Created
October 15, 2019 15:10
-
-
Save negedng/0c1382d336ef9b4f1a757c9bddf4d2a0 to your computer and use it in GitHub Desktop.
Project word2vec vectors to 2D
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# Word list | |
Wl = ['man', 'woman', 'rich', 'poor', 'queen', | |
'king', 'fisherman', 'teacher', 'actress', 'actor'] | |
Wv = [] | |
for i in range(len(Wl)): | |
# Embeddings | |
Wv.append(word2vec[Wl[i]]) | |
# To-be basis | |
b1 = (Wv[1]-Wv[0]) | |
b2 = (Wv[3]-Wv[2]) | |
# Get pseudo-inverse matrix | |
W = np.array(Wv) | |
B = np.array([b1,b2]) | |
Bi = np.linalg.pinv(B.T) | |
# Project all the words | |
Wp = np.matmul(Bi,W.T) | |
Wp = (Wp.T-[Wp[0,2],Wp[1,0]]).T |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment