apigekko
confused about gekko's deep learning interfaces
I am beginning to use gekko to train a model for tclab. I read the pages for NN demo on keras and gekko and SISO model for tclab. Also the API doc.
I found that the API doc for GEKKO is all about brain.brain and the demos are quite different. Which one should I follow or both is OK? I think brain.brain is quite clear and will give it a try.
Regards
Tibalt
Solution
I recommend the Brain module in Gekko because it simplifies the Deep Learning models. Here is an example in Gekko with the Brain module:
from gekko import brain
import numpy as np
import matplotlib.pyplot as plt
# generate training data
x = np.linspace(0.0,2*np.pi)
y = np.sin(x)
b = brain.Brain()
b.input_layer(1)
b.layer(linear=2)
b.layer(tanh=2)
b.layer(linear=2)
b.output_layer(1)
# train
b.learn(x,y)
# validate
xp = np.linspace(-2*np.pi,4*np.pi,100)
yp = b.think(xp)
plt.figure()
plt.plot(x,y,'bo')
plt.plot(xp,yp[0],'r-')
plt.show()
Here is that same example without the brain module in Gekko. You can see that writing out all of the network equations is quite tedious but it does give you additional flexibility to use any equation form or activation function.
from gekko import GEKKO
import numpy as np
import matplotlib.pyplot as plt
# generate training data
x = np.linspace(0.0,2*np.pi,20)
y = np.sin(x)
# option for fitting function
select = True # True / False
if select:
# Size with cosine function
nin = 1 # inputs
n1 = 1 # hidden layer 1 (linear)
n2 = 1 # hidden layer 2 (nonlinear)
n3 = 1 # hidden layer 3 (linear)
nout = 1 # outputs
else:
# Size with hyperbolic tangent function
nin = 1 # inputs
n1 = 2 # hidden layer 1 (linear)
n2 = 2 # hidden layer 2 (nonlinear)
n3 = 2 # hidden layer 3 (linear)
nout = 1 # outputs
# Initialize gekko
train = GEKKO()
test = GEKKO()
model = [train,test]
for m in model:
# input(s)
m.inpt = m.Param()
# layer 1
m.w1 = m.Array(m.FV, (nin,n1))
m.l1 = [m.Intermediate(m.w1[0,i]*m.inpt) for i in range(n1)]
# layer 2
m.w2a = m.Array(m.FV, (n1,n2))
m.w2b = m.Array(m.FV, (n1,n2))
if select:
m.l2 = [m.Intermediate(sum([m.cos(m.w2a[j,i]+m.w2b[j,i]*m.l1[j]) \
for j in range(n1)])) for i in range(n2)]
else:
m.l2 = [m.Intermediate(sum([m.tanh(m.w2a[j,i]+m.w2b[j,i]*m.l1[j]) \
for j in range(n1)])) for i in range(n2)]
# layer 3
m.w3 = m.Array(m.FV, (n2,n3))
m.l3 = [m.Intermediate(sum([m.w3[j,i]*m.l2[j] \
for j in range(n2)])) for i in range(n3)]
# output(s)
m.outpt = m.CV()
m.Equation(m.outpt==sum([m.l3[i] for i in range(n3)]))
# flatten matrices
m.w1 = m.w1.flatten()
m.w2a = m.w2a.flatten()
m.w2b = m.w2b.flatten()
m.w3 = m.w3.flatten()
# Fit parameter weights
m = train
m.inpt.value=x
m.outpt.value=y
m.outpt.FSTATUS = 1
for i in range(len(m.w1)):
m.w1[i].FSTATUS=1
m.w1[i].STATUS=1
m.w1[i].MEAS=1.0
for i in range(len(m.w2a)):
m.w2a[i].STATUS=1
m.w2b[i].STATUS=1
m.w2a[i].FSTATUS=1
m.w2b[i].FSTATUS=1
m.w2a[i].MEAS=1.0
m.w2b[i].MEAS=0.5
for i in range(len(m.w3)):
m.w3[i].FSTATUS=1
m.w3[i].STATUS=1
m.w3[i].MEAS=1.0
m.options.IMODE = 2
m.options.SOLVER = 3
m.options.EV_TYPE = 2
m.solve(disp=False)
# Test sample points
m = test
for i in range(len(m.w1)):
m.w1[i].MEAS=train.w1[i].NEWVAL
m.w1[i].FSTATUS = 1
print('w1['+str(i)+']: '+str(m.w1[i].MEAS))
for i in range(len(m.w2a)):
m.w2a[i].MEAS=train.w2a[i].NEWVAL
m.w2b[i].MEAS=train.w2b[i].NEWVAL
m.w2a[i].FSTATUS = 1
m.w2b[i].FSTATUS = 1
print('w2a['+str(i)+']: '+str(m.w2a[i].MEAS))
print('w2b['+str(i)+']: '+str(m.w2b[i].MEAS))
for i in range(len(m.w3)):
m.w3[i].MEAS=train.w3[i].NEWVAL
m.w3[i].FSTATUS = 1
print('w3['+str(i)+']: '+str(m.w3[i].MEAS))
m.inpt.value=np.linspace(-2*np.pi,4*np.pi,100)
m.options.IMODE = 2
m.options.SOLVER = 3
m.solve(disp=False)
plt.figure()
plt.plot(x,y,'bo',label='data')
plt.plot(test.inpt.value,test.outpt.value,'r-',label='predict')
plt.legend(loc='best')
plt.ylabel('y')
plt.xlabel('x')
plt.show()
Here is the same example in Scikit-Learn:
from sklearn.neural_network import MLPRegressor
import numpy as np
import matplotlib.pyplot as plt
# generate training data
x = np.linspace(0.0,2*np.pi)
xr = x.reshape(-1,1)
y = np.sin(x)
# train
nn = MLPRegressor(hidden_layer_sizes=(3),
activation='tanh',\
solver='lbfgs',max_iter=2000)
model = nn.fit(xr,y)
# validate
xp = np.linspace(-2*np.pi,4*np.pi,100)
xpr = xp.reshape(-1,1)
yp = nn.predict(xpr)
ypr = yp.reshape(-1,1)
r2 = nn.score(xpr,ypr)
print('R^2: ' + str(r2))
plt.figure()
plt.plot(x,y,'bo')
plt.plot(xpr,ypr,'r-')
plt.show()
Finally, here is the same example in Keras / TensorFlow:
import numpy as np
import pandas as pd
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential
from keras.layers import *
import matplotlib.pyplot as plt
#################################################################
### Generate Data ###############################################
#################################################################
# generate training data
x = np.linspace(0.0,2*np.pi,20)
y = np.sin(x)
# save training data to file
data = np.vstack((x,y)).T
np.savetxt('train_data.csv',data,header='x,y',comments='',delimiter=',')
# generate test data
x = np.linspace(0.0,2*np.pi,100)
y = np.sin(x)
# save test data to file
data = np.vstack((x,y)).T
np.savetxt('test_data.csv',data,header='x,y',comments='',delimiter=',')
#################################################################
### Scale data ##################################################
#################################################################
# load training and test data with pandas
train_df = pd.read_csv('train_data.csv')
test_df = pd.read_csv('test_data.csv')
# scale values to 0 to 1 for the ANN to work well
s = MinMaxScaler(feature_range=(0,1))
# scale training and test data
sc_train = s.fit_transform(train_df)
sc_test = s.transform(test_df)
# print scaling adjustments
print('Scalar multipliers')
print(s.scale_)
print('Scalar minimum')
print(s.min_)
# convert scaled values back to dataframe
sc_train_df = pd.DataFrame(sc_train, columns=train_df.columns.values)
sc_test_df = pd.DataFrame(sc_test, columns=test_df.columns.values)
# save scaled values to CSV files
sc_train_df.to_csv('train_scaled.csv', index=False)
sc_test_df.to_csv('test_scaled.csv', index=False)
#################################################################
### Train model #################################################
#################################################################
# create neural network model
model = Sequential()
model.add(Dense(1, input_dim=1, activation='linear'))
model.add(Dense(2, activation='linear'))
model.add(Dense(2, activation='tanh'))
model.add(Dense(2, activation='linear'))
model.add(Dense(1, activation='linear'))
model.compile(loss="mean_squared_error", optimizer="adam")
# load training data
train_df = pd.read_csv("train_scaled.csv")
X1 = train_df.drop('y', axis=1).values
Y1 = train_df[['y']].values
# train the model
model.fit(X1,Y1,epochs=5000,verbose=0,shuffle=True)
# Save the model to hard drive
#model.save('model.h5')
#################################################################
### Test model ##################################################
#################################################################
# Load the model from hard drive
#model.load('model.h5')
# load test data
test_df = pd.read_csv("test_scaled.csv")
X2 = test_df.drop('y', axis=1).values
Y2 = test_df[['y']].values
# test the model
mse = model.evaluate(X2,Y2, verbose=1)
print('Mean Squared Error: ', mse)
#################################################################
### Predictions Outside Training Region #########################
#################################################################
# generate prediction data
x = np.linspace(-2*np.pi,4*np.pi,100)
y = np.sin(x)
# scale input
X3 = x*s.scale_[0]+s.min_[0]
# predict
Y3P = model.predict(X3)
# unscale output
yp = (Y3P-s.min_[1])/s.scale_[1]
plt.figure()
plt.plot((X1-s.min_[0])/s.scale_[0], \
(Y1-s.min_[1])/s.scale_[1], \
'bo',label='train')
plt.plot(x,y,'r-',label='actual')
plt.plot(x,yp,'k--',label='predict')
plt.legend(loc='best')
plt.savefig('results.png')
plt.show()
- How do I get currency exchange rates via an API such as Google Finance?
- Create product in printful using API - Python
- Showing data stored in room database when offline
- getStaticProps return an empty array
- Spring boot - method 'POST' is not supported (multipart/form-data)
- The instance of entity type 'IngredientEntity' cannot be tracked because another instance with the same key value for {'Id'} is already being tracked
- Can't return multi-level resource/collection through API
- How can I read the last change even if the program is closed and opened in the differences in the flutter program?
- Using body in GET requests
- How to filter filter data after extracted data in Json form
- Connecting to Channel Advisor API
- Parsing and typing API response in Dart/Flutter
- How to Import XML table Google Sheets
- How to remove a specific link preview (auto-embed) in a specific message with Discord.py bot?
- Fetch JSON from API using XSLT
- How to change weather icon from json to a weather icon I drew myself
- The rendered input is acting as object rather than string in React
- PHP switch case to update array not affecting array
- Return the expected status code in Postman
- Python POST requests to an API with JSON data work on Requestbin, but not locally
- Querying API for every unique resource in a package using R
- Flask Api working Half the time- works flawlessly when sending post with python but not with javascript
- How to get a object key in a JSON in Microsoft Flow
- How to send markdown over a REST API, and render it in the frontend (React)
- How to be notified when a droplet is active after its creation
- CustomExceptionMapper only for a specific API Rest with Jersey
- Renaming a file with a name already being downloaded through an <a> (javascript)
- passing a csv through an API
- Java receiving text file inputs in REST
- How to loop through JSON from JSON.parse?