= X_random(Cord,:);
Y_train= Y_random(Cord);
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Save the dataset, please rename accordingly!
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
save('surfActive2D.mat')
% Clean-up
command='rm -rf *.txt';
system(command);
if size(gk,2)==2
    dlmwrite('GridLabels.txt',labelGrid);
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%   Call up evaluation script to compare with random learning
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Fig3a