深部内幕神经网络操场的数据生成
playground-data的Python项目详细描述
游乐场数据
深内幕神经网络操场的数据生成。
该项目/包作为辅助从{{a2}分叉的Nerural Network Playground - Deep Insider而存在。
官方网页
- The python package "playground-data" on PyPI for this project is available here
- The source for this package is available here
要求
- Python2:2.7+Python3:3.4、3.5、3.6+
- 努比
- matplotlib
使用pip
安装此软件包
pip install playground-data
用法
from__future__importprint_functionprint('Import plygdata package as pg')importplygdataaspg# Or, you can 'import' classes and functions directly like this:# from plygdata.datahelper import DatasetType# from plygdata.dataset import generate
print('Imported "playground-data" package version is ...')print(pg.__version__)
print('Code for plotting sample graph')importpprintpprint.pprint(dir(pg))# How to find class memberspprint.pprint(dir(pg.DatasetType))#['ClassifyCircleData',# 'ClassifySpiralData',# 'ClassifyTwoGaussData',# 'ClassifyXORData',# 'RegressGaussian',# 'RegressPlane',# ...]fig,ax=pg.plot_sample(pg.DatasetType.ClassifyCircleData)# # uncomment if a graph is not shown# import matplotlib.pyplot as plt# plt.show()
print('Basic code for generating and graphing data')data_noise=0.0validation_data_ratio=0.5# Generate datadata_array=pg.generate_data(pg.DatasetType.ClassifyCircleData,data_noise)#data_array = pg.generate_data(pg.DatasetType.ClassifyXORData, data_noise)#data_array = pg.generate_data(pg.DatasetType.ClassifyTwoGaussData, data_noise)#data_array = pg.generate_data(pg.DatasetType.ClassifySpiralData, data_noise)#data_array = pg.generate_data(pg.DatasetType.RegressPlane, data_noise)#data_array = pg.generate_data(pg.DatasetType.RegressGaussian, data_noise)# Divide the data for training and validating at a specified ratio (further, separate each data into Coordinate point data part and teacher label part)X_train,y_train,X_valid,y_valid=pg.split_data(data_array,validation_size=validation_data_ratio)# Plot the data on the standard graph for Playgroundfig,ax=pg.plot_points_with_playground_style(X_train,y_train,X_valid,y_valid,figsize=(6,6),dpi=100)# # get figure + axes of matplotlib graph and plot the data points# fig = pg.get_playground_figure(enable_colorbar=True)# ax = pg.get_playground_axes(fig)# pg.plot_points(ax, X_train, y_train, X_valid, y_valid)# # These 3 lines equal to `plot_points_with_playground_style` function# draw the decision boundary of X1 input (feature)pg.draw_decision_boundary(fig,ax,node_id=pg.InputType.X1,discretize=False)# # uncomment if a graph is not shown# import matplotlib.pyplot as plt# plt.show()
print('Signature of main @staticmethod')importsysifsys.version_info[0]<3:# inspect.signature was introduced at version Python 3.3!pipinstallfuncsigstry:frominspectimportsignatureexceptImportError:fromfuncsigsimportsignatureprint('pg.plot_sample',str(signature(pg.plot_sample)))# pg.plot_sample (data_type, noise=0.0, validation_size=0.5, visualize_validation_data=False, figsize=(5, 5), dpi=100, node_id=None, discretize=False)print('pg.generate',str(signature(pg.generate)))# pg.generate (data_type, noise=0.0)print('pg.split_data',str(signature(pg.data)))# pg.split_data (data, validation_size=0.5, label_num=1)print('pg.plot_points_with_playground_style',str(signature(pg.plot_points_with_playground_style)))# pg.plot_points_with_playground_style (X_train, y_train, X_valid=None, y_valid=None, figsize=(5, 5), dpi=100)print('pg.get_playground_figure',str(signature(pg.get_playground_figure)))# pg.get_playground_figure (enable_colorbar=False)print('pg.get_playground_axes',str(signature(pg.get_playground_axes)))# pg.get_playground_axes (fig)print('pg.plot_points',str(signature(pg.plot_points)))# pg.plot_points (ax, X_train, y_train, X_valid=None, y_valid=None)print('pg.draw_decision_boundary',str(signature(pg.draw_decision_boundary)))# pg.draw_decision_boundary (fig, ax, node_id='x', discretize=False, enable_colorbar=True)
示例Web应用程序
许可证
版权所有2018 Digital Advantage Co.，Ltd.保留所有权利。 在2.0版apache许可下授权。
使用开源代码的许可证
此项目使用以下开源代码的javascript到python转换：
tensorflow / playground (Deep playground) / dataset.ts，heatmap.ts，playground.ts，state.ts
版权所有2016谷歌公司。保留所有权利。
在2.0版apache许可下授权。
d3 / d3-scale / linear.js
版权所有2010-2015 Mike Bostock。保留所有权利。
根据BSD第3条“新”或“修订”许可进行许可。
欢迎加入QQ群-->： 979659372
