What is Machine Learning?

“Learning is any process by which a system improves

performance from experience.”
- Herbert Simon

Definition by Tom Mitchell (1998):

Machine Learning is the study of algorithms that
• improve their performance P
• at some task T
• with experience E.
A well-defined learning task is given by <P, T,
E>.
 Traditional Programming

Data
Computer Output

Program
Machine Learning

Data
Computer Program

Output 4
Slide credit: Pedro Domingos
 When Do We Use Machine Learning?
ML is used when:
• Human expertise does not exist (navigating on Mars)
• Humans can’t explain their expertise (speech recognition)
• Models must be customized (personalized medicine)
• Models are based on huge amounts of data (genomics)

Learning isn’t always useful:

• There is no need to “learn” to calculate payroll
5
Based on slide by E. Alpaydin
 A classic example of a task that requires machine learning:
It is very hard to say what makes a 2

6
Slide credit: Geoffrey Hinton
 Some more examples of tasks that are best
solved by using a learning algorithm
• Recognizing patterns:
– Facial identities or facial expressions
– Handwritten or spoken words
– Medical images
• Generating patterns:
– Generating images or motion sequences
• Recognizing anomalies:
– Unusual credit card transactions
– Unusual patterns of sensor readings in a nuclear power plant
• Prediction:
– Future stock prices or currency exchange rates
7
Slide credit: Geoffrey Hinton
 Sample Applications
• Web search
• Computational biology
• Finance
• E-commerce
• Space exploration
• Robotics
• Information extraction
• Social networks
• Debugging software
• [Your favorite area]

8
Slide credit: Pedro Domingos
 Samuel’s Checkers-Player
“Machine Learning: Field of study that gives
computers the ability to learn without being
explicitly programmed.” -Arthur Samuel
(1959)

9
 Defining the Learning Task
Improve on task T, with respect to
performance metric P, based on experience E
T: Playing checkers
P: Percentage of games won against an arbitrary opponent
E: Playing practice games against itself

T: Recognizing hand-written words

P: Percentage of words correctly classified
E: Database of human-labeled images of handwritten words

T: Driving on four-lane highways using vision sensors

P: Average distance traveled before a human-judged error
E: A sequence of images and steering commands recorded while
observing a human driver.

T: Categorize email messages as spam or legitimate.

P: Percentage of email messages correctly classified. 10
Slide credit: RayE:Mooney
Database of emails, some with human-given labels
State of the Art Applications of
Machine Learning

11
 Autonomous Cars

• Nevada made it legal for

autonomous cars to drive on
roads in June 2011
• As of 2013, four states (Nevada,
Florida, California, and
Michigan) have legalized
autonomous cars
Penn’s Autonomous Car 🡪
12
(Ben Franklin Racing Team)
Autonomous Car Sensors

13
 Autonomous Car Technology
Path

Planning

Laser Terrain Mapping

Learning from Human Drivers

Adaptive Vision

Sebastian

Stanley

Images and movies taken from Sebastian Thrun’s multimedia

1 4
Deep Learning in the Headlines

15
 Deep Belief Net on Face Images
object models

object parts
(combination
of edges)

edges

pixels
Based on materials 16
by Andrew Ng
 Learning of Object Parts

17
Slide credit: Andrew Ng
 Training on Multiple Objects

Trained on 4 classes (cars, faces,

motorbikes, airplanes).
Second layer: Shared-features
and object-specific features.
Third layer: More specific
features.

18
Slide credit: Andrew Ng
 Scene Labeling via Deep Learning

[Farabet et al. ICML 2012, PAMI 2013] 19

 Inference from Deep Learned Models
Generating posterior samples from faces by “filling in” experiments
(cf. Lee and Mumford, 2003). Combine bottom-up and top-down inference.

Input images

Samples from
feedforward
Inference
(control)

Samples from
Full posterior
inference

20
Slide credit: Andrew Ng
 Machine Learning in
Automatic Speech Recognition
A Typical Speech Recognition System

ML used to predict of phone states from the sound spectrogram

Deep learning has state-of-the-art results

# Hidden Layers 1 2 4 8 10 12

Word Error Rate % 16.0 12.8 11.4 10.9 11.0 11.1

Baseline GMM performance = 15.4%

[Zeiler et al. “On rectified linear units for speech
recognition” ICASSP 2013]
21
 Impact of Deep Learning in Speech Technology

22
Slide credit: Li Deng, MS Research
Types of Learning

23
 Types of Learning

• Supervised (inductive) learning

– Given: training data + desired outputs (labels)
• Unsupervised learning
– Given: training data (without desired outputs)
• Semi-supervised learning
– Given: training data + a few desired outputs
• Reinforcement learning
– Rewards from sequence of actions

24
Based on slide by Pedro Domingos
 Unsupervised Learning

Organize computing clusters Social network analysis

Image credit: NASA/JPL-Caltech/E. Churchwell (Univ. of Wisconsin, Madison)

Market segmentation Astronomical data analysis 33

Slide credit: Andrew Ng
 Unsupervised Learning
• Independent component analysis – separate a
combined signal into its original sources

34
Image credit: statsoft.com Audio from http://www.ism.ac.jp/~shiro/research/blindsep.html
 Unsupervised Learning
• Independent component analysis – separate a
combined signal into its original sources

35
Image credit: statsoft.com Audio from http://www.ism.ac.jp/~shiro/research/blindsep.html
 Reinforcement Learning
• Given a sequence of states and actions with
(delayed) rewards, output a policy
– Policy is a mapping from states 🡪 actions that
tells you what to do in a given state
• Examples:
– Credit assignment problem
– Game playing
– Robot in a maze
– Balance a pole on your hand
36
 The Agent-Environment Interface

Agent and environment interact at discrete time : t=

steps Agent observes state at step t: 0, 1, 2, K

t ∈S
sproduces action at step t : at ∈ A(st
)
gets resulting reward : rt +1
and resulting next ∈ℜ
state : s
t +1
... st r
t +1 st +1
r
t +2 st +2 r ...
t +3 s
at a a t +3 a
t +1 t +2 t +3
37
Slide credit: Sutton & Barto
 Reinforcement Learning

https://www.youtube.com/watch?v=4cgWya-wjgY 38
 Inverse Reinforcement Learning
• Learn policy from user demonstrations

Stanford Autonomous Helicopter

http://heli.stanford.edu/ https://
www.youtube.com/watch?v=VCdxqn0fcnE
39
Framing a Learning Problem

40
 Designing a Learning System
• Choose the training experience
• Choose exactly what is to be learned
– i.e. the target function
• Choose how to represent the target function
• Choose a learning algorithm to infer the target
function from the experience

Training data Learner

Environment/
Knowledge
Experience

Testing data
Performance
Based on slide by Ray Mooney
Element 41
 Training vs. Test Distribution
• We generally assume that the training and
test examples are independently drawn from
the same overall distribution of data
– We call this “i.i.d” which stands for “independent
and identically distributed”

• If examples are not independent, requires

collective classification
• If test distribution is different, requires
transfer learning
42
 ML in a Nutshell
• Tens of thousands of machine learning
algorithms
– Hundreds new every year

• Every ML algorithm has three components:

– Representation
– Optimization
– Evaluation

43
Slide credit: Pedro Domingos
 Various Function Representations
• Numerical functions
– Linear regression
– Neural networks
– Support vector machines
• Symbolic functions
– Decision trees
– Rules in propositional logic
– Rules in first-order predicate logic
• Instance-based functions
– Nearest-neighbor
– Case-based
• Probabilistic Graphical Models
– Naïve Bayes
– Bayesian networks
– Hidden-Markov Models (HMMs)
– Probabilistic Context Free Grammars (PCFGs)
– Markov networks 44
Slide credit: Ray Mooney
 Various Search/Optimization
Algorithms
• Gradient descent
– Perceptron
– Backpropagation
• Dynamic Programming
– HMM Learning
– PCFG Learning
• Divide and Conquer
– Decision tree induction
– Rule learning
• Evolutionary Computation
– Genetic Algorithms (GAs)
– Genetic Programming (GP)
– Neuro-evolution

45
Slide credit: Ray Mooney
 Evaluation
• Accuracy
• Precision and recall
• Squared error
• Likelihood
• Posterior probability
• Cost / Utility
• Margin
• Entropy
• K-L divergence
• etc.

47
Slide credit: Pedro Domingos
 ML in Practice
• Understand domain, prior knowledge, and goals
• Data integration, selection, cleaning, pre-processing, etc.
Loop • Learn models
• Interpret results
• Consolidate and deploy discovered knowledge

48
Based on a slide by Pedro Domingos
 Lessons Learned about Learning
• Learning can be viewed as using direct or indirect
experience to approximate a chosen target function.

• Function approximation can be viewed as a search

through a space of hypotheses (representations of
functions) for one that best fits a set of training data.

• Different learning methods assume different

hypothesis spaces (representation languages) and/or
employ different search techniques.

49
Slide credit: Ray Mooney
A Brief History of
Machine
Learning

50
 History of Machine Learning
• 1950s
– Samuel’s checker player
– Selfridge’s Pandemonium
• 1960s:
– Neural networks: Perceptron
– Pattern recognition
– Learning in the limit theory
– Minsky and Papert prove limitations of Perceptron
• 1970s:
– Symbolic concept induction
– Winston’s arch learner
– Expert systems and the knowledge acquisition bottleneck
– Quinlan’s ID3
– Michalski’s AQ and soybean diagnosis
– Scientific discovery with BACON
– Mathematical discovery with AM
51
Slide credit: Ray Mooney
 History of Machine Learning (cont.)
• 1980s:
– Advanced decision tree and rule learning
– Explanation-based Learning (EBL)
– Learning and planning and problem solving
– Utility problem
– Analogy
– Cognitive architectures
– Resurgence of neural networks (connectionism, backpropagation)
– Valiant’s PAC Learning Theory
– Focus on experimental methodology
• 1990s
– Data mining
– Adaptive software agents and web applications
– Text learning
– Reinforcement learning (RL)
– Inductive Logic Programming (ILP)
– Ensembles: Bagging, Boosting, and Stacking 52
Slide credit: Ray Mooney
 History of Machine Learning (cont.)
• 2000s
– Support vector machines & kernel methods
– Graphical models
– Statistical relational learning
– Transfer learning
– Sequence labeling
– Collective classification and structured outputs
– Computer Systems Applications (Compilers, Debugging, Graphics, Security)
– E-mail management
– Personalized assistants that learn
– Learning in robotics and vision
• 2010s
– Deep learning systems
– Learning for big data
– Bayesian methods
– Multi-task & lifelong learning
Based on slide by Ray Mooney
53