Will Nathan like Camille? Will Nathan vote for candidate T.? - - PowerPoint PPT Presentation

will nathan like camille will nathan vote for candidate t
SMART_READER_LITE
LIVE PREVIEW

Will Nathan like Camille? Will Nathan vote for candidate T.? - - PowerPoint PPT Presentation

Will Nathan like Camille? Will Nathan vote for candidate T.? 2 Will Nathan like Camille? Will Nathan vote for candidate T.? 3 4 5 word2vec can be


slide-1
SLIDE 1
slide-2
SLIDE 2

×

2

  • Will Nathan like Camille?

Will Nathan vote for candidate T.?

slide-3
SLIDE 3

3

  • Will Nathan like Camille?

Will Nathan vote for candidate T.?

slide-4
SLIDE 4

4

slide-5
SLIDE 5

5

≈ →

𝑋 𝑋𝑈

slide-6
SLIDE 6

6

  • word2vec can be

understood as matrix factorization!

dimensionality = number of nodes

slide-7
SLIDE 7

7

Random walks converge to Personalized PageRank 𝑡𝑗𝑛(𝑣,⋅)

𝑣 𝑤

𝑡𝑗𝑛(𝑤,⋅)

𝑣 𝑤

slide-8
SLIDE 8

𝑣

8

𝑋 𝑋𝑈

𝑣

slide-9
SLIDE 9
  • 9

Why should I bother about similarities?

slide-10
SLIDE 10

𝑡𝑗𝑛(𝑣,⋅)

𝑋~𝒪(0, 1)

𝑣 ∈ 𝑊

𝑋 𝑡𝑗𝑛(𝑣,⋅)

𝑃 𝑜2

10

𝑋 𝑋𝑈 𝑡𝑗𝑛(𝑣,⋅) 𝑣

We make it faster with sampling!

slide-11
SLIDE 11

11

Why not just using Negative Sampling?

slide-12
SLIDE 12
  • ×
  • 12
slide-13
SLIDE 13

×

13

slide-14
SLIDE 14

×

14

slide-15
SLIDE 15

×

15

slide-16
SLIDE 16

16

slide-17
SLIDE 17

Thank you for attention!

Anton Tsitsulin Knowledge Discovery and Data Mining Hasso Plattner Institute

Thank you for your attention!

bit.ly/www-verse github.com/xgfs/verse