本帖最后由 M丶Sulayman 于 2018-5-7 17:36 编辑
- #!/usr/bin/env python
- # -*- coding: UTF-8 -*-
- from __future__ import print_function
- import numpy as np
- from sklearn.model_selection import train_test_split
- import tensorflow as tf
- from tensorflow.contrib.tensor_forest.python import tensor_forest
- from tensorflow.python.ops import resources
- def load_data(file):
- features = []
- lables = []
- file = open(file, 'r')
- lines = file.readlines()
- for line in lines:
- items = line.strip().split(',')
- list_to_string = ','.join(items)
- for ch in ['Iris-setosa']:
- if ch in list_to_string:
- list_to_string = list_to_string.replace(ch, '0')
- for ch in ['Iris-versicolor']:
- if ch in list_to_string:
- list_to_string = list_to_string.replace(ch, '1')
- for ch in ['Iris-virginica']:
- if ch in list_to_string:
- list_to_string = list_to_string.replace(ch, '2')
- items = list_to_string.strip().split(',')
- features.append([float(items[i]) for i in range(len(items) - 1)])
- lables.append(float(items[-1]))
- return np.array(features), np.array(lables)
- if __name__ == '__main__':
- data, lables = load_data('iris.csv')
- train_data, test_data, train_lables, test_lables = train_test_split(data, lables, test_size=0.3, random_state=33)
- # Parameters
- num_steps = 1000
- num_classes = 3
- num_features = 4
- num_trees = 10
- max_nodes = 100
- # Input and Target data
- X = tf.placeholder(tf.float32, shape=[None, num_features])
- # For random forest, labels must be integers (the class id)
- Y = tf.placeholder(tf.int32, shape=[None])
- # Random Forest Parameters
- hparams = tensor_forest.ForestHParams(num_classes=num_classes,
- num_features=num_features,
- num_trees=num_trees,
- max_nodes=max_nodes).fill()
- # Build the Random Forest
- forest_graph = tensor_forest.RandomForestGraphs(hparams)
- # Get training graph and loss
- train_op = forest_graph.training_graph(X, Y)
- loss_op = forest_graph.training_loss(X, Y)
- # Measure the accuracy
- infer_op = forest_graph.inference_graph(X)
- correct_prediction = tf.equal(tf.argmax(infer_op, 1), tf.cast(Y, tf.int64))
- accuracy_op = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
- # Initialize the variables (i.e. assign their default value) and forest resources
- init_vars = tf.group(tf.global_variables_initializer(),
- resources.initialize_resources(resources.shared_resources()))
- # Start TensorFlow session
- sess = tf.Session()
- # Run the initializer
- sess.run(init_vars)
- # Training
- for i in range(1, num_steps + 1):
- _, l = sess.run([train_op, loss_op], feed_dict={X: train_data, Y: train_lables})
- if i % 50 == 0 or i == 1:
- acc = sess.run(accuracy_op, feed_dict={X: train_data, Y: train_lables})
- print('Step %i, Loss: %f, Acc: %f' % (i, l, acc))
- # Test Model
- print("Test Accuracy:", sess.run(accuracy_op, feed_dict={X: test_data, Y: test_lables}))
我知道答案
回答被采纳将会获得 10 金币 + 50 金币 已有25人回答
|
本帖子中包含更多资源
您需要 登录 才可以下载或查看，没有帐号？加入社区
x
-
green:
楼主好！ 我最近在学习tensor forest. 但是有个问题一直不明白，就是那个num_steps = 1000的作用是什么？random forest不是遍历一次就可以训练好了吗？为什么要有好多个step？如果是要训练一个很大的样本集，tensorforest到底是如何进行训练的？是通过这个steps用来将很大的样本集分成batch分段训练？每段之间的训练是如何连接的？
-
HiHotzenplotz:
回复 green :
At the start of training, the tree structure is initialized to a root node, and the leaf and growing statistics for it are both empty. Then, for each batch {(x_i, y_i)} of training data, the following steps are performed:
Given the current tree structure, each x_i is used to find the leaf assignment l_i.
y_i is used to update the leaf statistics of leaf l_i.
If the growing statistics for the leaf l_i do not yet contain num_splits_to_consider splits, x_i is used to generate another split. Specifically, a random feature value is chosen, and x_i's value at that feature is used for the split's threshold.
Otherwise, (x_i, y_i) is used to update the statistics of every split in the growing statistics of leaf l_i. If leaf l_i has now seen split_after_samples data points since creating all of its potential splits, the split with the best score is chosen, and the tree structure is grown.
-
|
加载中...
浙江移动 — 利用 TensorFlow 实现通
文 / 浙江移动网管中心业务系统工程师，邢彪
中国移动近年来积极探索人工智能在通讯
金山 WPS：基于 TensorFlow 的 AI 移
文 / 赵威 胡旭华，金山 WPS 团队
让 AI 的能力下沉到更多移动终端上，才能完全打开
扇贝 : 应用 TensorFlow 实现深度知
文 / 张汝宸 张志博，扇贝算法团队
背景
扇贝，作为一个拥有超过八千万用户的移动英
当 TensorFlow Lite 遇到有道翻译王
文 / 有道 AI 技术团队
2018 年 9 月 6 日下午，网易有道发布了有道翻译王 2.0 Pro。
爱作业口算同步练，TensorFlow Lite
文 / 杭州大拿科技股份有限公司
爱作业是一款通过 AI 技术自动批改数学作业的 APP，
，真是幸福
浙公网安备 33010802009243号