06 Word2Vec model (the first model dedicated to word vectors, CBOW and Skip-gram)


Neural Network Language Model (NNLM) – "To predict the next word

NNLM() – "predict the next word

Neural network + language model: use a neural network to solve two tasks related to human speech

softmax(w2(tanh((w1x+b1)))+b2)

get a by-product (word vector)

Q matrix, for any one-hot encoded word vector, a new word vector can be obtained through the Q matrix

  1. Dimensions can be converted
  2. There is also a relationship between word vectors between similar words

Guess you like

Origin blog.csdn.net/linjie_830914/article/details/132173893