1. 程式人生 > >利用 TensorFlow 入門 Word2Vec

利用 TensorFlow 入門 Word2Vec

我認為學習演算法的最好方法就是嘗試去實現它,因此這個教程我們就來學習如何利用 TensorFlow 來實現詞嵌入。

這篇文章我們不會去過多的介紹一些詞向量的內容,所以很多 king - man - woman - queue 的例子會被省去,直接進入編碼實踐過程。

我們如何設計這些詞嵌入?

對於如何設計詞嵌入有很多的技術,這裡我們討論一種非常有名的技術。與我們往常的認知不同,word2vec 並不是一個深層的網路,它只是一個三層的淺層網路。

注意:word2vec 有很多的技術細節,但是我們會跳過這些細節,來使得更加容易理解。

word2vec 如何工作?

word2vec 演算法的設計如下:

  1. 它是一個三層的網路(一個輸入層 + 一個隱藏層 + 一個輸出層)。
  2. 模型輸入一個詞,然後去預測它周圍的詞。
  3. 移除最後一層(輸出層),保留輸入層和隱藏層。
  4. 現在,輸入一個詞庫中的詞,然後隱藏層的輸出就是輸入詞的詞向量。

就是這麼簡單,這個三層網路就可以得到一個還不錯的詞向量。

接下來就讓我們來實現這個模型。完整的程式碼可以點選 Github,但我建議你先不要看完整的程式碼,先一步一步學習。

接下來,我們先定義我們要處理的原始文字:

import numpy as np
import tensorflow as tf
corpus_raw = 'He is the king . The king is
royal . She is the royal queen ' # convert to lower case corpus_raw = corpus_raw.lower()

現在,我們需要將輸入的原始文字資料轉換成一個輸入輸出對,以便我們對輸入的詞,可以去預測它附近的詞。比如,我們確定一箇中心詞, 視窗大小 window_size 設定為 n ,那麼我們就是去預測中心詞前面 n 個詞和後面 n 個詞。Chris McCormick 的這篇部落格給出了比較詳細的解釋。

A training sample generation with a window size of 2.

注意:如果中心詞是在句子的開頭或者末尾,那麼我們就忽略視窗無法獲得的詞。

在做這個之前,我們需要建立一個字典,用來確定每個單詞的索引,具體如下:

words = []
for word in corpus_raw.split():
    if word != '.': # because we don't want to treat . as a word
        words.append(word)
words = set(words) # so that all duplicate words are removed
word2int = {}
int2word = {}
vocab_size = len(words) # gives the total number of unique words
for i,word in enumerate(words):
    word2int[word] = i
    int2word[i] = word

這個字典的執行結果如下:

print(word2int['queen'])
-> 42 (say)

print(int2word[42])
-> 'queen'

接下來,我們將我們的句子向量轉換成單詞列表,如下:

# raw sentences is a list of sentences.
raw_sentences = corpus_raw.split('.')
sentences = []
for sentence in raw_sentences:
    sentences.append(sentence.split())

上面程式碼將幫助我們得到一個句子的列表,列表中的每一個元素是句子的單詞列表,如下:

print(sentences)

-> [['he', 'is', 'the', 'king'], ['the', 'king', 'is', 'royal'], ['she', 'is', 'the', 'royal', 'queen']]

接下來,我們要產生我們的訓練資料:

data = []

WINDOW_SIZE = 2

for sentence in sentences:
    for word_index, word in enumerate(sentence):
        for nb_word in sentence[max(word_index - WINDOW_SIZE, 0) : min(word_index + WINDOW_SIZE, len(sentence)) + 1] : 
            if nb_word != word:
                data.append([word, nb_word])

這個程式給出了單詞輸入輸出對,我們將視窗的大小設定為 2。

print(data)
[['he', 'is'],
 ['he', 'the'],
 ['is', 'he'],
 ['is', 'the'],
 ['is', 'king'],
 ['the', 'he'],
 ['the', 'is'], 
.
.
.
]

至此,我們有了我們的訓練資料,但是我們需要將它轉換成計算機可以理解的表示,即數字。也就是我們之前設計的 word2int 字典。

我們再進一步表示,將這些數字轉換成 0-1 向量。

i.e., 
say we have a vocabulary of 3 words : pen, pineapple, apple
where 
word2int['pen'] -> 0 -> [1 0 0]
word2int['pineapple'] -> 1 -> [0 1 0]
word2int['apple'] -> 2 -> [0 0 1]

那麼為什麼要表示成 0-1 向量呢?這個問題我們後續討論。

# function to convert numbers to one hot vectors
def to_one_hot(data_point_index, vocab_size):
    temp = np.zeros(vocab_size)
    temp[data_point_index] = 1
    return temp
x_train = [] # input word
y_train = [] # output word
for data_word in data:
    x_train.append(to_one_hot(word2int[ data_word[0] ], vocab_size))
    y_train.append(to_one_hot(word2int[ data_word[1] ], vocab_size))
# convert them to numpy arrays
x_train = np.asarray(x_train)
y_train = np.asarray(y_train)

現在,我們有了 x_trainy_train 資料:

print(x_train)
->
[[ 0.  0.  0.  0.  0.  0.  1.]
 [ 0.  0.  0.  0.  0.  0.  1.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  1.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  1.  0.  0.  0.  0.  0.]
 [ 0.  1.  0.  0.  0.  0.  0.]
 [ 0.  0.  1.  0.  0.  0.  0.]
 [ 0.  0.  1.  0.  0.  0.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  0.  1.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  0.  0.  0.  1.  0.  0.]
 [ 0.  1.  0.  0.  0.  0.  0.]
 [ 0.  1.  0.  0.  0.  0.  0.]
 [ 0.  1.  0.  0.  0.  0.  0.]
 [ 1.  0.  0.  0.  0.  0.  0.]
 [ 1.  0.  0.  0.  0.  0.  0.]]

這兩個資料的維度如下:

print(x_train.shape, y_train.shape)
->
(34, 7) (34, 7)
# meaning 34 training points, where each point has 7 dimensions

構造 TensorFlow 模型

# making placeholders for x_train and y_train

x = tf.placeholder(tf.float32, shape=(None, vocab_size))
y_label = tf.placeholder(tf.float32, shape=(None, vocab_size))

從上圖中可以看出,我們將訓練資料轉換成了另一種向量表示。

EMBEDDING_DIM = 5 # you can choose your own number
W1 = tf.Variable(tf.random_normal([vocab_size, EMBEDDING_DIM]))
b1 = tf.Variable(tf.random_normal([EMBEDDING_DIM])) #bias
hidden_representation = tf.add(tf.matmul(x,W1), b1)

接下來,我們對隱藏層的資料進行處理,並且對其附近的詞進行預測。預測詞的方法我們採用 softmax 方法。

W2 = tf.Variable(tf.random_normal([EMBEDDING_DIM, vocab_size]))
b2 = tf.Variable(tf.random_normal([vocab_size]))
prediction = tf.nn.softmax(tf.add( tf.matmul(hidden_representation, W2), b2))

所以,完整的模型是:

input_one_hot  --->  embedded repr. ---> predicted_neighbour_prob
predicted_prob will be compared against a one hot vector to correct it.

現在,我們可以訓練這個模型:

sess = tf.Session()
init = tf.global_variables_initializer()
sess.run(init) #make sure you do this!
# define the loss function:
cross_entropy_loss = tf.reduce_mean(-tf.reduce_sum(y_label * tf.log(prediction), reduction_indices=[1]))
# define the training step:
train_step = tf.train.GradientDescentOptimizer(0.1).minimize(cross_entropy_loss)
n_iters = 10000
# train for n_iter iterations
for _ in range(n_iters):
    sess.run(train_step, feed_dict={x: x_train, y_label: y_train})
    print('loss is : ', sess.run(cross_entropy_loss, feed_dict={x: x_train, y_label: y_train}))

在訓練的過程中,你在控制檯可以得到如下結果:

loss is :  2.73213
loss is :  2.30519
loss is :  2.11106
loss is :  1.9916
loss is :  1.90923
loss is :  1.84837
loss is :  1.80133
loss is :  1.76381
loss is :  1.73312
loss is :  1.70745
loss is :  1.68556
loss is :  1.66654
loss is :  1.64975
loss is :  1.63472
loss is :  1.62112
loss is :  1.6087
loss is :  1.59725
loss is :  1.58664
loss is :  1.57676
loss is :  1.56751
loss is :  1.55882
loss is :  1.55064
loss is :  1.54291
loss is :  1.53559
loss is :  1.52865
loss is :  1.52206
loss is :  1.51578
loss is :  1.50979
loss is :  1.50408
loss is :  1.49861
.
.
.

隨著損失值的不斷下降,最終會達到一個穩定值。即使我們無法獲得很精確的結果,但是我們也不在乎,因為我們感興趣的是 W1 和 b1 的值,即隱藏層的權重。

讓我們來看看這些權重,如下:

print(sess.run(W1))
print('----------')
print(sess.run(b1))
print('----------')

->
[[-0.85421133  1.70487809  0.481848   -0.40843448 -0.02236851]
 [-0.47163373  0.34260952 -2.06743765 -1.43854153 -0.14699034]
 [-1.06858993 -1.10739779  0.52600187  0.24079895 -0.46390489]
 [ 0.84426647  0.16476244 -0.72731972 -0.31994426 -0.33553854]
 [ 0.21508843 -1.21030915 -0.13006891 -0.24056002 -0.30445012]
 [ 0.17842589  2.08979321 -0.34172744 -1.8842833  -1.14538431]
 [ 1.61166084 -1.17404735 -0.26805425  0.74437028 -0.81183684]]
----------
[ 0.57727528 -0.83760375  0.19156453 -0.42394346  1.45631313]
----------

為什麼採用 0-1 向量?

again from Chris McCormick’s article (do read it)

當我們將一個 0-1 向量與 W1 相乘時,我們基本上可以將 W1 與 0-1 向量對應的那個 1 相乘的結果就是詞向量。也就是說, W1 就是一個數據查詢表。

在我們的程式中,我們也添加了一個偏置項 b1 ,所以我們也需要將它加上。

vectors = sess.run(W1 + b1)

# if you work it out, you will see that it has the same effect as running the node hidden representation
print(vectors)
->
[[-0.74829113 -0.48964909  0.54267412  2.34831429 -2.03110814]
 [-0.92472583 -1.50792813 -1.61014366 -0.88273793 -2.12359881]
 [-0.69424796 -1.67628145  3.07313657 -1.14802659 -1.2207377 ]
 [-1.7077738  -0.60641652  2.25586247  1.34536338 -0.83848488]
 [-0.10080346 -0.90931684  2.8825531  -0.58769202 -1.19922316]
 [ 1.49428082 -2.55578995  2.01545811  0.31536022  1.52662396]
 [-1.02735448  0.72176981 -0.03772151 -0.60208392  1.53156447]]

如果我們想得到 queen 的向量,我們可以用如下表示:

print(vectors[ word2int['queen'] ])
# say here word2int['queen'] is 2
-> 
[-0.69424796 -1.67628145  3.07313657 -1.14802659 -1.2207377 ]

那麼這些漂亮的向量有什麼用呢?

我們寫一個如何去查詢最相近向量的函式,當然這個寫法是非常簡單粗糙的。

def euclidean_dist(vec1, vec2):
    return np.sqrt(np.sum((vec1-vec2)**2))

def find_closest(word_index, vectors):
    min_dist = 10000 # to act like positive infinity
    min_index = -1
    query_vector = vectors[word_index]
    for index, vector in enumerate(vectors):
        if euclidean_dist(vector, query_vector) < min_dist and not np.array_equal(vector, query_vector):
            min_dist = euclidean_dist(vector, query_vector)
            min_index = index
    return min_index

接下來,讓我們來測試一下單詞 king ,queen 和 royal 這些詞。

print(int2word[find_closest(word2int['king'], vectors)])
print(int2word[find_closest(word2int['queen'], vectors)])
print(int2word[find_closest(word2int['royal'], vectors)])

->
queen
king
he

我們可以得到如下有趣的結果。

king is closest to queen
queen is closest to king
royal is closest to he

第三個資料是我們根據大型語料庫得出來的(看起來還不錯)。語料庫的資料更大,我們得到的結果會更好。(注意:由於權重是隨機初始化的,所以我們可能會得到不同的結果,如果有需要,我們可以多執行幾次。)

讓我們來畫出這個向量相關圖。

首先,我們需要利用將為技術將維度從 5 減小到 2,所用的技術是:tSNE(teesnee!)

from sklearn.manifold import TSNE
model = TSNE(n_components=2, random_state=0)
np.set_printoptions(suppress=True)
vectors = model.fit_transform(vectors)

然後,我們需要對結果進行規範化,以便我們可以在 matplotlib 中更好的對它進行檢視。

from sklearn import preprocessing
normalizer = preprocessing.Normalizer()
vectors =  normalizer.fit_transform(vectors, 'l2')

最後,我們將繪製出圖。

import matplotlib.pyplot as plt
fig, ax = plt.subplots()
for word in words:
    print(word, vectors[word2int[word]][1])
    ax.annotate(word, (vectors[word2int[word]][0],vectors[word2int[word]][1] ))
plt.show()

從圖中,我們可以看出。shequeen 的距離非常接近,kingroyal 的距離和 kingqueen 的距離相同。如果我們有一個更大的語料庫,我們可以得到更加複雜的關係圖。

為什麼會發生這些?

我們給神經網路的任務是預測單詞的相鄰詞。但是我們還沒有具體的分析神經網路是如何預測的。因此,神經網路找出單詞的向量表示,用來幫助它預測相鄰詞這個任務。預測相鄰詞這本身不是一個有趣的任務,我們關心的是隱藏層的向量表示。

為了得到這些表示,神經網路使用了上下文資訊。在我們的語料庫中,king 和 royal 是作為相鄰詞出現的,queen 和 royal 也是作為相鄰詞出現的。

為什麼把預測相鄰詞作為一個任務?

其他的任務也可以用來訓練這個詞向量任務,比如利用 n-gram 就可以訓練出很好的詞向量!這裡有一篇部落格有詳細解釋。

那麼,我們為什麼還要使用相鄰詞預測作為任務呢?因為有一個比較著名的模型稱為 skip gram 模型。我們可以使用中間詞的相鄰單詞作為輸入,並要求神經網路去預測中間詞。這被稱為連續詞袋模型。

總結

  • 詞向量是非常酷的一個工具。
  • 不要在實際生產環境中使用這個 TensorFlow 程式碼,我們這裡只是為了理解才這樣寫。生產環境建議使用一些成熟的工具包,比如 gensim

我希望這個簡單教程可以幫助到一些人,可以更加深刻的理解什麼是詞向量。

CoderPai 是一個專注於演算法實戰的平臺,從基礎的演算法到人工智慧演算法都有設計。如果你對演算法實戰感興趣,請快快關注我們吧。加入AI實戰微信群,AI實戰QQ群,ACM演算法微信群,ACM演算法QQ群。長按或者掃描如下二維碼,關注 “CoderPai” 微訊號(coderpai)
這裡寫圖片描述

相關推薦

利用 TensorFlow 入門 Word2Vec

我認為學習演算法的最好方法就是嘗試去實現它,因此這個教程我們就來學習如何利用 TensorFlow 來實現詞嵌入。 這篇文章我們不會去過多的介紹一些詞向量的內容,所以很多 king - man - woman - queue 的例子會被省去,直接進

利用Tensorflow進行自然語言處理(NLP)系列之二高階Word2Vec

一、概述 在上一篇中,我們介紹了Word2Vec即詞向量,對於Word Embeddings即詞嵌入有了些基礎,同時也闡述了Word2Vec演算法的兩個常見模型 :Skip-Gram模型和CBOW模型,本篇會對兩種演算法做出比較分析並給出其擴充套件模型-GloVe模型。

利用Tensorflow實現神經網絡模型

flow one 什麽 hold test ase tensor dom def 首先看一下神經網絡模型,一個比較簡單的兩層神經。 代碼如下: # 定義參數 n_hidden_1 = 256 #第一層神經元 n_hidden_2 = 128 #第

tensorflow實現Word2vec

while brush ber ear same split max ems red # coding: utf-8 ‘‘‘ Note: Step 3 is missing. That‘s why I left it. ‘‘‘ from __future__ impor

TensorFlow入門:mac 安裝 TensorFlow

sse ssi 來安 pan lan bin pat ont world 開發環境: mac os 10.12.5 Python 2.7.10 GCC 4.2.1 mac默認是不帶pip的,安裝pip。 sudo easy_install pip 1.安裝vir

TensorFlow入門:安裝常用的依賴模塊

包含 librosa 任務 nltk plot orf 入門 learn sci TensorFlow運行中常用到一些Python第三方模塊: numpy 存儲和處理大型矩陣的科學計算包 maplotlib 最著名的繪圖庫 jupyter scikit-image 圖像預

TensorFlow入門

imp 我們 class rank node strong ima orf 運行 Tensor(張量)是TensorFlow中最核心的數據結構單元,ta它可以表示任意維數的數組。wei‘du維度用rank(秩)表示。 例如: 3

TensorFlow文檔翻譯-01-TensorFlow入門

left https 你是 但是 sam return 很多 等級 license TensorFlow入門 這是關於如何開始tensorFlow的指南。開始之前,你需要先安裝TensorFlow。除此之外,你應該了解: 知道如何使用Python編程。 懂一點點數組 如

tensorflow:Google】三、tensorflow入門

als 管理 神經網絡 等價 問題 sign ria init 節點 【一】計算圖模型 節點是計算,邊是數據流, a = tf.constant( [1., 2.] )定義的是節點,節點有屬性 a.graph 取得默認計算圖  g1 = tf.get_default_gr

利用Tensorflow實現手寫字符識別

status ade 模式 數學 malloc interact tutorials x模型 項目 模式識別領域應用機器學習的場景非常多,手寫識別就是其中一種,最簡單的數字識別是一個多類分類問題,我們借這個多類分類問題來介紹一下google最新開源的tensorflow框架

tensorflow入門 (一)

統一 .sh 函數 soft acf layers 估計 一個地方 aid 轉載:作者:地球的外星人君鏈接:https://www.zhihu.com/question/49909565/answer/207609620來源:知乎著作權歸作者所有。商業轉載請聯系作者獲得授權

轉:TensorFlow入門(六) 雙端 LSTM 實現序列標註(分詞)

vsm max poc 代碼 單詞 arch 大致 雙端 fun http://blog.csdn.net/Jerr__y/article/details/70471066 歡迎轉載,但請務必註明原文出處及作者信息。 @author: huangyongye @creat_

TensorFlow入門必讀教程,拿走不謝!

活性 一起 操作系統 得到 haskell 開源 ask 服務 好的 TensorFlow 是一個開源的深度學習框架,於 2015 年末發布後,它成為了在全球得到最廣泛采用的深度學習框架之一。 深度學習神經網絡通常包含許多層。它們使用多維數組在不同層之間傳輸數據或執行操作。

TensorFlow入門筆記(一)基本操作

result 官方教程 with orf print blog res ont constant 0. 環境配置 安裝Anaconda,python3環境,然後利用conda命令配置的tensorflow環境。 參考極客學院翻譯TensorFlow官方教程:http://w

TensorFlow 入門

測試 art auto eight model img blog 選擇 edi 一、初識Tensorflow 編譯Tensorflow lite tflitecamerademo 安裝Tensorflow 通過下面鏈接安裝 http://www.tensorfly.

TensorFlow入門測試程序

ini closed ict dict clas pla 入門 ntop pos 1 import tensorflow as tf 2 from tensorflow.examples.tutorials.mnist import input_data 3

tensorflow入門——3解決問題——4讓我們開始吧

因此 海量 tensor flow 改變 80年 解決 識別 學習 深度學習適合解決海量數據和復雜問題 在機器學習中,語音識別,圖像識別,語意識別用的是不同的技術,從事相關工作的人合作幾乎不可能。 深度學習改變了這一切。 80年代計算機很慢,數據集很小,因此深度學習沒有得到

TensorFlow入門——hello

ESS constant TE 報錯 結果 port In 成功 說了 上一節說了TensorFlow的安裝,這一節說一下測試的問題 新建一個Python文件,輸入 1 import tensorflow as tf 2 hello = tf .constant (’He

TensorFlow入門——安裝(帶GPU)

incr per 下載地址 變量 環境 warning ice mman sign 這一系列基本上是屬於我自己進行到了那個步驟就做到那個步驟的 由於新裝了GPU (GTX750ti)和CUDA9.0、CUDNN7.1版本的軟件,所以希望TensorFlow能在GPU上運行,

知乎TensorFlow入門學習記錄

nbsp 網絡 lan ID span tar orf 形參 sta 知乎地址:https://zhuanlan.zhihu.com/p/30487008 import tensorflow as tf a=tf.placeholder(tf.int16) # 接受