Split (1)

train · 9.31M rows

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
훈련 후 테스트셋에 대한 정확도가 90%를 조금 웃돌 정도로 많이 향상되었다.	model = get_model() callbacks = [ keras.callbacks.ModelCheckpoint("binary_2gram.keras", save_best_only=True) ] model.fit(binary_2gram_train_ds.cache(), validation_data=binary_2gram_val_ds.cache(), epochs=10, callbacks=callbacks) model = keras.mode...	Epoch 1/10 625/625 [==============================] - 12s 18ms/step - loss: 0.3857 - accuracy: 0.8347 - val_loss: 0.2791 - val_accuracy: 0.9000 Epoch 2/10 625/625 [==============================] - 4s 6ms/step - loss: 0.2592 - accuracy: 0.9082 - val_loss: 0.2947 - val_accuracy: 0.8988 Epoch 3/10 625/625 [==============...	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
방식 3: 바이그램 TF-IDF 인코딩 N-그램을 벡터화할 때 사용 빈도를 함께 저장하는 방식을 사용할 수 있다.단어의 사용 빈도가 아무래도 문장 평가에 중요한 역할을 수행할 것이기 때문이다.아래 코드에서처럼 `output_mode="count"` 옵션을 사용하면 된다.	text_vectorization = TextVectorization( ngrams=2, max_tokens=20000, output_mode="count" )	_____no_output_____	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
그런데 이렇게 하면 "the", "a", "is", "are" 등의 사용 빈도는 매우 높은 반면에"Chollet" 등의 단어는 빈도가 거의 0에 가깝게 나온다.또한 생성된 벡터의 대부분은 0으로 채워질 것이다. `max_tokens=20000`을 사용한 반면에 하나의 문장엔 많아야 몇 십개 정도의 단어만 사용되었기 때문이다. ```pythoninputs[0]: tf.Tensor([1. 1. 1. ... 0. 0. 0.], shape=(20000,), dtype=float32)``` 이 점을 고려해서 사용 빈도를 정규화한다. 평균을 원점으로 만들지는 않고 TF-IDF ...	text_vectorization = TextVectorization( ngrams=2, max_tokens=20000, output_mode="tf_idf", )	_____no_output_____	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
훈련 후 테스트셋에 대한 정확도가 다시 89% 아래로 내려간다.여기서는 별 도움이 되지 않았지만 많은 텍스트 분류 모델에서는 1% 정도의 성능 향상을 가져온다.주의사항: 아래 코드는 현재(Tensorflow 2.6과 2.7) GPU를 사용하지 않는 경우에만 작동한다. 이유는 아직 모른다([여기 참조](https://github.com/fchollet/deep-learning-with-python-notebooks/issues/190)).	text_vectorization.adapt(text_only_train_ds) tfidf_2gram_train_ds = train_ds.map(lambda x, y: (text_vectorization(x), y)) tfidf_2gram_val_ds = val_ds.map(lambda x, y: (text_vectorization(x), y)) tfidf_2gram_test_ds = test_ds.map(lambda x, y: (text_vectorization(x), y)) model = get_model() model.summary() callbacks = ...	Model: "model_2" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= input_3 (InputLayer) [(None, 20000)] 0 ...	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
부록: 문자열 벡터화 전처리를 함께 처리하는 모델 내보내기 훈련된 모델을 실전에 배치하려면 텍스트 벡터화도 모델과 함께 내보내야 한다.이를 위해 `TextVectorization` 층의 결과를 재활용만 하면 된다.	inputs = keras.Input(shape=(1,), dtype="string") # 텍스트 벡터화 추가 processed_inputs = text_vectorization(inputs) # 훈련된 모델에 적용 outputs = model(processed_inputs) # 최종 모델 inference_model = keras.Model(inputs, outputs)	_____no_output_____	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
`inference_model`은 일반 텍스트 문장을 직접 인자로 받을 수 있다.예를 들어 "That was an excellent movie, I loved it."라는 리뷰는긍정일 확률이 매우 높다고 예측된다.	import tensorflow as tf raw_text_data = tf.convert_to_tensor([ ["That was an excellent movie, I loved it."], ]) predictions = inference_model(raw_text_data) print(f"{float(predictions[0] * 100):.2f} percent positive")	92.10 percent positive	MIT	notebooks/dlp11_part01_introduction.ipynb	codingalzi/dlp
Loading neurons from s3	import numpy as np from skimage import io from pathlib import Path from brainlit.utils.session import NeuroglancerSession from brainlit.utils.Neuron_trace import NeuronTrace import napari from napari.utils import nbscreenshot %gui qt	_____no_output_____	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
Loading entire neuron from AWS `s3_trace = NeuronTrace(s3_path,seg_id,mip)` to create a NeuronTrace object with s3 file path`swc_trace = NeuronTrace(swc_path)` to create a NeuronTrace object with swc file path1. `s3_trace.get_df()` to output the s3 NeuronTrace object as a pd.DataFrame2. `swc_trace.get_df()` to output...	""" s3_path = "s3://open-neurodata/brainlit/brain1_segments" seg_id = 2 mip = 1 s3_trace = NeuronTrace(s3_path, seg_id, mip) df = s3_trace.get_df() df.head() """	Downloading: 100%\|██████████\| 1/1 [00:00<00:00, 5.13it/s] Downloading: 100%\|██████████\| 1/1 [00:00<00:00, 5.82it/s]	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
2. `swc_trace.get_df()`This function outputs the swc NeuronTrace object as a pd.DataFrame. Each row is a vertex in the swc file with the following information: `sample number``structure identifier``x coordinate``y coordinate``z coordinate``radius of dendrite``sample number of parent`The coordinates are given in spatia...	""" swc_path = str(Path().resolve().parents[2] / "data" / "data_octree" / "consensus-swcs" / '2018-08-01_G-002_consensus.swc') swc_trace = NeuronTrace(path=swc_path) df = swc_trace.get_df() df.head() """	_____no_output_____	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
3. `swc_trace.generate_df_subset(list_of_voxels)`This function creates a smaller subset of the original dataframe with coordinates in img space. Each row is a vertex in the swc file with the following information: `sample number``structure identifier``x coordinate``y coordinate``z coordinate``radius of dendrite``sampl...	"""# Choose vertices to use for the subneuron subneuron_df = df[0:3] vertex_list = subneuron_df['sample'].array # Define a neuroglancer session url = "s3://open-neurodata/brainlit/brain1" mip = 1 ngl = NeuroglancerSession(url, mip=mip) # Get vertices seg_id = 2 buffer = 10 img, bounds, vox_in_img_list = ngl.pull_ve...	Downloading: 100%\|██████████\| 1/1 [00:00<00:00, 6.08it/s] Downloading: 100%\|██████████\| 1/1 [00:00<00:00, 6.95it/s] Downloading: 100%\|██████████\| 1/1 [00:00<00:00, 5.02it/s] Downloading: 0%\| \| 0/4 [00:01<?, ?it/s] sample structure x y z r parent 0 1 0 106 106 112 1.0 ...	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
4. `swc_trace.get_df_voxel()` If we want to overlay the swc file with a corresponding image, we need to make sure that they are in the same coordinate space. Because an image in an array of voxels, it makes sense to convert the vertices from spatial units into voxel units.Given the `spacing` (spatial units/voxel) and ...	# spacing = np.array([0.29875923,0.3044159,0.98840415]) # origin = np.array([70093.276,15071.596,29306.737]) # df_voxel = swc_trace.get_df_voxel(spacing=spacing, origin=origin) # df_voxel.head()	_____no_output_____	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
5. `swc_trace.get_graph()`A neuron is a graph with no cycles (tree). While napari does not support displaying graph objects, it can display multiple paths. The DataFrame already contains all the possible edges in the neurons. Each row in the DataFrame is an edge. For example, from the above we can see that `sample 2`...	# G = swc_trace.get_graph() # print('Number of nodes:', len(G.nodes)) # print('Number of edges:', len(G.edges)) # print('\n') # print('Sample 1 coordinates (x,y,z)') # print(G.nodes[1]['x'],G.nodes[1]['y'],G.nodes[1]['z'])	Number of nodes: 1650 Number of edges: 1649 Sample 1 coordinates (x,y,z) -387 1928 -1846	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
6. `swc_trace.get_paths()` This function returns the NeuronTrace object as a list of non-overlapping paths. The union of the paths forms the graph.The algorithm works by:1. Find longest path in the graph ([networkx.algorithms.dag.dag_longest_path](https://networkx.github.io/documentation/stable/reference/algorithms/ge...	# paths = swc_trace.get_paths() # print(f"The graph was decomposed into {len(paths)} paths")	The graph was decomposed into 179 paths	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
7. `ViewerModel.add_shapes`napari displays "layers". The most common layer is the image layer. In order to display the neuron, we use `path` from the [shapes](https://napari.org/tutorials/shapes) layer	# viewer = napari.Viewer(ndisplay=3) # viewer.add_shapes(data=paths, shape_type='path', edge_color='white', name='Skeleton 2') # nbscreenshot(viewer)	_____no_output_____	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
Loading sub-neuronThe image of the entire brain has dimensions of (33792, 25600, 13312) voxels. G-002 spans a sub-image of (7386, 9932, 5383) voxels. Both are too big to load in napari and overlay the neuron.To circumvent this, we can crop out a smaller region of the neuron, load the sub-neuron, and load the correspon...	# # Create an NGL session to get the bounding box # url = "s3://open-neurodata/brainlit/brain1" # mip = 1 # ngl = NeuroglancerSession(url, mip=mip) # img, bbbox, vox = ngl.pull_chunk(2, 300, 1) # bbox = bbbox.to_list() # box = (bbox[:3], bbox[3:]) # print(box) # G_sub = s3_trace.get_sub_neuron(box) # paths_sub = s3_tr...	459	Apache-2.0	docs/notebooks/visualization/loading.ipynb	neurodata/brainl
Deep Convolutional GANsIn this notebook, you'll build a GAN using convolutional layers in the generator and discriminator. This is called a Deep Convolutional GAN, or DCGAN for short. The DCGAN architecture was first explored in 2016 and has seen impressive results in generating new images; you can read the [original ...	# import libraries import matplotlib.pyplot as plt import numpy as np import pickle as pkl %matplotlib inline	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Getting the dataHere you can download the SVHN dataset. It's a dataset built-in to the PyTorch datasets library. We can load in training data, transform it into Tensor datatypes, then create dataloaders to batch our data into a desired size.	import torch from torchvision import datasets from torchvision import transforms # Tensor transform transform = transforms.ToTensor() # SVHN training datasets svhn_train = datasets.SVHN(root='data/', split='train', download=True, transform=transform) batch_size = 128 num_workers = 0 # build DataLoaders for SVHN dat...	Using downloaded and verified file: data/train_32x32.mat	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Visualize the DataHere I'm showing a small sample of the images. Each of these is 32x32 with 3 color channels (RGB). These are the real, training images that we'll pass to the discriminator. Notice that each image has _one_ associated, numerical label.	# obtain one batch of training images dataiter = iter(train_loader) images, labels = dataiter.next() # plot the images in the batch, along with the corresponding labels fig = plt.figure(figsize=(25, 4)) plot_size=20 for idx in np.arange(plot_size): ax = fig.add_subplot(2, plot_size/2, idx+1, xticks=[], yticks=[]) ...	<ipython-input-3-a55faf2ffde6>:9: MatplotlibDeprecationWarning: Passing non-integers as three-element position specification is deprecated since 3.3 and will be removed two minor releases later. ax = fig.add_subplot(2, plot_size/2, idx+1, xticks=[], yticks=[])	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Pre-processing: scaling from -1 to 1We need to do a bit of pre-processing; we know that the output of our `tanh` activated generator will contain pixel values in a range from -1 to 1, and so, we need to rescale our training images to a range of -1 to 1. (Right now, they are in a range from 0-1.)	# current range img = images[0] print('Min: ', img.min()) print('Max: ', img.max()) # helper scale function def scale(x, feature_range=(-1, 1)): ''' Scale takes in an image x and returns that image, scaled with a feature_range of pixel values from -1 to 1. This function assumes that the input x is a...	Scaled min: tensor(-0.4196) Scaled max: tensor(0.2627)	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
--- Define the ModelA GAN is comprised of two adversarial networks, a discriminator and a generator. DiscriminatorHere you'll build the discriminator. This is a convolutional classifier like you've built before, only without any maxpooling layers. * The inputs to the discriminator are 32x32x3 tensor images* You'll wan...	import torch.nn as nn import torch.nn.functional as F # helper conv function def conv(in_channels, out_channels, kernel_size, stride=2, padding=1, batch_norm=True): """Creates a convolutional layer, with optional batch normalization. """ layers = [] conv_layer = nn.Conv2d(in_channels, out_channels, ...	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
GeneratorNext, you'll build the generator network. The input will be our noise vector `z`, as before. And, the output will be a $tanh$ output, but this time with size 32x32 which is the size of our SVHN images.What's new here is we'll use transpose convolutional layers to create our new images. * The first layer is a ...	# helper deconv function def deconv(in_channels, out_channels, kernel_size, stride=2, padding=1, batch_norm=True): """Creates a transposed-convolutional layer, with optional batch normalization. """ ## TODO: Complete this function ## create a sequence of transpose + optional batch norm layers layers...	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Build complete networkDefine your models' hyperparameters and instantiate the discriminator and generator from the classes defined above. Make sure you've passed in the correct input arguments.	# define hyperparams conv_dim = 32 z_size = 100 # define discriminator and generator D = Discriminator(conv_dim) G = Generator(z_size=z_size, conv_dim=conv_dim) print(D) print() print(G)	Discriminator( (conv1): Sequential( (0): Conv2d(3, 32, kernel_size=(4, 4), stride=(2, 2), padding=(1, 1), bias=False) ) (conv2): Sequential( (0): Conv2d(32, 64, kernel_size=(4, 4), stride=(2, 2), padding=(1, 1), bias=False) (1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats...	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Training on GPUCheck if you can train on GPU. If you can, set this as a variable and move your models to GPU. > Later, we'll also move any inputs our models and loss functions see (real_images, z, and ground truth labels) to GPU as well.	train_on_gpu = torch.cuda.is_available() if train_on_gpu: # move models to GPU G.cuda() D.cuda() print('GPU available for training. Models moved to GPU') else: print('Training on CPU.')	GPU available for training. Models moved to GPU	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
--- Discriminator and Generator LossesNow we need to calculate the losses. And this will be exactly the same as before. Discriminator Losses> * For the discriminator, the total loss is the sum of the losses for real and fake images, `d_loss = d_real_loss + d_fake_loss`. * Remember that we want the discriminator to outp...	def real_loss(D_out, smooth=False): batch_size = D_out.size(0) # label smoothing if smooth: # smooth, real labels = 0.9 labels = torch.ones(batch_size)*0.9 else: labels = torch.ones(batch_size) # real labels = 1 # move labels to GPU if available if train_on_gpu: ...	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
OptimizersNot much new here, but notice how I am using a small learning rate and custom parameters for the Adam optimizers, This is based on some research into DCGAN model convergence. HyperparametersGANs are very sensitive to hyperparameters. A lot of experimentation goes into finding the best hyperparameters such th...	import torch.optim as optim # params lr = 0.0002 beta1=0.5 beta2=0.999 # Create optimizers for the discriminator and generator d_optimizer = optim.Adam(D.parameters(), lr, [beta1, beta2]) g_optimizer = optim.Adam(G.parameters(), lr, [beta1, beta2])	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
--- TrainingTraining will involve alternating between training the discriminator and the generator. We'll use our functions `real_loss` and `fake_loss` to help us calculate the discriminator losses in all of the following cases. Discriminator training1. Compute the discriminator loss on real, training images 2. ...	import pickle as pkl # training hyperparams num_epochs = 30 # keep track of loss and generated, "fake" samples samples = [] losses = [] print_every = 300 # Get some fixed data for sampling. These are images that are held # constant throughout training, and allow us to inspect the model's performance sample_size=16 ...	Epoch [ 1/ 30] \| d_loss: 1.4085 \| g_loss: 0.9993 Epoch [ 1/ 30] \| d_loss: 0.6737 \| g_loss: 1.9478 Epoch [ 2/ 30] \| d_loss: 0.7026 \| g_loss: 2.6182 Epoch [ 2/ 30] \| d_loss: 0.4292 \| g_loss: 2.3596 Epoch [ 3/ 30] \| d_loss: 0.2889 \| g_loss: 2.7350 Epoch [ 3/ 30] \| d_loss: 0.1361 \| g_loss: 4.5...	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Training lossHere we'll plot the training losses for the generator and discriminator, recorded after each epoch.	fig, ax = plt.subplots() losses = np.array(losses) plt.plot(losses.T[0], label='Discriminator', alpha=0.5) plt.plot(losses.T[1], label='Generator', alpha=0.5) plt.title("Training Losses") plt.legend()	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
Generator samples from trainingHere we can view samples of images from the generator. We'll look at the images we saved during training.	# helper function for viewing a list of passed in sample images def view_samples(epoch, samples): fig, axes = plt.subplots(figsize=(16,4), nrows=2, ncols=8, sharey=True, sharex=True) for ax, img in zip(axes.flatten(), samples[epoch]): img = img.detach().cpu().numpy() img = np.transpose(img, (1, ...	_____no_output_____	MIT	DCGAN_Exercise.ipynb	ng572/DCGAN_SVHN
get names of each condition for later	pd.Categorical(luminescence_raw_df.condition) names = luminescence_raw_df.condition.unique() for name in names: print(name) #get list of promoters pd.Categorical(luminescence_raw_df.Promoter) prom_names = luminescence_raw_df.Promoter.unique() for name in prom_names: print(name)	UBQ10 NIR1 NOS STAP4 NRP	MIT	src/plotting/luminescence/24.11.19/luminescence_plots.ipynb	Switham1/PromoterArchitecture
test normality	#returns test statistic, p-value for name1 in prom_names: for name in names: print('{}: {}'.format(name, stats.shapiro(luminescence_raw_df['nluc/fluc'][luminescence_raw_df.condition == name])))	nitrate_free: (0.7033216953277588, 0.0002697518502827734) 100mM nitrate_2hrs_morning: (0.7973607182502747, 0.00463036959990859) 100mM nitrate_overnight: (0.8101227879524231, 0.004972793627530336) nitrate_free: (0.7033216953277588, 0.0002697518502827734) 100mM nitrate_2hrs_morning: (0.7973607182502747, 0.004630369599908...	MIT	src/plotting/luminescence/24.11.19/luminescence_plots.ipynb	Switham1/PromoterArchitecture
not normal	#test variance stats.levene(luminescence_raw_df['nluc/fluc'][luminescence_raw_df.condition == names[0]], luminescence_raw_df['nluc/fluc'][luminescence_raw_df.condition == names[1]], luminescence_raw_df['nluc/fluc'][luminescence_raw_df.condition == names[2]]) test = luminescence_raw_df.gr...	_____no_output_____	MIT	src/plotting/luminescence/24.11.19/luminescence_plots.ipynb	Switham1/PromoterArchitecture
Загрузка библиотек и данных	!pip install simpletransformers==0.61.13 !pip uninstall transformers !pip install transformers==4.10.0 !git clone https://github.com/GoldenRMT/WikiSearch.git !pip install googledrivedownloader import nltk nltk.download('stopwords') nltk.download('punkt') from nltk.corpus import stopwords nltk.download('wordnet') stopwo...	Downloading 13Nuwm7BV-4RXI9JqjPTDE9rcdupkKqlF into /Data/AIIJC/aiijc_1578_goodFromTrain_pretrained.model... Done.	MIT	solution_AIIJC(NLP)/Notebooks/singleAnswering_Aiijc.ipynb	Makual/AIIJC_NLP
Функции для предобработки текста	def normal_form(word): #Получение нормальной формы слова word = word.lower() return word def clean_html(html): #Очистка html soup = BeautifulSoup(BeautifulSoup(html, "lxml").text) return str(soup.body) def get_good_tokens(text): #Выделение ключевых слов good_tokens = [] for tokens in tokenizer(text)[1]:...	_____no_output_____	MIT	solution_AIIJC(NLP)/Notebooks/singleAnswering_Aiijc.ipynb	Makual/AIIJC_NLP
Загрузка модели и функция для ответов на вопрос	model = joblib.load('/Data/AIIJC/aiijc_1578_goodFromTrain_pretrained.model') model.args.max_seq_length = 512 model.args.silent = True def answering(question): text = question good_tokens = get_good_tokens(text) try: urls = wikipedia.search(text,results=2) except: link_1 = '-' link_2 = '-' tr...	_____no_output_____	MIT	solution_AIIJC(NLP)/Notebooks/singleAnswering_Aiijc.ipynb	Makual/AIIJC_NLP
Проверка работоспособности и времени работы функции	import time time_1 = time.time() print(answering("What is the name of Trump first daughter?")) print('Время обработки запроса: ' + str(time.time()-time_1))	Ivana Marie "Ivanka" Trump Время обработки запроса: 1.6687853336334229	MIT	solution_AIIJC(NLP)/Notebooks/singleAnswering_Aiijc.ipynb	Makual/AIIJC_NLP
CNTK 101: Logistic Regression and ML PrimerThis tutorial is targeted to individuals who are new to CNTK and to machine learning. In this tutorial, you will train a simple yet powerful machine learning model that is widely used in industry for a variety of applications. The model trained below scales to massive data se...	# Figure 1 Image(url="https://www.cntk.ai/jup/cancer_data_plot.jpg", width=400, height=400)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Goal:Our goal is to learn a classifier that can automatically label any patient into either the benign or malignant categories given two features (age and tumor size). In this tutorial, we will create a linear classifier, a fundamental building-block in deep networks.	# Figure 2 Image(url= "https://www.cntk.ai/jup/cancer_classify_plot.jpg", width=400, height=400)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
In the figure above, the green line represents the model learned from the data and separates the blue dots from the red dots. In this tutorial, we will walk you through the steps to learn the green line. Note: this classifier does make mistakes, where a couple of blue dots are on the wrong side of the green line. Howev...	# Figure 3 Image(url= "https://www.cntk.ai/jup/logistic_neuron.jpg", width=300, height=200)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
In the above figure, contributions from different input features are linearly weighted and aggregated. The resulting sum is mapped to a (0, 1) range via a [sigmoid]( https://en.wikipedia.org/wiki/Sigmoid_function) function. For classifiers with more than two output labels, one can use a [softmax](https://en.wikipedia.o...	# Import the relevant components from __future__ import print_function import numpy as np import sys import os import cntk as C import cntk.tests.test_utils cntk.tests.test_utils.set_device_from_pytest_env() # (only needed for our build system) C.cntk_py.set_fixed_random_seed(1) # fix the random seed so that LR exampl...	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Data GenerationLet us generate some synthetic data emulating the cancer example using the `numpy` library. We have two input features (represented in two-dimensions) and two output classes (benign/blue or malignant/red). In our example, each observation (a single 2-tuple of features - age and size) in the training dat...	# Define the network input_dim = 2 num_output_classes = 2	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Input and LabelsIn this tutorial we are generating synthetic data using the `numpy` library. In real-world problems, one would use a [reader](https://docs.microsoft.com/en-us/cognitive-toolkit/brainscript-and-python---understanding-and-extending-readers), that would read feature values (`features`: age and *tumor si...	# Ensure that we always get the same results np.random.seed(0) # Helper function to generate a random data sample def generate_random_data_sample(sample_size, feature_dim, num_classes): # Create synthetic data using NumPy. Y = np.random.randint(size=(sample_size, 1), low=0, high=num_classes) # Make sure ...	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Let us visualize the input data.Note: If the import of `matplotlib.pyplot` fails, please run `conda install matplotlib`, which will fix the `pyplot` version dependencies. If you are on a python environment different from Anaconda, then use `pip install matplotlib`.	# Plot the data import matplotlib.pyplot as plt %matplotlib inline # let 0 represent malignant/red and 1 represent benign/blue colors = ['r' if label == 0 else 'b' for label in labels[:,0]] plt.scatter(features[:,0], features[:,1], c=colors) plt.xlabel("Age (scaled)") plt.ylabel("Tumor size (in cm)") plt.show()	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Model CreationA logistic regression (a.k.a. LR) network is a simple building block, but has powered many ML applications in the past decade. LR is a simple linear model that takes as input a vector of numbers describing the properties of what we are classifying (also known as a feature vector, $\bf{x}$, the blue nodes...	# Figure 4 Image(url= "https://www.cntk.ai/jup/logistic_neuron2.jpg", width=300, height=200)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
The first step is to compute the evidence for an observation. $$z = \sum_{i=1}^n w_i \times x_i + b = \textbf{w} \cdot \textbf{x} + b$$ where $\bf{w}$ is the weight vector of length $n$ and $b$ is known as the [bias](https://www.quora.com/What-does-the-bias-term-represent-in-logistic-regression) term. Note: we use **bo...	feature = C.input_variable(input_dim, np.float32)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Network setupThe `linear_layer` function is a straightforward implementation of the equation above. We perform two operations:0. multiply the weights ($\bf{w}$) with the features ($\bf{x}$) using the CNTK `times` operator,1. add the bias term ($b$).These CNTK operations are optimized for execution on the available ha...	# Define a dictionary to store the model parameters mydict = {} def linear_layer(input_var, output_dim): input_dim = input_var.shape[0] weight_param = C.parameter(shape=(input_dim, output_dim)) bias_param = C.parameter(shape=(output_dim)) mydict['w'], mydict['b'] = weight_param, bias_param ...	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
`z` will be used to represent the output of the network.	output_dim = num_output_classes z = linear_layer(feature, output_dim)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Learning model parametersNow that the network is set up, we would like to learn the parameters $\bf w$ and $b$ for our simple linear layer. To do so we convert, the computed evidence ($z$) into a set of predicted probabilities ($\textbf p$) using a `softmax` function.$$ \textbf{p} = \mathrm{softmax}(z)$$ The `softmax`...	label = C.input_variable(num_output_classes, np.float32) loss = C.cross_entropy_with_softmax(z, label)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
EvaluationIn order to evaluate the classification, we can compute the [classification_error](https://www.cntk.ai/pythondocs/cntk.metrics.htmlcntk.metrics.classification_error), which is 0 if our model was correct (it assigned the true label the most probability), otherwise 1.	eval_error = C.classification_error(z, label)	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Configure trainingThe trainer strives to minimize the `loss` function using an optimization technique. In this tutorial, we will use [Stochastic Gradient Descent](https://en.wikipedia.org/wiki/Stochastic_gradient_descent) (`sgd`), one of the most popular techniques. Typically, one starts with random initialization of ...	# Instantiate the trainer object to drive the model training learning_rate = 0.5 lr_schedule = C.learning_rate_schedule(learning_rate, C.UnitType.minibatch) learner = C.sgd(z.parameters, lr_schedule) trainer = C.Trainer(z, (loss, eval_error), [learner])	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
First, let us create some helper functions that will be needed to visualize different functions associated with training. Note: these convenience functions are for understanding what goes on under the hood.	# Define a utility function to compute the moving average. # A more efficient implementation is possible with np.cumsum() function def moving_average(a, w=10): if len(a) < w: return a[:] return [val if idx < w else sum(a[(idx-w):idx])/w for idx, val in enumerate(a)] # Define a utility that prints...	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Run the trainerWe are now ready to train our Logistic Regression model. We want to decide what data we need to feed into the training engine.In this example, each iteration of the optimizer will work on 25 samples (25 dots w.r.t. the plot above) a.k.a. `minibatch_size`. We would like to train on 20000 observations. If...	# Initialize the parameters for the trainer minibatch_size = 25 num_samples_to_train = 20000 num_minibatches_to_train = int(num_samples_to_train / minibatch_size) from collections import defaultdict # Run the trainer and perform model training training_progress_output_freq = 50 plotdata = defaultdict(list) for i in ...	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Run evaluation / Testing Now that we have trained the network, let us evaluate the trained network on data that hasn't been used for training. This is called testing. Let us create some new data and evaluate the average error and loss on this set. This is done using `trainer.test_minibatch`. Note the error on this...	# Run the trained model on a newly generated dataset test_minibatch_size = 25 features, labels = generate_random_data_sample(test_minibatch_size, input_dim, num_output_classes) trainer.test_minibatch({feature : features, label : labels})	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Checking prediction / evaluation For evaluation, we softmax the output of the network into a probability distribution over the two classes, the probability of each observation being malignant or benign.	out = C.softmax(z) result = out.eval({feature : features})	_____no_output_____	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
Let us compare the ground-truth label with the predictions. They should be in agreement.Question: - How many predictions were mislabeled? Can you change the code below to identify which observations were misclassified?	print("Label :", [np.argmax(label) for label in labels]) print("Predicted:", [np.argmax(x) for x in result])	Label : [1, 0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1] Predicted: [1, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 1]	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
VisualizationIt is desirable to visualize the results. In this example, the data can be conveniently plotted using two spatial dimensions for the input (patient age on the x-axis and tumor size on the y-axis), and a color dimension for the output (red for malignant and blue for benign). For data with higher dimensions...	# Model parameters print(mydict['b'].value) bias_vector = mydict['b'].value weight_matrix = mydict['w'].value # Plot the data import matplotlib.pyplot as plt # let 0 represent malignant/red, and 1 represent benign/blue colors = ['r' if label == 0 else 'b' for label in labels[:,0]] plt.scatter(features[:,0], featu...	[ 8.00007153 -8.00006485]	MIT	Tutorials/CNTK_101_LogisticRegression.ipynb	shyamalschandra/CNTK
3장 처음 시작하는 머신러닝	# 必要ライブラリの導入 !pip install japanize_matplotlib \| tail -n 1 !pip install torchviz \| tail -n 1 # 必要ライブラリのインポート %matplotlib inline import numpy as np import matplotlib.pyplot as plt #import japanize_matplotlib from IPython.display import display # PyTorch関連ライブラリ import torch from torchviz import make_dot # デフォルトフォントサイズ変更...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.4 경사 하강법의 구현 방법	def L(u, v): return 3 * u*2 + 3 v*2 - uv + 7u - 7v + 10 def Lu(u, v): return 6* u - v + 7 def Lv(u, v): return 6* v - u - 7 u = np.linspace(-5, 5, 501) v = np.linspace(-5, 5, 501) U, V = np.meshgrid(u, v) Z = L(U, V) # 勾配降下法のシミュレーション W = np.array([4.0, 4.0]) W1 = [W[0]] W2 = [W[1]] N = 21 alpha = 0....	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.5 データ前処理5人の人の身長と体重のデータを使う。 1次関数で身長から体重を予測する場合、最適な直線を求めることが目的。	# サンプルデータの宣言 sampleData1 = np.array([ [166, 58.7], [176.0, 75.7], [171.0, 62.1], [173.0, 70.4], [169.0,60.1] ]) print(sampleData1) # 機械学習モデルで扱うため、身長だけを抜き出した変数xと # 体重だけを抜き出した変数yをセットする x = sampleData1[:,0] y = sampleData1[:,1] import matplotlib # '맑은 고딕'으로 폰트 설정 matplotlib.rcParams['font.family'] = '...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
座標系の変換機械学習モデルでは、データは0に近い値を持つことが望ましい。そこで、x, y ともに平均値が0になるように平行移動し、新しい座標系をX, Yとする。	X = x - x.mean() Y = y - y.mean() # 散布図表示で結果の確認 fig1 = plt.gcf() plt.scatter(X, Y, c='k', s=50) plt.xlabel('$X$') plt.ylabel('$Y$') plt.title('데이터 가공 후 신장과 체중의 관계') plt.show() plt.draw() fig1.savefig('ex03-04.tif', format='tif', dpi=300)	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.6 予測計算	# XとYをテンソル変数化する X = torch.tensor(X).float() Y = torch.tensor(Y).float() # 結果確認 print(X) print(Y) # 重み変数の定義 # WとBは勾配計算をするので、requires_grad=Trueとする W = torch.tensor(1.0, requires_grad=True).float() B = torch.tensor(1.0, requires_grad=True).float() # 予測関数は一次関数 def pred(X): return W * X + B # 予測値の計算 Yp = pred(X) #...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.7 損失計算	# 損失関数は誤差二乗平均 def mse(Yp, Y): loss = ((Yp - Y) ** 2).mean() return loss # 損失計算 loss = mse(Yp, Y) # 結果標示 print(loss) # 損失の計算グラフ可視化 params = {'W': W, 'B': B} g = make_dot(loss, params=params) display(g) g.render('ex03-11', format='tif') !dot -Ttif -Gdpi=300 ex03-11 -o ex03-11_large.tif	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.8 勾配計算	# 勾配計算 loss.backward() # 勾配値確認 print(W.grad) print(B.grad)	tensor(-19.0400) tensor(2.0000)	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.9 パラメータ修正	# 学習率の定義 lr = 0.001 # 경사를 기반으로 파라미터 수정 W -= lr * W.grad B -= lr * B.grad	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
WとBは一度計算済みなので、この状態で値の更新ができない次の書き方にする必要がある	# 勾配を元にパラメータ修正 # with torch.no_grad() を付ける必要がある with torch.no_grad(): W -= lr * W.grad B -= lr * B.grad # 計算済みの勾配値をリセットする W.grad.zero_() B.grad.zero_() # パラメータと勾配値の確認 print(W) print(B) print(W.grad) print(B.grad)	tensor(1.0190, requires_grad=True) tensor(0.9980, requires_grad=True) tensor(0.) tensor(0.)	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
元の値はどちらも1.0だったので、Wは微少量増加、Bは微少量減少したことがわかる。この計算を繰り返すことで、最適なWとBを求めるのが勾配降下法となる。 3.10 繰り返し計算	# 初期化 # WとBを変数として扱う W = torch.tensor(1.0, requires_grad=True).float() B = torch.tensor(1.0, requires_grad=True).float() # 繰り返し回数 num_epochs = 500 # 学習率 lr = 0.001 # 記録用配列初期化 history = np.zeros((0, 2)) # ループ処理 for epoch in range(num_epochs): # 予測計算 Yp = pred(X) # 損失計算 loss = mse(Yp, Y) ...	epoch = 0 loss = 13.3520 epoch = 10 loss = 10.3855 epoch = 20 loss = 8.5173 epoch = 30 loss = 7.3364 epoch = 40 loss = 6.5858 epoch = 50 loss = 6.1047 epoch = 60 loss = 5.7927 epoch = 70 loss = 5.5868 epoch = 80 loss = 5.4476 epoch = 90 loss = 5.3507 epoch = 100 loss = 5.2805 epoch = 110 loss = 5.2275 epoch...	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.11 結果確認	# パラメータの最終値 print('W = ', W.data.numpy()) print('B = ', B.data.numpy()) #損失の確認 print(f'초기상태: 손실:{history[0,1]:.4f}') print(f'최종상태: 손실:{history[-1,1]:.4f}') # 学習曲線の表示 (損失) fig1 = plt.gcf() plt.plot(history[:,0], history[:,1], 'b') plt.xlabel('반복 횟수') plt.ylabel('손실') plt.title('학습 곡선(손실)') plt.show() plt.draw() fig1...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
散布図に回帰直線を重ね書きする	# xの範囲を求める(Xrange) X_max = X.max() X_min = X.min() X_range = np.array((X_min, X_max)) X_range = torch.from_numpy(X_range).float() print(X_range) # 対応するyの予測値を求める Y_range = pred(X_range) print(Y_range.data) # グラフ描画 fig1 = plt.gcf() plt.scatter(X, Y, c='k', s=50) plt.xlabel('$X$') plt.ylabel('$Y$') plt.plot(X_range.d...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
加工前データへの回帰直線描画	# y座標値とx座標値の計算 x_range = X_range + x.mean() yp_range = Y_range + y.mean() # グラフ描画 fig1 = plt.gcf() plt.scatter(x, y, c='k', s=50) plt.xlabel('$x$') plt.ylabel('$y$') plt.plot(x_range, yp_range.data, lw=2, c='b') plt.title('신장과 체중의 상관 직선(가공 전)') plt.show() plt.draw() fig1.savefig('ex03-21.tif', format='tif', dpi=30...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.12 最適化関数とstep関数の利用	# 初期化 # WとBを変数として扱う W = torch.tensor(1.0, requires_grad=True).float() B = torch.tensor(1.0, requires_grad=True).float() # 繰り返し回数 num_epochs = 500 # 学習率 lr = 0.001 # optimizerとしてSGD(確率的勾配降下法)を指定する import torch.optim as optim optimizer = optim.SGD([W, B], lr=lr) # 記録用配列初期化 history = np.zeros((0, 2)) # ループ処理 for epo...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
3.7の結果と見比べるとまったく同じであることがわかる。つまり、step関数でやっていることは、次のコードと同じ。```py3 with torch.no_grad(): パラメータ修正 (フレームワークを使う場合はstep関数) W -= lr * W.grad B -= lr * B.grad``` 最適化関数のチューニング	# 初期化 # WとBを変数として扱う W = torch.tensor(1.0, requires_grad=True).float() B = torch.tensor(1.0, requires_grad=True).float() # 繰り返し回数 num_epochs = 500 # 学習率 lr = 0.001 # optimizerとしてSGD(確率的勾配降下法)を指定する import torch.optim as optim optimizer = optim.SGD([W, B], lr=lr, momentum=0.9) # 記録用配列初期化 history2 = np.zeros((0, 2)) #...	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
コラム　局所最適解	def f(x): return x * (x+1) * (x+2) * (x-2) x = np.arange(-3, 2.7, 0.05) y = f(x) plt.plot(x, y) plt.axis('off') plt.show()	_____no_output_____	Apache-2.0	notebooks/ch03_first_ml.ipynb	ychoi-kr/pytorch_book_info
Assignment 2: Machine learning with tree based models In this assignment, you will work on the Titanic dataset and use machine learning to create a model that predicts which passengers survived the Titanic shipwreck. --- About the dataset:---* The column named `Survived` is the label and the remaining c...	import pandas as pd import numpy as np import matplotlib.pyplot as plt from sklearn.impute import SimpleImputer import seaborn as sns from sklearn.tree import DecisionTreeClassifier from sklearn.model_selection import train_test_split,cross_val_score,GridSearchCV from sklearn.linear_model import LinearRegression,Logis...	_____no_output_____	MIT	Assignment_06/Assignment_ML_L2_Sankalp_Jain_ipynb_txt.ipynb	Sankalp679/SHALA
Read the CSV and Perform Basic Data Cleaning	df = pd.read_csv("exoplanet_data.csv") # Drop the null columns where all values are null df = df.dropna(axis='columns', how='all') # Drop the null rows df = df.dropna() df.head() df.describe()	_____no_output_____	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Select your features (columns)	# Set features. This will also be used as your x values. target = df["koi_disposition"] data = df.drop("koi_disposition", axis=1) feature_names = data.columns data.head()	_____no_output_____	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Create a Train Test SplitUse `koi_disposition` for the y values	from sklearn.model_selection import train_test_split X_train, X_test, y_train, y_test = train_test_split(data, target, random_state=42) X_train.head()	_____no_output_____	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Pre-processingScale the data using the MinMaxScaler and perform some feature selection	# Scale your data from sklearn.preprocessing import MinMaxScaler X_minmax = MinMaxScaler().fit(X_train) X_train_minmax = X_minmax.transform(X_train) X_test_minmax = X_minmax.transform(X_test) from sklearn.svm import SVC model = SVC(kernel='linear') model.fit(X_train_minmax, y_train)	_____no_output_____	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Train the Model	print(f"Training Data Score: {model.score(X_train_minmax, y_train)}") print(f"Testing Data Score: {model.score(X_test_minmax, y_test)}")	Training Data Score: 0.8455082967766546 Testing Data Score: 0.8415331807780321	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Hyperparameter TuningUse `GridSearchCV` to tune the model's parameters	# Create the GridSearchCV model from sklearn.model_selection import GridSearchCV param_grid = {'C': [1, 5, 10, 50], 'gamma': [0.0001, 0.0005, 0.001, 0.005]} grid = GridSearchCV(model, param_grid, verbose=3) # Train the model with GridSearch grid.fit(X_train_minmax, y_train) print(grid.best_params_) print(...	precision recall f1-score support CANDIDATE 0.81 0.67 0.73 411 CONFIRMED 0.76 0.85 0.80 484 FALSE POSITIVE 0.98 1.00 0.99 853 accuracy 0.88 1748 macro avg 0.85 0.84 ...	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Save the Model	# save your model by updating "your_name" with your name # and "your_model" with your model variable # be sure to turn this in to BCS # if joblib fails to import, try running the command to install in terminal/git-bash import joblib filename = 'models/bridgette_svm.sav' joblib.dump(model, filename)	_____no_output_____	MIT	exoplanet1.ipynb	bshub6/machine-learning-challenge
Watershed Distance Transform for 3D Data---Implementation of papers:[Deep Watershed Transform for Instance Segmentation](http://openaccess.thecvf.com/content_cvpr_2017/papers/Bai_Deep_Watershed_Transform_CVPR_2017_paper.pdf)[Learn to segment single cells with deep distance estimator and deep cell detector](https://arx...	import os import errno import datetime import numpy as np import deepcell	Using TensorFlow backend.	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Load the Training Data	# Download the data (saves to ~/.keras/datasets) filename = 'mousebrain.npz' test_size = 0.1 # % of data saved as test seed = 0 # seed for random train-test split (X_train, y_train), (X_test, y_test) = deepcell.datasets.mousebrain.load_data(filename, test_size=test_size, seed=seed) print('X.shape: {}\ny.shape: {}'.fo...	Downloading data from https://deepcell-data.s3.amazonaws.com/nuclei/mousebrain.npz 1730158592/1730150850 [==============================] - 106s 0us/step X.shape: (176, 15, 256, 256, 1) y.shape: (176, 15, 256, 256, 1)	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Set up filepath constants	# the path to the data file is currently required for `train_model_()` functions # change DATA_DIR if you are not using `deepcell.datasets` DATA_DIR = os.path.expanduser(os.path.join('~', '.keras', 'datasets')) # DATA_FILE should be a npz file, preferably from `make_training_data` DATA_FILE = os.path.join(DATA_DIR, f...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Set up training parameters	from tensorflow.keras.optimizers import SGD from deepcell.utils.train_utils import rate_scheduler fgbg_model_name = 'conv_fgbg_3d_model' conv_model_name = 'conv_watershed_3d_model' n_epoch = 10 # Number of training epochs norm_method = 'whole_image' # data normalization - `whole_image` for 3d conv receptive_field =...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
First, create a foreground/background separation model Instantiate the fgbg model	from deepcell import model_zoo fgbg_model = model_zoo.bn_feature_net_skip_3D( receptive_field=receptive_field, n_features=2, # segmentation mask (is_cell, is_not_cell) n_frames=frames_per_batch, n_skips=n_skips, n_conv_filters=32, n_dense_filters=128, input_shape=tuple([frames_per_batch] +...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Train the fgbg model	from deepcell.training import train_model_conv fgbg_model = train_model_conv( model=fgbg_model, dataset=DATA_FILE, # full path to npz file model_name=fgbg_model_name, test_size=test_size, seed=seed, transform='fgbg', optimizer=optimizer, batch_size=batch_size, frames_per_batch=fram...	X_train shape: (198, 15, 256, 256, 1) y_train shape: (198, 15, 256, 256, 1) X_test shape: (22, 15, 256, 256, 1) y_test shape: (22, 15, 256, 256, 1) Output Shape: (None, 3, 256, 256, 2) Number of Classes: 2 Training on 1 GPUs Epoch 1/10 197/198 [============================>.] - ETA: 0s - loss: 0.8965 - model_loss: 0.21...	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Next, Create a model for the watershed energy transform Instantiate the distance transform model	from deepcell import model_zoo watershed_model = model_zoo.bn_feature_net_skip_3D( fgbg_model=fgbg_model, receptive_field=receptive_field, n_skips=n_skips, n_features=distance_bins, n_frames=frames_per_batch, n_conv_filters=32, n_dense_filters=128, multires=False, last_only=False, ...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Train the model	from deepcell.training import train_model_conv watershed_model = train_model_conv( model=watershed_model, dataset=DATA_FILE, # full path to npz file model_name=conv_model_name, test_size=test_size, seed=seed, transform=transform, distance_bins=distance_bins, erosion_width=erosion_width...	X_train shape: (198, 15, 256, 256, 1) y_train shape: (198, 15, 256, 256, 1) X_test shape: (22, 15, 256, 256, 1) y_test shape: (22, 15, 256, 256, 1) Output Shape: (None, 3, 256, 256, 4) Number of Classes: 4 Training on 1 GPUs Epoch 1/10 197/198 [============================>.] - ETA: 0s - loss: 3.8927 - model_5_loss: 0....	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Run the modelThe model was trained on only a `frames_per_batch` frames at a time. In order to run this data on a full set of frames, a new model must be instantiated, which will load the trained weights. Save weights of trained models	fgbg_weights_file = os.path.join(MODEL_DIR, '{}.h5'.format(fgbg_model_name)) fgbg_model.save_weights(fgbg_weights_file) watershed_weights_file = os.path.join(MODEL_DIR, '{}.h5'.format(conv_model_name)) watershed_model.save_weights(watershed_weights_file)	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Initialize the new models	from deepcell import model_zoo # All training parameters should match except for the `input_shape` run_fgbg_model = model_zoo.bn_feature_net_skip_3D( receptive_field=receptive_field, n_features=2, n_frames=frames_per_batch, n_skips=n_skips, n_conv_filters=32, n_dense_filters=128, input_sha...	(4, 15, 256, 256, 1)	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Make predictions on test data	test_images = run_watershed_model.predict(X_test)[-1] test_images_fgbg = run_fgbg_model.predict(X_test)[-1] print('watershed transform shape:', test_images.shape) print('segmentation mask shape:', test_images_fgbg.shape)	watershed transform shape: (4, 15, 256, 256, 4) segmentation mask shape: (4, 15, 256, 256, 2)	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Watershed post-processing	argmax_images = [] for i in range(test_images.shape[0]): max_image = np.argmax(test_images[i], axis=-1) argmax_images.append(max_image) argmax_images = np.array(argmax_images) argmax_images = np.expand_dims(argmax_images, axis=-1) print('watershed argmax shape:', argmax_images.shape) # threshold the foreground...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Plot the results	import matplotlib.pyplot as plt import matplotlib.animation as animation index = np.random.randint(low=0, high=watershed_images.shape[0]) frame = np.random.randint(low=0, high=watershed_images.shape[1]) print('Image:', index) print('Frame:', frame) fig, axes = plt.subplots(ncols=3, nrows=2, figsize=(15, 15), sharex=...	_____no_output_____	Apache-2.0	scripts/watershed/Watershed Transform 3D Fully Convolutional.ipynb	esgomezm/deepcell-tf
Tutorial 6 - Handle Missing Data replace function	import pandas as pd import numpy as np df = pd.read_csv('sample_data_tutorial_06.csv') df newdf = df.replace(-99999,np.NaN) newdf newdf = df.replace([-99999, -88888],np.NaN) newdf newdf = df.replace({ 'temperature': -99999, 'windspeed': [-99999, -88888], 'event': 'No event' }, np.NaN) newdf ...	_____no_output_____	MIT	Python Pandas Tutorials 06.ipynb	HenriqueArgentieri/Tutoriais
PARSE SINGLE ABSTRACT WITH NON-INDEXED AUTHORS LIST	abstract = text_dict['P123'] #abstract abstract_info = re.findall(r"\w+[A-Z\w+]\w+.(?=TNF\stherapy.)", abstract) abstract_head = str(abstract_info[0]) abstract_head authors_info = re.findall(r"\w+[^A-Z\d)\W]\s\w.(?=TNF\stherapy)", abstract) authors = str(authors_info[0]) authors author_name = re.findall(r"\w.+(?=Sp...	_____no_output_____	MIT	file_parse.ipynb	ivanlohvyn/beetroot_parse_pdf
PARSE SINGLE ABSTRACT WITH INDEXED AUTHORS LIST	abstract = text_dict['P120'] #abstract abstract_info = re.findall(r"\w+[A-Z\w+]\w+.(?=Introduction.)", abstract) abstract_head = str(abstract_info[0]) abstract_head authors_info = re.findall(r"\w+[^A-Z\d)\W]\s\w.(?=Introduction.)", abstract) authors = str(authors_info[0]) authors author_name = re.findall(r"(\w+.\s[...	_____no_output_____	MIT	file_parse.ipynb	ivanlohvyn/beetroot_parse_pdf
![Self Check Exercises check mark image](files/art/check.png) 6.3.2 Self Check 2. _(IPython Session)_ Given the sets `{10, 20, 30}` and `{5, 10, 15, 20}` use the mathematical set operators to produce the following results:a. `{30}` b. `{5, 15, 30}` c. `{5, 10, 15, 20, 30}` d. `{10, 20}`*Answer:...	{10, 20, 30} - {5, 10, 15, 20} {10, 20, 30} ^ {5, 10, 15, 20} {10, 20, 30} \| {5, 10, 15, 20} {10, 20, 30} & {5, 10, 15, 20} ########################################################################## # (C) Copyright 2019 by Deitel & Associates, Inc. and # # Pearson Education, Inc. All Rights Reserved....	_____no_output_____	Apache-2.0	examples/ch06/snippets_ipynb/06.03.02selfcheck.ipynb	germanngc/PythonFundamentals
All the IPython Notebooks in this lecture series by Dr. Milan Parmar are available @ [GitHub](https://github.com/milaan9/03_Python_Flow_Control) Python Nested `if` statementWe can have a nested-[if-else](https://github.com/milaan9/03_Python_Flow_Control/blob/main/002_Python_if_else_statement.ipynb) or nested-*...	# Example 1: a=10 if a>=20: # Condition FALSE print ("Condition is True") else: # Code will go to ELSE body if a>=15: # Condition FALSE print ("Checking second value") else: # Code will go to ELSE body print ("All Conditions are false") # Example 2: x = 10 y = 12 if x > y: print( "...	96 is greater than 66 96 and 96 are equal	MIT	004_Python_Nested_if_statement.ipynb	chen181016/03_Python_Flow_Control
	import torch import torch.nn as nn import torchvision.transforms.functional as TF	_____no_output_____	MIT	notebooks/Original_U-Net_PyTorch.ipynb	jimmiemunyi/fastai-experiments
The Original U-Net Architecture ![image.png](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAA5wAAAJmCAIAAAC/rhhNAAAgAElEQVR4Aey9Xask2Zmol//A5GDw7bB/gnb7zjfDSRn8xagv9oznMDAqms7hCKMpyoJE6GBCwxwUgmKodJ2yiGkOZYX7widGJfko58wUHepCqk73xxDqciPnxqAOpDZS9LGEwiNEO1DfLNP7zVpaHV8ZEbl27hWZT5J0R0ZGvGvF867IeGrtFSsmihcEIAABCEAAAhCAAARG...	def conv_block(ni, nf): return nn.Sequential( nn.Conv2d(ni, nf, kernel_size=3, stride=1), nn.ReLU(inplace=True), nn.Conv2d(nf, nf, kernel_size=3, stride=1), nn.ReLU(inplace=True) )	_____no_output_____	MIT	notebooks/Original_U-Net_PyTorch.ipynb	jimmiemunyi/fastai-experiments
Implementing the origal architecture:	class UNET(nn.Module): def __init__(self, in_channels=1, out_channels=1, features = [64, 128, 256, 512]): super(UNET, self).__init__() self.encoder = nn.ModuleList() self.pool = nn.MaxPool2d(kernel_size=2, stride=2) # create the contracting path (encoder + bottleneck) for feature...	_____no_output_____	MIT	notebooks/Original_U-Net_PyTorch.ipynb	jimmiemunyi/fastai-experiments

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.