-
Notifications
You must be signed in to change notification settings - Fork 0
/
U14222460_PRAC3_Task2.py
296 lines (227 loc) · 11 KB
/
U14222460_PRAC3_Task2.py
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
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
# -*- coding: utf-8 -*-
"""
Created on Mon Apr 13 13:39:46 2020
@author: Lefa Raleting
"""
import numpy as np
import matplotlib.pyplot as plt
import time
import csv
import pandas as pd
#This part of the code. opens up a specified csv file, then adds it to a turple
""""
with open('data1.csv', 'r') as csvFile:
reader = csv.reader(csvFile)
line= [tuple(row) for row in reader]
csvFile.close()
database=line
#function to split datainto chunks
def chunks(l,n):
for i in range(0,len(l),n):
yield l[i:i+n]
#print(list(chunks(database,90))[0][89])
#This part of the code then goes on to create a data base of the the data collected
#counting ever instances and classifying it in a dictionary
#database=collections.Counter()
#database.update(line)
"""
#these are my helper dictionaries to better help navigate my functions
indexmoves= {
0:'RRRR', 1:'RRRP', 2:'RRRS', 3:'RRPR', 4:'RRPP', 5:'RRPS', 6:'RRSR',
7:'RRSP', 8:'RRSS', 9: 'RPRR', 10:'RPRP',11:'RPRS', 12:'RPPR', 13:'RPPP',
14:'RPPS',15:'RPSR',16:'RPSP',17:'RPSS', 18:'RSRR', 19:'RSRP',20: 'RSRS',
21: 'RSPR', 22: 'RSPP', 23: 'RSPS', 24: 'RSSR', 25: 'RSSP', 26: 'RSSS',
27: 'PRRR', 28: 'PRRP', 29: 'PRRS', 30: 'PRPR', 31: 'PRPP', 32: 'PRPS',
33: 'PRSR', 34: 'PRSP', 35: 'PRSS', 36: 'PPRR', 37: 'PPRP', 38: 'PPRS',
39: 'PPPR', 40: 'PPPP', 41: 'PPPS', 42: 'PPSR', 43: 'PPSP', 44: 'PPSS',
45: 'PSRR', 46: 'PSRP', 47: 'PSRS', 48: 'PSPR', 49: 'PSPP', 50: 'PSPS',
51: 'PSSR', 52: 'PSSP', 53: 'PSSS', 54: 'SRRR', 55: 'SRRP', 56: 'SRRS',
57: 'SRPR', 58: 'SRPP', 59: 'SRPS', 60: 'SRSR', 61: 'SRSP', 62: 'SRSS',
63: 'SPRR', 64: 'SPRP', 65: 'SPRS', 66: 'SPPR', 67: 'SPPP', 68: 'SPPS',
69: 'SPSR', 70: 'SPSP', 71: 'SPSS', 72: 'SSRR', 73: 'SSRP', 74: 'SSRS',
75: 'SSPR', 76: 'SSPP', 77: 'SSPS', 78: 'SSSR', 79: 'SSSP', 80: 'SSSS'}
plays={0:'R',1:'P',2:'S'}
players={'R':0,'P':1,'S':2}
outputEncoder={'R':[1,0,0],'P':[0,1,0],'S':[0,0,1]}
movesIndex={
"RRRR": 0,"RRRP": 1,"RRRS": 2,"RRPR": 3,"RRPP": 4,"RRPS": 5,"RRSR": 6,
"RRSP": 7,"RRSS": 8,"RPRR":9,"RPRP": 10,"RPRS":11,"RPPR": 12,"RPPP": 13,
"RPPS":14,"RPSR":15,"RPSP":16,"RPSS":17,"RSRR":18,"RSRP":19,"RSRS":20,
"RSPR":21,"RSPP":22,"RSPS":23,"RSSR":24,"RSSP":25,"RSSS":26,"PRRR":27,
"PRRP":28,"PRRS":29,"PRPR":30,"PRPP":31,"PRPS":32,"PRSR":33,"PRSP":34,
"PRSS":35,"PPRR":36,"PPRP":37,"PPRS":38,"PPPR":39,"PPPP":40,"PPPS":41,
"PPSR":42,"PPSP":43,"PPSS":44,"PSRR":45,"PSRP":46,"PSRS":47,"PSPR":48,
"PSPP":49,"PSPS":50,"PSSR":51,"PSSP":52,"PSSS":53,"SRRR":54,"SRRP":55,
"SRRS":56,"SRPR":57,"SRPP":58,"SRPS":59,"SRSR":60,"SRSP":61,"SRSS":62,
"SPRR":63,"SPRP":64,"SPRS":65,"SPPR":66,"SPPP":67,"SPPS":68,"SPSR":69,
"SPSP":70,"SPSS":71,"SSRR":72,"SSRP":73,"SSRS":74,"SSPR":75,"SSPP":76,
"SSPS":77,"SSSR":78,"SSSP":79,"SSSS":80
}
encoder={
"RRRR": 0.0,"RRRP": 0.01,"RRRS": 0.02,"RRPR": 0.03,"RRPP": 0.04,"RRPS": 0.05,
"RRSR": 0.06,"RRSP": 0.07,"RRSS": 0.08,"RPRR": 0.09,"RPRP": 0.10,"RPRS":0.11,
"RPPR": 0.12,"RPPP": 0.13, "RPPS":0.14,"RPSR": 0.15,"RPSP": 0.16,"RPSS":0.17,
"RSRR": 0.18,"RSRP": 0.19,"RSRS": 0.20,"RSPR": 0.21,"RSPP": 0.22,"RSPS":0.23,
"RSSR": 0.24,"RSSP": 0.25,"RSSS": 0.26,"PRRR": 0.27,"PRRP": 0.28,"PRRS":0.29,
"PRPR": 0.30,"PRPP": 0.31,"PRPS": 0.32,"PRSR": 0.33,"PRSP": 0.34, "PRSS":0.35,
"PPRR": 0.36,"PPRP": 0.37,"PPRS": 0.38,"PPPR": 0.39,"PPPP": 0.40,"PPPS":0.41,
"PPSR": 0.42,"PPSP": 0.43,"PPSS": 0.44,"PSRR": 0.45,"PSRP": 0.46,"PSRS":0.47,
"PSPR": 0.48,"PSPP": 0.49,"PSPS": 0.50,"PSSR": 0.51,"PSSP": 0.52,"PSSS":0.53,
"SRRR": 0.54,"SRRP": 0.55,"SRRS": 0.56,"SRPR": 0.57,"SRPP": 0.58,"SRPS":0.59,
"SRSR": 0.60,"SRSP": 0.61,"SRSS": 0.62,"SPRR": 0.63,"SPRP": 0.64,"SPRS":0.65,
"SPPR": 0.66,"SPPP": 0.67,"SPPS": 0.68,"SPSR": 0.69,"SPSP": 0.70,"SPSS":0.71,
"SSRR": 0.72,"SSRP": 0.73,"SSRS": 0.74,"SSPR": 0.75,"SSPP": 0.76,"SSPS":0.77,
"SSSR": 0.78,"SSSP": 0.79,"SSSS": 0.80
}
#_______________Global parameters______________________________\
Input_number = 1 # This the number of nodes in the input layer
Hidden_layers = 2 #This is the number hidden layers
Hidden_number1= 3 #number of neurons in the first hidden layer
Hidden_number2= 3 #Number of neurons in the second hidden layer
Output_number= 3 #Number of nodes in output layer
#__________________Intializations______________________________
#________________________Matrix's_______________________________
#Matrix for between input and hidden layer 1
#Weight_IH= np.array([[-1.30256608, -0.63358578 , 0.46809963]])#np.random.randn(Input_number,Hidden_number1)
#Weight_IH= np.array([[ 0.03439296, -0.84921801, -0.67630459]])
Weight_IH=np.array([[0.98546655, 1.0, 0.68011305]]) #np.array([[ 0.62028726, -0.23636479, -0.56722029]])
#matrix for hidden layer 1
Bias_H1 = np.array([[0.65685711, 0.6665443, 0.45332548]])#np.random.rand(1,Hidden_number1)
#Matrix for between input and hidden layer 2
#Weight_HH= np.array([[-1.82436288e-05, 1.11140185e-03, -2.23581046e-05],[ 6.63464877e-02, -3.05224354e-01, 2.11662861e-01],[ 8.98090548e-02, -4.11788163e-01, 2.86453940e-01]])# np.random.randn(Hidden_number1,Hidden_number2)
#Weight_HH=np.array([[-1.85954332, 0.99382871, 0.19574764],[-0.32298423, 0.14265085, -0.03934919],[-1.48797382, 0.83164407, 0.2485742 ]])
Weight_HH=np.array([[0.98546655, 0.33667067, 0.38093468],
[1.0 , 0.34163581, 0.38655262],
[0.68011305, 0.23235098, 0.26289948]])
#matrix for hidden layer 2
Bias_H2= np.array([[0.3889918, 0.13289353, 0.1503658 ]])#np.random.rand(1,Hidden_number2)
#Weight Matrix for hidden layer 2 to output
#Weight_HO= np.array([[ 0.3024957 , -0.39158243, -0.38557635],[-0.30391214, -0.63549297, 0.11393838],[-0.1940526, -0.70688921 ,-0.70417734]])#np.random.randn(Hidden_number2,Output_number)
#Weight_HO=np.array([[ 0.36541192, -0.77949451, -0.85524656],[-1.1897309, -0.59789276, 0.93715327],[-0.22587574, -0.26511212, -0.71704524]])
Weight_HO=np.array([[0.80960078, 1.0, 0.92964603],
[0.27658862, 0.34163581, 0.31760038],
[0.3129533, 0.38655262, 0.35935711]])
#Bias matrix for output layer
Bias_O =np.array([[0.93448987, 0.47523067, 0.48086404]])#np.random.rand(1,Output_number)
#Activation Functions
#Leaky Relu
def Relu(x,a=0.1):
#if x is less than zero
#return the multiplication of the value and small gradient
#else return the value as is
x[x<0]=a*x[x<0]
return x
def DerivativeRelu(x,a=0.1):
x[x<=0]=a #let the diravitive equal to the slope which is a for x<=0
x[x>0]=1 #slope equal 1 for x>0
return x
#Soft max function
def Softmax(x):
# return np.exp(x)/np.exp(x).sum(axis=0)
return np.exp(x-np.max(x))/np.sum(np.exp(x-np.max(x)),axis=0)
#softmax derivative
def Softmaxderv(x):
#soft= x.reshape(-1,1)
#return np.diagflat(soft) - np.dot(soft, soft.T)
return (1-Softmax(x[0]))
#This Function is for forward propergation. Explained in the report
def Foward_Propagation(Inputs,Weight_IH,Weight_HH,Weight_HO,a=0.1):
S1= np.dot(Inputs,Weight_IH)+ Bias_H1
A1= Relu(S1[0],a)
S2= np.dot(A1,Weight_HH)+ Bias_H2
A2= Relu(S2[0],a)
Y= np.dot(A2,Weight_HO)+ Bias_O
Out= Softmax(Y[0])
return S1,A1,S2,A2,Y,Out
#This function is for backpopagation and it works as explained in the report
def Back_Propagation(Out,S1,A1,S2,A2,Y,LearningRate,Target,inputs,a=0.1):
global Weight_HO
global Weight_HH
global Weight_IH
Alpha1= np.multiply(-(Target-Out),Softmaxderv(Y)) #np.dot(np.array((Target-Out)),Softmaxderv(Y)) #3by3 matrix
Link_H2O = np.dot(np.array([A2]).T, np.array([Alpha1])) #gradient calculation
Alpha2= np.dot(Alpha1,Weight_HO.T)*DerivativeRelu(S2,a)
Link_H1H2= np.dot(np.array([A1]).T,Alpha2)
Alpha3=np.dot(Alpha2,Weight_HH.T)*DerivativeRelu(S1[0],a)
Link_IH1=np.dot(np.array([inputs]).T,Alpha3)#
Weight_HO=Weight_HO-LearningRate*Link_H2O
Weight_HO=Weight_HO/np.max(Weight_HO)
Weight_HH=Weight_HH-LearningRate*Link_H1H2
Weight_HH=Weight_HH/np.max(Weight_HH)
Weight_IH=Weight_IH-LearningRate*Link_IH1
Weight_IH=Weight_IH/np.max(Weight_IH)
return Weight_HO,Weight_HH,Weight_IH #Link_H2O,Link_H1H2,Link_IH1
#loss function to help corret the error in training
def Loss_function(EOut,Out):
L=0.5*sum((EOut-Out)**2)
return L
#_________________________TASK1________________________________________
gama=0.05
number_of_iterations=10 #Epoch
Loss=[]
def train(Inputs,Target,IH,HH,HO,LearningRate,Epoch,alpha):
# DataBase_length= len(database)
error=0
for i in range(Epoch):#[encoder[Inputs[k][0]]] outputEncoder[Target[k][1]]
for k in Inputs: #range(0,1000000):#DataBase_length
S1,A1,S2,A2,Y,Out=Foward_Propagation([encoder[k[0]]],IH,HH,HO,alpha)
HO,HH,IH=Back_Propagation(Out,S1,A1,S2,A2,Y,LearningRate,outputEncoder[k[1]],[encoder[k[0]]] ,alpha)
#arrayOut.append(Out)
#print(Y)
#L=Loss_functionDB(Target,arrayOut )
L=Loss_function(outputEncoder[k[1]],Out)
Loss.append(L)
if(i>1):
if(((Loss[i-1]-Loss[i])/Loss[i-1])<0.10 and (Loss[i-1]>=Loss[i]) ):
error+=1
if(error==3):
break
else:
error=0
# plt.plot(range(i+1),Loss)
# plt.title('Loss Function')
# plt.xlabel('Epoch')
# plt.ylabel('Loss')
# print("After\n")
# print("Weight_IH \n")
# print(Weight_IH)
# print("Weight_HH \n")
# print(Weight_HH)
# print("Weight_HO")
# print(Weight_HO)
return HO,HH,IH,Loss
#Task 2
#timea=time.time()
#train(database,database,Weight_IH,Weight_HH,Weight_HO,gama,number_of_iterations)
#timeb=time.time()
#print("Time:", timeb-timea)
#This is a helper function to determine output
def move(x):
b=list(x)
pos=b.index(max(b))
if(pos==0):
return 'R'
elif(pos==1):
return 'P'
else:
return 'S'
if input == "":
previous=np.random.choice(["R", "P", "S"])
moves=""
matches=0
temp=""
alpha=0.05
else:
if(matches<2):
moves+=previous
moves+=input
previous=np.random.choice(["R", "P", "S"])
matches+=1
else:
temp=previous
S1,A1,S2,A2,Y,previous=Foward_Propagation(encoder[moves],Weight_IH,Weight_HH,Weight_HO,alpha)
previous=move(previous)
moves+=temp
moves+=input
moves=moves[2:]
output=previous