COMP 135 Introduction to Machine Learning First day of class - - PowerPoint PPT Presentation

▶
comp 135 introduction to machine learning
SMART_READER_LITE
LIVE PREVIEW

COMP 135 Introduction to Machine Learning First day of class - - PowerPoint PPT Presentation

Sep 14, 2022 114 likes •497 views

COMP 135 Introduction to Machine Learning First day of class Spring 2019 https://www.cs.tufts.edu/comp/135/2019s/ Many slides attributable to: Emily Fox (UW), Finale Doshi-Velez (Harvard), Erik Sudderth (UCI), & Liping Liu (Tufts) 1

slide-1
SLIDE 1

COMP 135 Introduction to Machine Learning

First day of class Spring 2019

1

https://www.cs.tufts.edu/comp/135/2019s/

Many slides attributable to: Emily Fox (UW), Finale Doshi-Velez (Harvard), Erik Sudderth (UCI), & Liping Liu (Tufts)

slide-2
SLIDE 2

2

Mike Hughes - Tufts COMP 135 - Spring 2019

Why Machine Learning?

slide-3
SLIDE 3

Mike Hughes - Tufts COMP 135 - Spring 2019

3

Image Credit: Emily Fox

slide-4
SLIDE 4

4

Mike Hughes - Tufts COMP 135 - Spring 2019

Artificial Intelligence (AI)

  • Study of “intelligent systems”, with many parts:

logic, planning, search, probabilistic reasoning, learning from experience, interacting with other agents, etc.

slide-5
SLIDE 5

Machine Learning (ML)

  • Study of algorithms that learn from

experience/data to perform a task

  • Task output: a prediction or a decision

5

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-6
SLIDE 6

The Machine Learning Process

6

Mike Hughes - Tufts COMP 135 - Spring 2019

Past data (or “experience”) Performance measure Task / Goal input data (now) prediction / decision (now)

slide-7
SLIDE 7

7

Why take this course?

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-8
SLIDE 8

Why take this course?

Our goal is to prepare students to effectively apply machine learning methods to problems that might arise in "the real world" -- in industry, medicine, education, and beyond. Gain skills and understanding for a future as:

  • Developer using ML “out-of-the-box”
  • ML methods researcher

8

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-9
SLIDE 9

Why take this course?

Students will be able to:

  • Think systematically
  • Compare/contrast each method’s strengths & limitations
  • Deploy rapidly
  • Hands-on experience with open-source libraries
  • Evaluate carefully
  • Design experiments with task-appropriate performance

metrics and strong baselines

  • Report uncertainty in performance

9

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-10
SLIDE 10

What skills will we learn?

10

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-11
SLIDE 11

What will we learn?

11

Mike Hughes - Tufts COMP 135 - Spring 2019

Supervised Learning Unsupervised Learning Reinforcement Learning

Data, Label Pairs Performance measure Task data x label y

{xn, yn}N

n=1

slide-12
SLIDE 12

12

Mike Hughes - Tufts COMP 135 - Spring 2019

Task: Regression

Supervised Learning Unsupervised Learning Reinforcement Learning

regression

x y y

is a continuous variable e.g. sales in $$

slide-13
SLIDE 13

13

Mike Hughes - Tufts COMP 135 - Spring 2019

Regression Example: Uber

Supervised Learning Unsupervised Learning Reinforcement Learning

regression

slide-14
SLIDE 14

Regression Example: Uber

14

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-15
SLIDE 15

Regression Example: Uber

15

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-16
SLIDE 16

16

Mike Hughes - Tufts COMP 135 - Spring 2019

Task: Classification

Supervised Learning Unsupervised Learning Reinforcement Learning

classification

y

x2 x1

is a discrete variable (red or blue)

slide-17
SLIDE 17

Classification Example: Swype

17

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-18
SLIDE 18

What will we learn?

18

Mike Hughes - Tufts COMP 135 - Spring 2019

Supervised Learning Unsupervised Learning Reinforcement Learning

Data, Label Pairs Performance measure Task data x label y

{xn, yn}N

n=1

slide-19
SLIDE 19

19

Mike Hughes - Tufts COMP 135 - Spring 2019

What will we learn?

Data Examples data x

Supervised Learning Unsupervised Learning Reinforcement Learning

{xn}N

n=1

Task summary

  • f x

Performance measure

slide-20
SLIDE 20

Task: Clustering

20

Mike Hughes - Tufts COMP 135 - Spring 2019

Supervised Learning Unsupervised Learning Reinforcement Learning

clustering

x2 x1

slide-21
SLIDE 21

Clustering Example: News

21

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-22
SLIDE 22

Task: Embedding

22

Mike Hughes - Tufts COMP 135 - Spring 2019

Supervised Learning Unsupervised Learning Reinforcement Learning

embedding

x2 x1

slide-23
SLIDE 23

Example: Genes vs. geography

23

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-24
SLIDE 24

24

Mike Hughes - Tufts COMP 135 - Spring 2019

Task: Recommendation

Supervised Learning Unsupervised Learning Reinforcement Learning

recommendation

slide-25
SLIDE 25

Recommendation Example

25

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-26
SLIDE 26

Histories of states, actions, rewards Task Next action a Performance measure

What will we learn?

26

Mike Hughes - Tufts COMP 135 - Spring 2019

Supervised Learning Unsupervised Learning Reinforcement Learning

{st, at, rt, st+1, . . .}N

n=1

Recent history

slide-27
SLIDE 27

RL example: Pancake robot

27

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-28
SLIDE 28

28

Mike Hughes - Tufts COMP 135 - Spring 2019

What will we learn this semester?

Supervised Learning Unsupervised Learning Reinforcement Learning

~ 1 week 0.5 homeworks ~ 2 weeks 1.5 homeworks ~ 10 weeks 6 homeworks 3 projects

Take COMP 136 Take COMP 150 – RL (Prof. Jivko Sinapov) Take COMP 136 Take COMP 150 – Deep Learning If I want more?

slide-29
SLIDE 29

What we won’t cover

  • Active learning
  • Transfer learning
  • Semi-supervised learning
  • Learning theory
  • ….. lots more

29

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-30
SLIDE 30

Who is teaching?

  • Instructor: Prof. Mike Hughes
  • TA Staff
  • Mike Pietras
  • Rui Chen
  • Minh Nguyen
  • Duc Nguyen
  • Wayne Tang

30

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-31
SLIDE 31

Problem: When will ICU patient need intervention?

Interventions:

  • mechanical ventilation
  • blood pressure drugs

Early prediction helps: prepare patient plan staffing try less aggressive options early

Ghassemi, Wu, Hughes, et al. AMIA CRI 2017

slide-32
SLIDE 32

Cohort from MIMIC-III dataset

36,050 patients

  • from Beth-Israel Deaconess in Boston 2001-2012
  • kept all adults with record within 6-360 hours

mimic.physionet.org

(Johnson et al. Sci. Data 2016)

slide-33
SLIDE 33

Observed data

7 nurse-validated vital signs (hourly) heart rate, blood pressure, temp., SpO2, … 11 lab measurements (much less than hourly) hematocrit, lactate, …

each channel standardized to mean=0, var=1 with carry-and-hold for missing data

slide-34
SLIDE 34

34

Mike Hughes - Tufts COMP 135 - Spring 2019

Task: predict need in advance

+2 hrs ahead

slide-35
SLIDE 35

35

Mike Hughes - Tufts COMP 135 - Spring 2019

Vasopressor prediction : 1 hr ahead

Area-under-ROC

static demographics (age, race, etc) dynamic patient vitals at time t “embedding” using time-series model

slide-36
SLIDE 36

Key Stakeholder Questions

  • How should we fill in missing values?
  • How to deal with imbalance data?
  • most patients never get drug X
  • How to deal with imbalanced mistake costs?
  • remove ventilator too early ends life, too late costs $$
  • How uncertain are predictions?
  • How will this generalize to a new hospital?

36

Mike Hughes - Tufts COMP 135 - Spring 2019

We can start to answer many of these in COMP 135

slide-37
SLIDE 37

Logistics

  • Course website (take a tour)
  • https://www.cs.tufts.edu/comp/135/2019s/
  • Discussions on Piazza
  • Lectures every Mon & Wed
  • Recitation Sessions with TAs every Mon
  • Deliverables: 2 exams, 8 homeworks, 3 projects
  • Collaboration Policy

37

Mike Hughes - Tufts COMP 135 - Spring 2019

slide-38
SLIDE 38

Let’s Get Started!

  • Setup your Python environment ASAP
  • First recitation is Mon 1/21
  • Get help with your Python env.
  • Learn basics of arrays, plotting, etc
  • HW0 due NEXT WEEK (Wed 1/23 at 11:59pm)

38

Mike Hughes - Tufts COMP 135 - Spring 2019