markdown
stringlengths
0
1.02M
code
stringlengths
0
832k
output
stringlengths
0
1.02M
license
stringlengths
3
36
path
stringlengths
6
265
repo_name
stringlengths
6
127
A custom function to pick a default device
def get_default_device(): """Pick GPU if available else CPU""" if torch.cuda.is_available(): return torch.device('cuda') else: return torch.device('cpu') device = get_default_device() device def to_device(data,device): """Move tensors to chosen device""" if isinstance(data,(list,tuple)): return [t...
_____no_output_____
MIT
VGG/VGG.ipynb
gowriaddepalli/papers
Training the model
@torch.no_grad() def evaluate(model, val_loader): model.eval() outputs = [model.validation_step(batch) for batch in val_loader] return model.validation_epoch_end(outputs) def fit(epochs, lr, model, train_loader, val_loader, opt_func=torch.optim.SGD): history = [] train_losses =[] optimizer = o...
_____no_output_____
MIT
VGG/VGG.ipynb
gowriaddepalli/papers
*Accompanying code examples of the book "Introduction to Artificial Neural Networks and Deep Learning: A Practical Guide with Applications in Python" by [Sebastian Raschka](https://sebastianraschka.com). All code examples are released under the [MIT license](https://github.com/rasbt/deep-learning-book/blob/master/LICEN...
%load_ext watermark %watermark -a 'Sebastian Raschka' -v -p torch
Sebastian Raschka CPython 3.6.6 IPython 7.1.1 torch 0.4.1
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Model Zoo -- CNN Gender Classifier (ResNet-50 Architecture, CelebA) with Data Parallelism Network Architecture The network in this notebook is an implementation of the ResNet-50 [1] architecture on the CelebA face dataset [2] to train a gender classifier. References - [1] He, K., Zhang, X., Ren, S., & Sun, J. (20...
import os import time import numpy as np import pandas as pd import torch import torch.nn as nn import torch.nn.functional as F from torch.utils.data import Dataset from torch.utils.data import DataLoader from torchvision import datasets from torchvision import transforms import matplotlib.pyplot as plt from PIL i...
_____no_output_____
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Dataset Downloading the Dataset Note that the ~200,000 CelebA face image dataset is relatively large (~1.3 Gb). The download link provided below was provided by the author on the official CelebA website at http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html. 1) Download and unzip the file `img_align_celeba.zip`, which ...
df1 = pd.read_csv('list_attr_celeba.txt', sep="\s+", skiprows=1, usecols=['Male']) # Make 0 (female) & 1 (male) labels instead of -1 & 1 df1.loc[df1['Male'] == -1, 'Male'] = 0 df1.head() df2 = pd.read_csv('list_eval_partition.txt', sep="\s+", skiprows=0, header=None) df2.columns = ['Filename', 'Partition'] df2 = df2....
(218, 178, 3)
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Implementing a Custom DataLoader Class
class CelebaDataset(Dataset): """Custom Dataset for loading CelebA face images""" def __init__(self, csv_path, img_dir, transform=None): df = pd.read_csv(csv_path, index_col=0) self.img_dir = img_dir self.csv_path = csv_path self.img_names = df.index.values self.y =...
Epoch: 1 | Batch index: 0 | Batch size: 1024 Epoch: 2 | Batch index: 0 | Batch size: 1024
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Model
########################## ### SETTINGS ########################## # Hyperparameters random_seed = 1 learning_rate = 0.001 num_epochs = 5 # Architecture num_features = 128*128 num_classes = 2
_____no_output_____
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
The following code cell that implements the ResNet-34 architecture is a derivative of the code provided at https://pytorch.org/docs/0.4.0/_modules/torchvision/models/resnet.html.
########################## ### MODEL ########################## def conv3x3(in_planes, out_planes, stride=1): """3x3 convolution with padding""" return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride, padding=1, bias=False) class Bottleneck(nn.Module): expansion = 4 ...
Using 4 GPUs
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Training
def compute_accuracy(model, data_loader): correct_pred, num_examples = 0, 0 for i, (features, targets) in enumerate(data_loader): features = features.to(device) targets = targets.to(device) logits, probas = model(features) _, predicted_labels = torch.max(probas, 1) ...
Epoch: 001/005 | Batch 0000/0158 | Cost: 0.7133 Epoch: 001/005 | Batch 0050/0158 | Cost: 0.1586 Epoch: 001/005 | Batch 0100/0158 | Cost: 0.1041 Epoch: 001/005 | Batch 0150/0158 | Cost: 0.1345 Epoch: 001/005 | Train: 93.080% | Valid: 94.050% Time elapsed: 2.74 min Epoch: 002/005 | Batch 0000/0158 | Cost: 0.1176 Epoch: 0...
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Evaluation
with torch.set_grad_enabled(False): # save memory during inference print('Test accuracy: %.2f%%' % (compute_accuracy(model, test_loader))) for batch_idx, (features, targets) in enumerate(test_loader): features = features targets = targets break plt.imshow(np.transpose(features[0], (1, 2, 0))) mode...
Probability Female 99.19%
MIT
code/model_zoo/pytorch_ipynb/convnet-resnet50-celeba-dataparallel.ipynb
wpsliu123/Sebastian_Raschka-Deep-Learning-Book
Filtering data
df.filter('Close>500').show() df.filter('Close>500').select('Open').show() df.filter((df["Close"]> 500) &(df["Open"]< 495)).show() df.filter((df["Close"]>200) | (df["Open"]< 200)).show() result = df.filter(df["open"] == 208.330002 ).collect() type(result[0]) row = result[0] row.asDict() for item in result[0]: prin...
2010-01-19 00:00:00 208.330002 215.18999900000003 207.240004 215.039995 182501900 27.860484999999997
MIT
spark/SparkBasic/DataFrames_Basic_Operations.ipynb
AlphaSunny/RecSys
Examples of matrix products
import pyJvsip as pjv
_____no_output_____
MIT
doc/notebooks/MatrixProduct.ipynb
rrjudd/jvsip
Example of matrix product prod
inA=pjv.create('mview_d',2,5).randn(5) inB=pjv.create('mview_d',5,5).identity outC=inA.prod(inB) outC.mprint('%.3f') print("Frobenius of difference %.2f"%(inA-outC).normFro) inB=pjv.create('mview_d',2,2).identity outC=inB.prod(inA) outC.mprint('%.3f') print("Frobenius of difference %.2f"%(inA-outC).normFro)
[ 0.508 0.535 0.699 -0.960 0.231; 0.040 -0.477 0.208 0.506 -0.383] Frobenius of difference 0.00
MIT
doc/notebooks/MatrixProduct.ipynb
rrjudd/jvsip
Example of prodjConjugate matrix product
inA=pjv.create('cmview_f',3,4).randn(3) inB=pjv.create('cmview_f',4,2).randn(4) outC=inA.prodj(inB) print('C=A.prodj(B)'); print('A');inA.mprint('%.3f') print('B');inB.mprint('%.3f') print('C');outC.mprint('%.3f') print('test using prod and inB.conj');pjv.prod(inA,(inB.conj),outC).mprint('%.3f')
test using prod and inB.conj [ 1.384-2.247i -0.287-2.598i; 2.453+1.101i 0.404+0.091i; -1.963+3.397i -3.496+0.722i]
MIT
doc/notebooks/MatrixProduct.ipynb
rrjudd/jvsip
Example of prodhHermitian matrix product
inA=pjv.create('cmview_f',3,4).randn(3) inB=pjv.create('cmview_f',2,4).randn(4) outC=inA.prodh(inB) print('C=A.prodj(B)'); print('A');inA.mprint('%.3f') print('B');inB.mprint('%.3f') print('C');outC.mprint('%.3f') print('test using prod and inB.herm');pjv.prod(inA,(inB.herm),outC).mprint('%.3f')
test using prod and inB.herm [-2.193+0.832i -0.311+4.938i; 0.322-0.693i -3.759-1.877i; -0.758+1.496i -2.282+2.687i]
MIT
doc/notebooks/MatrixProduct.ipynb
rrjudd/jvsip
Example of prodtTranspose matrix product.
inA=pjv.create('cmview_f',3,4).randn(3) inB=pjv.create('cmview_f',2,4).randn(4) outC=inA.prodt(inB) print('C=A.prodj(B)'); print('A');inA.mprint('%.3f') print('B');inB.mprint('%.3f') print('C');outC.mprint('%.3f') print('test using prod and inB.herm');pjv.prod(inA,(inB.transview),outC).mprint('%.3f') inA=pjv.create('mv...
[ 0.555 -1.017 -0.534 0.872 -1.029 1.373 -1.534 -2.648 -0.214 2.112; -0.796 -0.389 0.591 -0.835 -0.654 0.024 3.188 2.993 0.176 -2.690; -0.202 1.608 -0.422 0.629 1.406 -1.663 -2.301 -2.322 -0.565 2.163]
MIT
doc/notebooks/MatrixProduct.ipynb
rrjudd/jvsip
Priority Queue Reference Implementation Operations:For the sake of simplicity, all inputs assumed to be valid**enqueue(data, priority)*** Insert data to the priority queue**dequeue()*** Remove one node from the priority queue with highest priority* If queue is empty, return None
class Node(object): def __init__(self, data, priority, next=None): self.data = data self.priority = priority self.next = next class PriorityQueue(object): def __init__(self): self.head = None def enqueue(self, data, priority): if self.head is None: se...
2 with priority of 30 1 with priority of 20 3 with priority of 15 Dequeue 2 with priority of 30 1 with priority of 20 3 with priority of 15
MIT
DataStructures/PriorityQueue.ipynb
varian97/ComputerScience-Notebook
May
features_hierarchical_may, transformed_tokens_may, linkage_matrix_may, clusters_may = hierarchical_clustering(best_model_may, tfidf_matrix_may, 2) agglomerative_clustering(6, features_hierarchical_may, df_may, 2, best_model_may, transformed_tokens_may, clusters_may) elbow_method(tfidf_matrix_may[clusters_may==2], linka...
_____no_output_____
MIT
Mediacloud_Hierarchical_clustering.ipynb
gesiscss/media_frames
September
features_hierarchical_sep, transformed_tokens_sep, linkage_matrix_sep, clusters_sep = hierarchical_clustering(best_model_sep, tfidf_matrix_sep, 10) agglomerative_clustering(2, features_hierarchical_sep, df_sep, 10, best_model_sep, transformed_tokens_sep, clusters_sep)
_____no_output_____
MIT
Mediacloud_Hierarchical_clustering.ipynb
gesiscss/media_frames
REINFORCE in PyTorchJust like we did before for Q-learning, this time we'll design a PyTorch network to learn `CartPole-v0` via policy gradient (REINFORCE).Most of the code in this notebook is taken from approximate Q-learning, so you'll find it more or less familiar and even simpler.
import sys, os if 'google.colab' in sys.modules and not os.path.exists('.setup_complete'): !wget -q https://raw.githubusercontent.com/yandexdataschool/Practical_RL/master/setup_colab.sh -O- | bash !touch .setup_complete # This code creates a virtual display to draw game images on. # It will have no effect if y...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
A caveat: with some versions of `pyglet`, the following cell may crash with `NameError: name 'base' is not defined`. The corresponding bug report is [here](https://github.com/pyglet/pyglet/issues/134). If you see this error, try restarting the kernel.
env = gym.make("CartPole-v0") # gym compatibility: unwrap TimeLimit if hasattr(env, '_max_episode_steps'): env = env.env env.reset() n_actions = env.action_space.n state_dim = env.observation_space.shape plt.imshow(env.render("rgb_array"))
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Building the network for REINFORCE For REINFORCE algorithm, we'll need a model that predicts action probabilities given states.For numerical stability, please __do not include the softmax layer into your network architecture__.We'll use softmax or log-softmax where appropriate.
import torch import torch.nn as nn # Build a simple neural network that predicts policy logits. # Keep it simple: CartPole isn't worth deep architectures. model = nn.Sequential( <YOUR CODE: define a neural network that predicts policy logits> )
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Predict function Note: output value of this function is not a torch tensor, it's a numpy array.So, here gradient calculation is not needed.Use [no_grad](https://pytorch.org/docs/stable/autograd.htmltorch.autograd.no_grad)to suppress gradient calculation.Also, `.detach()` (or legacy `.data` property) can be used instea...
def predict_probs(states): """ Predict action probabilities given states. :param states: numpy array of shape [batch, state_shape] :returns: numpy array of shape [batch, n_actions] """ # convert states, compute logits, use softmax to get probability <YOUR CODE> return <YOUR CODE> test_s...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Play the gameWe can now use our newly built agent to play the game.
def generate_session(env, t_max=1000): """ Play a full session with REINFORCE agent. Returns sequences of states, actions, and rewards. """ # arrays to record session states, actions, rewards = [], [], [] s = env.reset() for t in range(t_max): # action probabilities array aka p...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Computing cumulative rewards$$\begin{align*}G_t &= r_t + \gamma r_{t + 1} + \gamma^2 r_{t + 2} + \ldots \\&= \sum_{i = t}^T \gamma^{i - t} r_i \\&= r_t + \gamma * G_{t + 1}\end{align*}$$
def get_cumulative_rewards(rewards, # rewards at each step gamma=0.99 # discount for reward ): """ Take a list of immediate rewards r(s,a) for the whole session and compute cumulative returns (a.k.a. G(s,a) in Sutton '16). G_t = r_t + gamma*r...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Loss function and updatesWe now need to define objective and update over policy gradient.Our objective function is$$ J \approx { 1 \over N } \sum_{s_i,a_i} G(s_i,a_i) $$REINFORCE defines a way to compute the gradient of the expected reward with respect to policy parameters. The formula is as follows:$$ \nabla_\theta ...
def to_one_hot(y_tensor, ndims): """ helper: take an integer vector and convert it to 1-hot matrix. """ y_tensor = y_tensor.type(torch.LongTensor).view(-1, 1) y_one_hot = torch.zeros( y_tensor.size()[0], ndims).scatter_(1, y_tensor, 1) return y_one_hot # Your code: define optimizers optimizer = ...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
The actual training
for i in range(100): rewards = [train_on_session(*generate_session(env)) for _ in range(100)] # generate new sessions print("mean reward:%.3f" % (np.mean(rewards))) if np.mean(rewards) > 500: print("You Win!") # but you can train even further break
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Results & video
# Record sessions import gym.wrappers with gym.wrappers.Monitor(gym.make("CartPole-v0"), directory="videos", force=True) as env_monitor: sessions = [generate_session(env_monitor) for _ in range(100)] # Show video. This may not work in some setups. If it doesn't # work for you, you can download the videos and view...
_____no_output_____
Unlicense
week06_policy_based/reinforce_pytorch.ipynb
RomaKoks/Practical_RL
Equivalent layer technique for estimating total magnetization direction: Analysis of the result Importing libraries
% matplotlib inline import sys import numpy as np import matplotlib.pyplot as plt import matplotlib.mlab as mlab import cPickle as pickle import datetime import timeit import string as st from mpl_toolkits.axes_grid1.inset_locator import inset_axes from fatiando.gridder import regular notebook_name = 'airborne_EQL_ma...
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Plot style
plt.style.use('ggplot')
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Importing my package
dir_modules = '../../../mypackage' sys.path.append(dir_modules) import auxiliary_functions as fc
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Loading model
with open('data/model_multi.pickle') as f: model_multi = pickle.load(f)
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Loading observation points
with open('data/airborne_survey.pickle') as f: airborne = pickle.load(f)
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Loading data set
with open('data/data_set.pickle') as f: data = pickle.load(f)
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Loading results
with open('data/result_RM_airb.pickle') as f: results = pickle.load(f)
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
List of saved files
saved_files = []
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Observation area
print 'Area limits: \n x_max = %.1f m \n x_min = %.1f m \n y_max = %.1f m \n y_min = %.1f m' % (airborne['area'][1], airborne['area'][0], ...
Area limits: x_max = 5500.0 m x_min = -6500.0 m y_max = 6500.0 m y_min = -5500.0 m
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Airborne survey information
print 'Shape : (%.0f,%.0f)'% airborne['shape'] print 'Number of data: %.1f' % airborne['N'] print 'dx: %.1f m' % airborne['dx'] print 'dy: %.1f m ' % airborne['dy']
Shape : (49,25) Number of data: 1225.0 dx: 250.0 m dy: 500.0 m
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Properties of the model Main field
inc_gf,dec_gf = model_multi['main_field'] print'Main field inclination: %.1f degree' % inc_gf print'Main field declination: %.1f degree' % dec_gf
Main field inclination: -40.0 degree Main field declination: -22.0 degree
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Magnetization direction
print 'Inclination: %.1f degree' % model_multi['inc_R'] print 'Declination: %.1f degree' % model_multi['dec_R'] inc_R,dec_R = model_multi['inc_R'],model_multi['dec_R']
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Coordinates equivalent sources
h = results['layer_depth'] shape_layer = (airborne['shape'][0],airborne['shape'][1]) xs,ys,zs = regular(airborne['area'],shape_layer,h)
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
The best solution using L-curve
m_LM = results['magnetic_moment'][4] inc_est = results['inc_est'][4] dec_est = results['dec_est'][4] mu = results['reg_parameter'][4] phi = results['phi'][4] print mu
350000.0
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Visualization of the convergence
phi = (np.array(phi)/airborne['x'].size) title_font = 22 bottom_font = 20 saturation_factor = 1. plt.close('all') plt.figure(figsize=(10,10), tight_layout=True) plt.plot(phi,'b-',linewidth=1.5) plt.title('Convergence', fontsize=title_font) plt.xlabel('iteration', fontsize = title_font) plt.ylabel('Goal function ', fon...
/home/andrelreis/anaconda3/envs/py2/lib/python2.7/site-packages/matplotlib/figure.py:2299: UserWarning: This figure includes Axes that are not compatible with tight_layout, so results might be incorrect. warnings.warn("This figure includes Axes that are not compatible "
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Estimated magnetization direction
print (inc_est,dec_est) print (inc_R,dec_R)
(-25.0, 30.0)
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Comparison between observed data and predicted data
pred = fc.tfa_layer(airborne['x'],airborne['y'],airborne['z'], xs,ys,zs,inc_gf,dec_gf,m_LM,inc_est,dec_est) res = pred - data['tfa_obs_RM_airb'] r_norm,r_mean,r_std = fc.residual(data['tfa_obs_RM_airb'],pred) title_font = 22 bottom_font = 20 plt.figure(figsize=(28,11), tight_layout=True) ranges = n...
/home/andrelreis/anaconda3/envs/py2/lib/python2.7/site-packages/matplotlib/axes/_axes.py:6571: UserWarning: The 'normed' kwarg is deprecated, and has been replaced by the 'density' kwarg. warnings.warn("The 'normed' kwarg is deprecated, and has been " /home/andrelreis/anaconda3/envs/py2/lib/python2.7/site-packages/ip...
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Positive magnetic-moment distribution
title_font = 22 bottom_font = 20 plt.close('all') plt.figure(figsize=(10,10), tight_layout=True) plt.title('Magnetic moment distribution', fontsize=title_font) plt.contourf(1e-3*ys.reshape(shape_layer),1e-3*xs.reshape(shape_layer), m_LM.reshape(shape_layer), 40, cmap='inferno') cb = plt.colorbar(pad=0.01...
_____no_output_____
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Figure for paper
#title_font = 17 title_font = 5 #bottom_font = 14 bottom_font = 4 hist_font = 5 height_per_width = 17./15. plt.figure(figsize=(4.33,4.33*height_per_width), tight_layout=True) ranges = np.abs([data['tfa_obs_RM_airb'].max(), data['tfa_obs_RM_airb'].min(), pred.max(), pred.min()]).max()...
/home/andrelreis/anaconda3/envs/py2/lib/python2.7/site-packages/ipykernel_launcher.py:71: MatplotlibDeprecationWarning: scipy.stats.norm.pdf
BSD-3-Clause
code/notebooks/synthetic_tests/model_multibody_shallow-seated/airborne_EQL_magdirection_RM_analysis.ipynb
pinga-lab/eqlayer-magnetization-direction
Parsing Natural Language in Python **(C) 2018 by [Damir Cavar](http://damir.cavar.me/)** **License:** [Creative Commons Attribution-ShareAlike 4.0 International License](https://creativecommons.org/licenses/by-sa/4.0/) ([CA BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)) This is a tutorial related to the ...
import sys
_____no_output_____
Apache-2.0
notebooks/Parsing Natural Language in Python.ipynb
peey/python-tutorial-notebooks
The Grammar Class Let us assume that our Phrase Structure Grammar consists of rules that contain one symbol in the Left-Hand Side, followed by a production symbol, an arrow, and by a list of at least one terminal and symbol. Comments can be introduced using the ** symbol. Every rule has to be contained in one line.
grammarText = """ # PSG1 # small English grammar # (C) 2005 by Damir Cavar, Indiana University # Grammar: S -> NP VP NP -> N NP -> Adj N NP -> Art Adj N NP -> Art N NP -> Art N PP #NP -> Art N NP VP -> V VP -> V NP VP -> Adv V NP VP -> V PP VP -> V NP PP PP -> P NP # Lexicon: N -> John N -> Mary N -> bench...
_____no_output_____
Apache-2.0
notebooks/Parsing Natural Language in Python.ipynb
peey/python-tutorial-notebooks
We can parse this grammar into a representation that allows us to fetch the left- and the right-hand side of a rule for top- or bottom-up parsing.
class PSG: def __init__(self, grammar): self.LHS = {} self.RHS = {} self.__read__(grammar) def __str__(self): text = "" for i in self.LHS.keys(): if len(text) > 0: text += "\n" for x in self.LHS[i]: text += i + " ->...
_____no_output_____
Apache-2.0
notebooks/Parsing Natural Language in Python.ipynb
peey/python-tutorial-notebooks
The grammar file: The Top-Down Parser Defining some parameters:
LIFO = -1 FIFO = 0 strategy = FIFO def tdparse(inp, goal, grammar, agenda): print("Got : %s\tinput: %s" % (goal, inp)) if goal == inp == []: print("Success") elif goal == [] or inp == []: if agenda == []: print("Fail: Agenda empty!") else: entry = agenda.pop(strategy) ...
S -> NP VP NP -> N NP -> Adj N NP -> Art Adj N NP -> Art N NP -> Art N PP VP -> V VP -> V NP VP -> Adv V NP VP -> V PP VP -> V NP PP PP -> P NP N -> John N -> Mary N -> bench N -> cat N -> mouse Art -> the Art -> a Adj -> green Adj -> yellow Adj -> big Adj -> small Adv -> often Adv -> yesterday V -> kissed V ->...
Apache-2.0
notebooks/Parsing Natural Language in Python.ipynb
peey/python-tutorial-notebooks
Mask R-CNN DemoA quick intro to using the pre-trained model to detect and segment objects.
import os import sys import random import math import numpy as np import skimage.io import matplotlib import matplotlib.pyplot as plt import cv2,time,json,glob from IPython.display import clear_output # Root directory of the project ROOT_DIR = os.path.abspath("../") # Import Mask RCNN sys.path.append(ROOT_DIR) # To...
tensorflow version: 1.15.0 using gpu: True
MIT
samples/demo.ipynb
jtchilders/Mask_RCNN
ConfigurationsWe'll be using a model trained on the MS-COCO dataset. The configurations of this model are in the ```CocoConfig``` class in ```coco.py```.For inferencing, modify the configurations a bit to fit the task. To do so, sub-class the ```CocoConfig``` class and override the attributes you need to change.
class InferenceConfig(coco.CocoConfig): # Set batch size to 1 since we'll be running inference on # one image at a time. Batch size = GPU_COUNT * IMAGES_PER_GPU GPU_COUNT = 1 IMAGES_PER_GPU = 10 BATCH_SIZE=10 config = InferenceConfig() config.display()
_____no_output_____
MIT
samples/demo.ipynb
jtchilders/Mask_RCNN
Create Model and Load Trained Weights
# Create model object in inference mode. model = modellib.MaskRCNN(mode="inference", model_dir=MODEL_DIR, config=config) # Load weights trained on MS-COCO model.load_weights(COCO_MODEL_PATH, by_name=True)
_____no_output_____
MIT
samples/demo.ipynb
jtchilders/Mask_RCNN
Class NamesThe model classifies objects and returns class IDs, which are integer value that identify each class. Some datasets assign integer values to their classes and some don't. For example, in the MS-COCO dataset, the 'person' class is 1 and 'teddy bear' is 88. The IDs are often sequential, but not always. The CO...
# COCO Class names # Index of the class in the list is its ID. For example, to get ID of # the teddy bear class, use: class_names.index('teddy bear') class_names = ['BG', 'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus', 'train', 'truck', 'boat', 'traffic light', 'fire hydrant',...
_____no_output_____
MIT
samples/demo.ipynb
jtchilders/Mask_RCNN
Run Object Detection
# Load a random image from the images folder file_names = next(os.walk(IMAGE_DIR))[2] image = skimage.io.imread(os.path.join(IMAGE_DIR, random.choice(file_names))) # Run detection results = model.detect([image], verbose=1) # Visualize results r = results[0] visualize.display_instances(image, r['rois'], r['masks'], r[...
files: 345 0 of 345 frames per second: 25 WARNING:tensorflow:From /home/jchilders/conda/mask_rcnn/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py:422: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.
MIT
samples/demo.ipynb
jtchilders/Mask_RCNN
The perceptron - Recognising the MNIST digits Table of contents
%matplotlib inline from pylab import * from utils import *
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Let us implement a perceptron that categorize the MNIST images as numbers. As you will see below the behaviour of the network is far from optimal. As we see the network [learns well the training set](Plotting-the-results-of-test). Nevertheless its behaviour in a [test with new digits](Spreading-of-the-network-during-t...
#----------------------------------------------------------- # training # Set the number of patterns n_patterns = 500 # Take 'n_patterns' rows indices = arange(training_length) shuffle(indices) indices = indices[:n_patterns] # Get patterns patterns = array(mndata.train_images)[indices] # Rescale all patterns betwe...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Let us visualize the first 20 patterns of the trining set:
for i in xrange(20): # Create a new figure after each 10-th item if i%10 == 0: fig = figure(figsize = (20, 1)) # Plot current item (we use the # function plot_img in our utils.py) plot_img( to_mat(patterns[i]), fig, (i%10)+1, windows = 20 ) # show figure after all 1o ...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Spreading of the network during trainingHere starts the core part, iterating the timesteps. We also divide the training phase in epochs. Each epoch is a single presentation of the whole input pattern series. The sum of squared errors will be grouped by epochs.
# counter of repetitions # of the series of patterns epoch = -1 # Iterate trials for t in xrange(stime) : # Reiterate the input pattern # sequence through timesteps # Reshuffle at the end # of the series if t%n_patterns == 0: shuffle(pattern_indices) epoch += 1 ...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Plotting the results of trainingWe plot the history of the squared errors through epocs a
fig = figure() ax = fig.add_subplot(111) ax.plot(squared_errors) xlabel("Epochs") ylabel("Sum of squared errors")
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
and a visualization of the weights to each ouput unit. Each set of weights seems to reproduce (in a very raugh manner) a generalization of the target digit.
figure(figsize=(15,2)) for i in xrange(m) : subplot(1,m,i+1) title(i) im = to_mat(w[i,1:]) imshow(im, cmap=cm.bone) axis('off') show()
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Testing Initializing data and parametersNow we create a new dataset to test the network and reset some variables:
#----------------------------------------------------------- # test # Set the number of patterns n_patterns = 1000 # Take 'n_patterns' rows indices = arange(test_length) shuffle(indices) indices = indices[:n_patterns] # Get patterns patterns = array(mndata.test_images)[indices] # Rescale all patterns between 0 and...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Spreading of the network during testThe network react to each test pattern in each spreading timestep:
# Iterate trials for p in xrange(n_patterns) : # Aggregate inputs and the bias unit x = hstack([ 1, patterns[p] ]) # !!!! The dot product becomes a matrix # product with more than one output unit !!!! net = dot(w,x) # output function y = step(net) y_index = squeez...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Let us see what is the proportion of correct answers of the network:
print "Proportion of correct answers:{}" \ .format(sum(error_store)/float(n_patterns))
Proportion of correct answers:0.655
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
Plotting the results of testNow we plot few test samples to get the real idea. For each sample we plot the input digit on the top, the answer of the network on the center and the target digit on the left. Squared brakets indicate that the network gave zero or more than one answer.
import matplotlib.gridspec as gridspec gs = gridspec.GridSpec(8, 4*10) n_patterns = 20 for p in xrange(n_patterns) : im = to_mat(input_store[:,p]) k = p%10 if k==0 : fig = figure(figsize=(15,2)) ax1 = fig.add_subplot(gs[:4,(k*4):(k*4+4)]) ax1.imshow(im, cmap...
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
The next cell is just for styling
from IPython.core.display import HTML def css_styling(): styles = open("../style/ipybn.css", "r").read() return HTML(styles) css_styling()
_____no_output_____
MIT
course/perceptron-MNIST-simulation.ipynb
cecconeurale/neunet-basics
BE 240 Lecture 4 Sub-SBML Modeling diffusion, shared resources, and compartmentalized systems _Ayush Pandey_
# This notebook is designed to be converted to a HTML slide show # To do this in the command prompt type (in the folder containing the notebook): # jupyter nbconvert BE240_Lecture4_Sub-SBML.ipynb --to slides
_____no_output_____
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
![image.png](attachment:image.png) ![image.png](attachment:image.png) An example: Three different "subsystems" - each with its SBML model Another "signal in mixture" subsystem - models signal in the environment / mixture Using Sub-SBML we can obtain the combined model for such a system with* transport across membrane...
# Import statements from subsbml.Subsystem import createNewSubsystem, createSubsystem import numpy as np import pylab as plt
_____no_output_____
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
Transcription Model:Consider the following simple transcription-only model where $G$ is a gene, $T$ is a transcript, and $S$ is the signaling molecule.We can write the following reduced order dynamics:1. $G \xrightarrow[]{\rho_{tx}(G, S)} G + T$; \begin{align} \rho_{tx}(G, S) = G K_{X}\frac{S^{2}}{K_{S}^{2}+S^{2}}\\\e...
# Import SBML models by creating Subsystem class objects ss1 = createSubsystem('transcription_SBML_model.xml') ss2 = createSubsystem('translation_SBML_model.xml') ss1.renameSName('mRNA_T', 'T') # Combine the two subsystems together tx_tl_subsystem = ss1 + ss2 # The longer way to do the same thing: # tx_tl_subsystem...
{'default_bioscrape_generated_model_47961': 'default_bioscrape_generated_model_245758_combined', 'default_bioscrape_generated_model_245758': 'default_bioscrape_generated_model_245758_combined', 'mRNA_T_bioscrape_generated_model_245758': 'mRNA_T_bioscrape_generated_model_245758_1_combined', 'T_bioscrape_generated_model_...
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
Signal induction model:1. $\varnothing \xrightarrow[]{\rho(I)} S$; \begin{align} \rho(S) = K_{0} \frac{I^2}{K_{I} + I^2}\\\end{align}Here $S$ is the signal produced on induction by an inducer $I$.The lumped parameters $K_{0}$ and $K_S$ model effects of cooperative production of the signal by the inducer. This is the s...
ss3 = createSubsystem('signal_in_mixture.xml') # Signal subsystem (production of signal molecule) combined_ss = ss1 + ss2 + ss3 # Alternatively combined_ss = createNewSubsystem() combined_ss.combineSubsystems([ss1,ss2,ss3]) # Writing a Subsystem to an SBML file (Export SBML) combined_ss.writeSBML('txtl_combined.xml...
_____no_output_____
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
What does Sub-SBML look for?1. For compartments: if two compartments have the same `name` and the same `size` attributes => they are combined together.1. For species: if two species have the same `name` attribute => they are combined together. If initial amount is not the same, the first amount is set. It is easy to s...
from subsbml.System import System, combineSystems cell_1 = System('cell_1') ss1 = createSubsystem('txtl_ss.xml') ss1.renameSName('S', 'IPTG') ss2 = createSubsystem('IPTG_reservoir.xml') IPTG_external_conc = ss2.getSpeciesByName('IPTG').getInitialConcentration() cell_1.setInternal([ss1]) cell_1.setExternal([ss2]) # ...
_____no_output_____
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
System 2 : TX-TL with IPTG reservoir and a simple membrane Membrane : IPTG external and internal diffusion in a one step reversible reaction
from subsbml import System, createSubsystem, combineSystems, createNewSubsystem ss1 = createSubsystem('txtl_ss.xml') ss1.renameSName('S','IPTG') ss2 = createSubsystem('IPTG_reservoir.xml') # Create a simple IPTG membrane where IPTG goes in an out of the membrane via a reversible reaction mb2 = createSubsystem('membran...
The subsystem from membrane_IPTG.xml has multiple compartments
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
System 3 : TX-TL with IPTG reservoir and a detailed membrane diffusion Membrane : IPTG external binds to a transport protein and forms a complex. This complex causes the diffusion of IPTG in the internal of the cell.
# Create a more detailed IPTG membrane where IPTG binds to an intermediate transporter protein, forms a complex # then transports out of the cell system to the external environment mb3 = createSubsystem('membrane_IPTG_detailed.xml', membrane = True) cell_3 = System('cell_3',ListOfInternalSubsystems = [ss1], ...
_____no_output_____
BSD-3-Clause
examples/BE240_Lecture4_Sub-SBML.ipynb
BuildACell/subsbml
[Table of Contents](http://nbviewer.ipython.org/github/rlabbe/Kalman-and-Bayesian-Filters-in-Python/blob/master/table_of_contents.ipynb) The Extended Kalman Filter
#format the book %matplotlib inline from __future__ import division, print_function from book_format import load_style load_style()
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
At this point in the book we have developed the theory for the linear Kalman filter. Then, in the last two chapters we broached the topic of using Kalman filters for nonlinear problems. In this chapter we will learn the Extended Kalman filter (EKF). The EKF handles nonlinearity by linearizing the system at the point of...
import ekf_internal ekf_internal.show_linearization()
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
If the curve above is the process model, then the dotted lines shows the linearization of that curve for the estimate $x=1.5$.We linearize systems by finding the slope of the curve at the given point:$$\begin{aligned}f(x) &= x^2 -2x \\\frac{df}{dx} &= 2x - 2\end{aligned}$$and then finding its value at the evaluation po...
import ekf_internal ekf_internal.show_radar_chart()
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
This gives us the equality $x=\sqrt{slant^2 - altitude^2}$. Design the State Variables We want to track the position of an aircraft assuming a constant velocity and altitude, and measurements of the slant distance to the aircraft. That means we need 3 state variables - horizontal distance, horizonal velocity, and alt...
from math import sqrt def HJacobian_at(x): """ compute Jacobian of H matrix at x """ horiz_dist = x[0] altitude = x[2] denom = sqrt(horiz_dist**2 + altitude**2) return array ([[horiz_dist/denom, 0., altitude/denom]])
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Finally, let's provide the code for $h(\mathbf x)$
def hx(x): """ compute measurement for slant range that would correspond to state x. """ return (x[0]**2 + x[2]**2) ** 0.5
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Now lets write a simulation for our radar.
from numpy.random import randn import math class RadarSim(object): """ Simulates the radar signal returns from an object flying at a constant altityude and velocity in 1D. """ def __init__(self, dt, pos, vel, alt): self.pos = pos self.vel = vel self.alt = alt self....
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Design Process and Measurement NoiseThe radar returns the range distance. A good radar can achieve accuracy of $\sigma_{range}= 5$ meters, so we will use that value. This gives us$$\mathbf R = \begin{bmatrix}\sigma_{range}^2\end{bmatrix} = \begin{bmatrix}25\end{bmatrix}$$The design of $\mathbf Q$ requires some discuss...
from filterpy.common import Q_discrete_white_noise from filterpy.kalman import ExtendedKalmanFilter from numpy import eye, array, asarray import numpy as np dt = 0.05 rk = ExtendedKalmanFilter(dim_x=3, dim_z=1) radar = RadarSim(dt, pos=0., vel=100., alt=1000.) # make an imperfect starting guess rk.x = array([radar.po...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Using SymPy to compute Jacobians Depending on your experience with derivatives you may have found the computation of the Jacobian difficult. Even if you found it easy, a slightly more difficult problem easily leads to very difficult computations.As explained in Appendix A, we can use the SymPy package to compute the J...
import sympy sympy.init_printing(use_latex=True) x, x_vel, y = sympy.symbols('x, x_vel y') H = sympy.Matrix([sympy.sqrt(x**2 + y**2)]) state = sympy.Matrix([x, x_vel, y]) H.jacobian(state)
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
This result is the same as the result we computed above, and with much less effort on our part! Robot LocalizationSo, time to try a real problem. I warn you that this is far from a simple problem. However, most books choose simple, textbook problems with simple answers, and you are left wondering how to implement a re...
ekf_internal.plot_bicycle()
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
In the **Unscented Kalman Filter** chapter we derived these equations describing for this model:$$\begin{aligned} x &= x - R\sin(\theta) + R\sin(\theta + \beta) \\y &= y + R\cos(\theta) - R\cos(\theta + \beta) \\\theta &= \theta + \beta\end{aligned}$$where $\theta$ is the robot's heading.You do not need to understand t...
import sympy from sympy.abc import alpha, x, y, v, w, R, theta from sympy import symbols, Matrix sympy.init_printing(use_latex="mathjax", fontsize='16pt') time = symbols('t') d = v*time beta = (d/w)*sympy.tan(alpha) r = w/sympy.tan(alpha) fxu = Matrix([[x-r*sympy.sin(theta) + r*sympy.sin(theta+beta)], [y...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
That looks a bit complicated. We can use SymPy to substitute terms:
# reduce common expressions B, R = symbols('beta, R') J = J.subs((d/w)*sympy.tan(alpha), B) J.subs(w/sympy.tan(alpha), R)
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
In that form we can see that our computation of the Jacobian is correct.Now we can turn our attention to the noise. Here, the noise is in our control input, so it is in *control space*. In other words, we command a specific velocity and steering angle, but we need to convert that into errors in $x, y, \theta$. In a rea...
V = fxu.jacobian(Matrix([v, alpha])) V = V.subs(sympy.tan(alpha)/w, 1/R) V = V.subs(time*v/R, B) V = V.subs(time*v, 'd') V
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
This should give you an appreciation of how quickly the EKF become mathematically intractable. This gives us the final form of our prediction equations:$$\begin{aligned}\mathbf{\overline x} &= \mathbf x + \begin{bmatrix}- R\sin(\theta) + R\sin(\theta + \beta) \\R\cos(\theta) - R\cos(\theta + \beta) \\\beta\end{bmatrix}...
px, py = symbols('p_x, p_y') z = Matrix([[sympy.sqrt((px-x)**2 + (py-y)**2)], [sympy.atan2(py-y, px-x) - theta]]) z.jacobian(Matrix([x, y, theta]))
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Now we need to write that as a Python function. For example we might write:
from math import sqrt def H_of(x, landmark_pos): """ compute Jacobian of H matrix where h(x) computes the range and bearing to a landmark for state x """ px = landmark_pos[0] py = landmark_pos[1] hyp = (px - x[0, 0])**2 + (py - x[1, 0])**2 dist = sqrt(hyp) H = array( [[-(px - x[0...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
We also need to define a function that converts the system state into a measurement.
from math import atan2 def Hx(x, landmark_pos): """ takes a state variable and returns the measurement that would correspond to that state. """ px = landmark_pos[0] py = landmark_pos[1] dist = sqrt((px - x[0, 0])**2 + (py - x[1, 0])**2) Hx = array([[dist], [atan2(py - x[1, 0...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Design Measurement NoiseThis is quite straightforward as we need to specify measurement noise in measurement space, hence it is linear. It is reasonable to assume that the range and bearing measurement noise is independent, hence$$R=\begin{bmatrix}\sigma_{range}^2 & 0 \\ 0 & \sigma_{bearing}^2\end{bmatrix}$$ Implement...
from filterpy.kalman import ExtendedKalmanFilter as EKF from numpy import dot, array, sqrt class RobotEKF(EKF): def __init__(self, dt, wheelbase, std_vel, std_steer): EKF.__init__(self, 3, 2, 2) self.dt = dt self.wheelbase = wheelbase self.std_vel = std_vel self.std_steer = s...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
Now we have another issue to handle. The residual is notionally computed as $y = z - h(x)$ but this will not work because our measurement contains an angle in it. Suppose z has a bearing of $1^\circ$ and $h(x)$ has a bearing of $359^\circ$. Naively subtracting them would yield a bearing difference of $-358^\circ$, whic...
def residual(a, b): """ compute residual (a-b) between measurements containing [range, bearing]. Bearing is normalized to [-pi, pi)""" y = a - b y[1] = y[1] % (2 * np.pi) # force in range [0, 2 pi) if y[1] > np.pi: # move to [-pi, pi) y[1] -= 2 * np.pi return y
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
The rest of the code runs the simulation and plots the results, and shouldn't need too much comment by now. I create a variable `landmarks` that contains the coordinates of the landmarks. I update the simulated robot position 10 times a second, but run the EKF only once. This is for two reasons. First, we are not using...
from filterpy.stats import plot_covariance_ellipse from math import sqrt, tan, cos, sin, atan2 import matplotlib.pyplot as plt dt = 1.0 def z_landmark(lmark, sim_pos, std_rng, std_brg): x, y = sim_pos[0, 0], sim_pos[1, 0] d = np.sqrt((lmark[0] - x)**2 + (lmark[1] - y)**2) a = atan2(lmark[1] - y, lmark[0...
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
I have plotted the landmarks as solid squares. The path of the robot is drawn with black line. The covariance ellipses for the predict step is light gray, and the covariances of the update are shown in green. To make them visible at this scale I have set the ellipse boundary at 6$\sigma$.From this we can see that there...
landmarks = array([[5, 10], [10, 5], [15, 15], [20, 5]]) ekf = run_localization( landmarks, std_vel=0.1, std_steer=np.radians(1), std_range=0.3, std_bearing=0.1) plt.show() print('Final P:', ekf.P.diagonal())
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
The uncertainly in the estimates near the end of the track are smaller with the additional landmark. We can see the fantastic effect that multiple landmarks has on our uncertainty by only using the first two landmarks.
ekf = run_localization( landmarks[0:2], std_vel=1.e-10, std_steer=1.e-10, std_range=1.4, std_bearing=.05) print('Final P:', ekf.P.diagonal())
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
The estimate quickly diverges from the robot's path after passing the landmarks. The covariance also grows quickly. Let's see what happens with only one landmark:
ekf = run_localization( landmarks[0:1], std_vel=1.e-10, std_steer=1.e-10, std_range=1.4, std_bearing=.05) print('Final P:', ekf.P.diagonal())
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
As you probably suspected, only one landmark produces a very bad result. Conversely, a large number of landmarks allows us to make very accurate estimates.
landmarks = array([[5, 10], [10, 5], [15, 15], [20, 5], [15, 10], [10,14], [23, 14], [25, 20], [10, 20]]) ekf = run_localization( landmarks, std_vel=0.1, std_steer=np.radians(1), std_range=0.3, std_bearing=0.1, ylim=(0, 21)) print('Final P:', ekf.P.diagonal())
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
DiscussionI said that this was a 'real' problem, and in some ways it is. I've seen alternative presentations that used robot motion models that led to much easier Jacobians. On the other hand, my model of a automobile's movement is itself simplistic in several ways. First, it uses a bicycle model. A real car has two s...
import nonlinear_plots nonlinear_plots.plot_ekf_vs_mc()
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
We can see from both the graph and the print out at the bottom that the EKF has introduced quite a bit of error.In contrast, here is the performance of the UKF:
nonlinear_plots.plot_ukf_vs_mc(alpha=0.001, beta=3., kappa=1.)
_____no_output_____
CC-BY-4.0
11-Extended-Kalman-Filters.ipynb
galuardi/Kalman-and-Bayesian-Filters-in-Python
!nvidia-smi !git clone https://github.com/venkat2319/MIRnet %cd MIRNet !pip install -qq wandb from glob import glob import tensorflow as tf from mirnet.train import LowLightTrainer from mirnet.utils import init_wandb, download_dataset tf.compat.v1.logging.set_verbosity(tf.compat.v1.logging.ERROR) download_dataset('LO...
_____no_output_____
Apache-2.0
notebook/MIRNet_Low_Light_Train.ipynb
venkat2319/MIRnet
Using ObjectScript in a notebookThis notebook uses a kernel written in Python, which plugs into Jupyter to enable execution of ObjectScript inside IRIS. See `misc/kernels/objectscript/*` and `src/ObjectScript/Kernel/CodeExecutor.cls` for how this is done.Indenting each line with at least one space allows InterSystems ...
Set hello = "helloworld2" zw hello
hello="helloworld2"
MIT
src/Notebooks/ObjectScript.ipynb
gjsjohnmurray/iris-python-template