Explain it like Im 5 AI, ML, NLP, and Deep Learning Kathryn Hume, - - PowerPoint PPT Presentation

▶
explain it like i m 5
SMART_READER_LITE
LIVE PREVIEW

Explain it like Im 5 AI, ML, NLP, and Deep Learning Kathryn Hume, - - PowerPoint PPT Presentation

Mar 29, 2024 296 likes •801 views

Explain it like Im 5 AI, ML, NLP, and Deep Learning Kathryn Hume, Sales & Marketing @humekathryn | kathryn@fastforwardlabs.com Arti fj cial intelligence is whatever computers cannot do until they can. Arti fj cial intelligence is

slide-1
SLIDE 1

Kathryn Hume, Sales & Marketing @humekathryn | kathryn@fastforwardlabs.com

Explain it like I’m 5

AI, ML, NLP, and Deep Learning

slide-2
SLIDE 2
slide-3
SLIDE 3
slide-4
SLIDE 4

“Artifjcial intelligence is whatever computers cannot do until they can.”

slide-5
SLIDE 5
slide-6
SLIDE 6
slide-7
SLIDE 7
slide-8
SLIDE 8
slide-9
SLIDE 9
slide-10
SLIDE 10

Artifjcial intelligence is uncoupled from consciousness

slide-11
SLIDE 11

Artifjcially intelligent systems are idiot savants, not Renaissance Men

slide-12
SLIDE 12

“Machine learning is the study of computer systems that automatically improve with experience.”

slide-13
SLIDE 13

AI? Machine Learning Data Science Analytics “Big Data”

slide-14
SLIDE 14

Supervised and Unsupervised Learning

slide-15
SLIDE 15

Unsupervised Learning

slide-16
SLIDE 16
slide-17
SLIDE 17

Supervised Learning Find a function that defjnes a correlation between P and C Use this function to make guesses about C Find a proxy (P) for something hard to know (C)

slide-18
SLIDE 18

Use square footage (P) to predict housing prices (C)

slide-19
SLIDE 19

Use “Nigerian Prince” (P) to predict if emails are spam (C)

slide-20
SLIDE 20

Use past behavior (P) to predict future preferences (C)

slide-21
SLIDE 21

What P should we pick to decide if it’s a cat or dog?

slide-22
SLIDE 22

Deep Learning

  • Use layers to transform complex input into mathematical expressions
  • Remove need for human to select which features matter
slide-23
SLIDE 23

Universal Approximation Theorem Neural networks can approximate arbitrary functions

slide-24
SLIDE 24
slide-25
SLIDE 25

Dog!

slide-26
SLIDE 26

X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 …. W1 W2 W3 W4 W5 W6 W7 W8 W9 W10 ….

“x” =

Y1 Y2 Y3 …

slide-27
SLIDE 27

X “x” W = Y

  • ne equation

three variables

slide-28
SLIDE 28

X “x” W = Y

Known Known Unknown

slide-29
SLIDE 29

X “x” W = Y

Known Unknown Known

slide-30
SLIDE 30

2 x 3 = Y

slide-31
SLIDE 31

2 x w = 6

slide-32
SLIDE 32

w = 6 / 2

slide-33
SLIDE 33

w = 6 / 2

slide-34
SLIDE 34

0 = 2 x w - 6

slide-35
SLIDE 35

Error = |2 x w - 6|

slide-36
SLIDE 36

6 4 2 1 0.5 0.2 0.1 0.06 0.02 0.0002

1 2 3 4 5 6 7 1 2 3 4 5 6 7 8 9 10

slide-37
SLIDE 37

w = 2.999

(close enough)

slide-38
SLIDE 38

Supervised Learning: Recap

Find a function that describes how these two things are correlated. (Solve for W through iteration) Use this function to make guesses about the thing that’s hard to know. (Use W to solve for new Ys) Identify a correlation between something easy to know and hard to know. (X and Y)

slide-39
SLIDE 39

Natural Language Processing

slide-40
SLIDE 40
slide-41
SLIDE 41
slide-42
SLIDE 42
slide-43
SLIDE 43

The real impact lies in making complex data simple.

There’s been a rise in sales!

slide-44
SLIDE 44

Developments in Language Processing

Traditional NLP N-grams Word Embeddings

slide-45
SLIDE 45

Bolukbasi, Chang, Zou, Saligrama, Kalai, 2016

Man : King :: Woman : Queen Man : Computer Programmer :: Woman : Homemaker Black Male : Assaulted :: White Male: Entitled To

Inherent Bias in Word Embeddings

slide-46
SLIDE 46
slide-47
SLIDE 47
slide-48
SLIDE 48
slide-49
SLIDE 49
slide-50
SLIDE 50

Thank you!

@Humekathryn | kathryn@fastforwardlabs.com