kudkudak

      def act(self):
        if len(self.command_queue) == 0:
            if self.state == Rotator.STATE_ROTATE:
                self.command_queue.append([TURN, \
		self.state_helper * int((0.5*pi)/ TICK_ROTATE )]) # Rotate by 90 deegres
                self.state_helper *= -1
        return self.command_queue.pop(0)

kudkudak

def act(self):
    if len(self.command_queue) == 0:
        if self.state == Rotator.STATE_ROTATE:
             self.command_queue.append([TURN, \
	     self.state_helper * int((0.5*pi)/ TICK_ROTATE )]) # Rotate by 90 deegres
             self.state_helper *= -1
    return self.command_queue.pop(0)

kudkudak

 if self.state == Turner.STATE_SCANNING:
    x_disc, y_disc = int(self.x + 0.5), int(self.y + 0.5) 
    # +0.5 because axis origin is in the middle of a field

    print "Scanning on ",x_disc, " ",y_disc

    import math
    # Determine vector in which direction we are scanning
    vector = (math.cos((self.angle - 90.0)/180.0 * pi), \
math.sin((self.angle - 90.0)/180.0 * pi))

kudkudak

dataset = DenseDesignMatrix(X=X_train, y=Y_train[:,11:12].reshape(-1,1).astype("int64"))
hidden_layer = mlp.Sigmoid(layer_name='hidden', dim=1000, sparse_init =15, init_bias=1.)
output_layer = mlp.Softmax(n_classes=1, sparse_init =1, layer_name='output')
layers = [hidden_layer, output_layer]
ann = mlp.MLP(layers, nvis=X_train.shape[1])
trainer = sgd.SGD(
learning_rate=.05, batch_size=250, termination_criterion=EpochCounter(10), monitoring_dataset=dataset)
trainer.setup(ann,dataset)

while True:

kudkudak

import pandas as pd
import numpy as np

def get_city_meteo(city):
    a = [pd.read_csv(os.path.join(c["DATA_DIR"], str(city)+"-201"+str(i)+".csv"), sep=",") for i in range(4)]
    csv = pd.concat(a)
    # Do simple transformations
    columns = ["date_year", "date_month", "date_day", "date_hour", "cloud_amount", "pressure", "temperature", "wind_speed", "percipitation_amount"]
    csv.index = range(len(csv))
    csv = csv[columns]

kudkudak

import odespy
import numpy as np
dt = np.float64

mu = dt(1.0)/dt(82.45)
lamb = dt(1.0) - mu

def gprim_1(t,z):
    return z[1]

kudkudak

In file included from cec/kmeanspp_assignment.cpp:1:
In file included from ../inst/include/cec/kmeanspp_assignment.hpp:4:
In file included from /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/list:178:
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/algorithm:685:97: error: invalid operands to binary expression ('const gmum::Pair' and 'const gmum::Pair')
    _LIBCPP_INLINE_VISIBILITY bool operator()(const _T1& __x, const _T1& __y) const {return __x < __y;}
                                                                                            ~~~ ^ ~~~
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/list:2188:13: note: in instantiation of member function 'std::__1::__less<gmum::Pair, gmum::Pair>::operator()' requested here
        if (__comp(*--__e2, *__f1))
            ^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault

kudkudak

# coding: utf-8

## Praktyczny Machine Learning w Pythonie
### Czy wszystko jest poprawnie zainstalowane?

# In[14]:

import matplotlib.pylab as plt # pakiet do rysowania
import numpy as np # pakiet do manipulacji macierzami
import sklearn # scikit-learn, który będzie podstawowym tematem dzisiejszych warsztatów

kudkudak

- przypomnienie lub wprowadzenie do sieci feedforward i ich trenowanie przy pomocy stochastic gradient descent
- sieci konwolucyjnych (CNN):
	- local connectivity
	- parameter sharing
	- convolution
	- pooling / subsampling
- trenowanie CNN
- zastosowanie CNN w computer vision
	- object recognition
	- deep rank

kudkudak

Protein:  5ht1a
Satisfied? Not satisfied
Sizes of clusters [5669, 6, 12, 30, 11, 3]
-----

Protein:  5ht2a
Satisfied? Not satisfied
Sizes of clusters [2828, 191, 29, 55, 11, 6, 2]
-----