Machine Learning
Content
MACHINE LEARNING
Presenter:
Junaid Khan Department of Computer Science University of Peshawar Pakistan [email protected]
MACHINE LEARNING
³« the design and development of algorithms and techniques that allow computers to µlearn¶ ´ ³« these programs develop concepts, infer new concepts from existing concepts and revise incorrect concepts´ Getting computers to Programme themselves
2
11/27/2010
MACHINE LEARNING
CONTD«.
Machine learning is a scientific discipline that is concerned with the design and development of algorithms that allow computers to change behavior based on data, such as from sensor data or databases. A major focus of machine learning research is to automatically learn to recognize complex patterns and make intelligent decisions based on data. Hence, machine learning is closely related to fields such as statistics, probability theory
11/27/2010 MACHINE LEARNING
3
EXAMPLE
A machine learning algorithm for face recognition in images would try to define what a face is (round skin-like-colored disk, with dark area where you expect the eyes etc). A machine learning algorithm would not have such coded definition, but will "learnby-examples": you'll show several images of faces and not-faces and a good algorithm will eventually learn and be able to predict whether or not an unseen image is a face.
4
11/27/2010
MACHINE LEARNING
MAGIC???
No, more like gardening Seeds = Algorithms Nutrients = Data Gardener = You Plants = Programs
5
11/27/2010
MACHINE LEARNING
TYPES OF LEARNING\ALGORITHMS
Supervised Unsupervised Semi-supervised Reinforcement Transduction Learning to learn
6
11/27/2010
MACHINE LEARNING
1:SUPERVISED LEARNING
Generates a function that maps inputs to desired outputs. the model defines the effect one set of observations, called inputs, has on another set of observations, called outputs other words, the inputs are assumed to be at the beginning and outputs at the end of the causal chain.
7
11/27/2010
MACHINE LEARNING
2:UN SUPERVISED LEARNING
the observations are assumed to be caused by latent variables, that is, the observations are assumed to be at the end of the causal chain Only have inputs Want to find regularities in the input Density Estimation: finding patterns in the input space ± Clustering: find groupings in the input
8
11/27/2010
MACHINE LEARNING
FIG 1:
9
11/27/2010
MACHINE LEARNING
LATENT VARIABLES;
. The latent variables in the higher levels of abstraction are the causes for both sets of observations and mediate the dependence between inputs and outputs.
10
11/27/2010
MACHINE LEARNING
EXAMPLE:
Decision trees do classification ± Classifies instances into one of a discrete set of possible categories
±
Learned function represented by tree Each node in tree is test on some attribute of an instance Branches represent values of attributes Follow the tree from root to leaves to find the output value. 11/27/2010 MACHINE LEARNING
±
±
±
11
3:SEMI SUPERVISED LEARNING
o
y y
Combines both labeled and unlabeled examples to generate an appropriate function or classifier. Deduce function from training data Training data is a pair of input object and desired output
12
11/27/2010
MACHINE LEARNING
4:REINFORCEMENT LEARNING
Learns how to act given an observation of the world. Policy: generating correct actions to reach the goal Learn from past good policies Example: robot navigating unknown environment in search of a goal ± Some data may be missing ± May be multiple agents in the system
11/27/2010 MACHINE LEARNING
13
5:TRANSDUCTION
y
Tries to predict new outputs based on training inputs, training outputs, and test inputs. Example: Navigation of autonomous car ± Training Data: actions of human drivers in various situations ± Input: data from sensors (like GPS or video) ± Output: angle to turn steering wheel
14
11/27/2010
MACHINE LEARNING
DIFFERNENCE:
TRADITIONAL PROGRAMMING DATA OUTPUT PROGRAM
MACHINE LEARNING DATA OUTPUT PROGRAM
15
11/27/2010
MACHINE LEARNING
HUMAN INTERFACE
Some machine learning systems attempt to eliminate the need for human intuition in data analysis Human intuition cannot, however, be entirely eliminated, since the system's designer must specify how the data is to be represented
16
11/27/2010
MACHINE LEARNING
APPLICATIONS
MACHINE PERCEPTION NLP ROBOTICS CHEMINFORMATICS CREDIT CARD FRAUD SEARCH ENGINES HANDWRITTING RECOGNITION
11/27/2010 MACHINE LEARNING
17
Why Learning Programs?
Differences in learning styles Saves human time Helps make advances in research
18
11/27/2010
MACHINE LEARNING
Why Not Learning Programs?
Difficult to program May require advance knowledge Supervision required in some cases Not suited for all tasks Human interaction cannot be eliminated
19
11/27/2010
MACHINE LEARNING
REFFRENCES :
http://en.wikipedia.org/wiki/Machine_learning http://www.lisa.org/globalizationinsider/2004/09/teac hing_comput.html http://teach-computers.org/ http://hunch.net/?p=290 http://www.machinelearning.net/
20
11/27/2010
MACHINE LEARNING
MESSAGE:
´ A life dedicated chiefly towards the fulfillment of personal desires sooner or later always leads to bitter disappointmentµ ALBERT EINSTIEN
21
11/27/2010
MACHINE LEARNING
Presenter:
Junaid Khan Department of Computer Science University of Peshawar Pakistan [email protected]
MACHINE LEARNING
³« the design and development of algorithms and techniques that allow computers to µlearn¶ ´ ³« these programs develop concepts, infer new concepts from existing concepts and revise incorrect concepts´ Getting computers to Programme themselves
2
11/27/2010
MACHINE LEARNING
CONTD«.
Machine learning is a scientific discipline that is concerned with the design and development of algorithms that allow computers to change behavior based on data, such as from sensor data or databases. A major focus of machine learning research is to automatically learn to recognize complex patterns and make intelligent decisions based on data. Hence, machine learning is closely related to fields such as statistics, probability theory
11/27/2010 MACHINE LEARNING
3
EXAMPLE
A machine learning algorithm for face recognition in images would try to define what a face is (round skin-like-colored disk, with dark area where you expect the eyes etc). A machine learning algorithm would not have such coded definition, but will "learnby-examples": you'll show several images of faces and not-faces and a good algorithm will eventually learn and be able to predict whether or not an unseen image is a face.
4
11/27/2010
MACHINE LEARNING
MAGIC???
No, more like gardening Seeds = Algorithms Nutrients = Data Gardener = You Plants = Programs
5
11/27/2010
MACHINE LEARNING
TYPES OF LEARNING\ALGORITHMS
Supervised Unsupervised Semi-supervised Reinforcement Transduction Learning to learn
6
11/27/2010
MACHINE LEARNING
1:SUPERVISED LEARNING
Generates a function that maps inputs to desired outputs. the model defines the effect one set of observations, called inputs, has on another set of observations, called outputs other words, the inputs are assumed to be at the beginning and outputs at the end of the causal chain.
7
11/27/2010
MACHINE LEARNING
2:UN SUPERVISED LEARNING
the observations are assumed to be caused by latent variables, that is, the observations are assumed to be at the end of the causal chain Only have inputs Want to find regularities in the input Density Estimation: finding patterns in the input space ± Clustering: find groupings in the input
8
11/27/2010
MACHINE LEARNING
FIG 1:
9
11/27/2010
MACHINE LEARNING
LATENT VARIABLES;
. The latent variables in the higher levels of abstraction are the causes for both sets of observations and mediate the dependence between inputs and outputs.
10
11/27/2010
MACHINE LEARNING
EXAMPLE:
Decision trees do classification ± Classifies instances into one of a discrete set of possible categories
±
Learned function represented by tree Each node in tree is test on some attribute of an instance Branches represent values of attributes Follow the tree from root to leaves to find the output value. 11/27/2010 MACHINE LEARNING
±
±
±
11
3:SEMI SUPERVISED LEARNING
o
y y
Combines both labeled and unlabeled examples to generate an appropriate function or classifier. Deduce function from training data Training data is a pair of input object and desired output
12
11/27/2010
MACHINE LEARNING
4:REINFORCEMENT LEARNING
Learns how to act given an observation of the world. Policy: generating correct actions to reach the goal Learn from past good policies Example: robot navigating unknown environment in search of a goal ± Some data may be missing ± May be multiple agents in the system
11/27/2010 MACHINE LEARNING
13
5:TRANSDUCTION
y
Tries to predict new outputs based on training inputs, training outputs, and test inputs. Example: Navigation of autonomous car ± Training Data: actions of human drivers in various situations ± Input: data from sensors (like GPS or video) ± Output: angle to turn steering wheel
14
11/27/2010
MACHINE LEARNING
DIFFERNENCE:
TRADITIONAL PROGRAMMING DATA OUTPUT PROGRAM
MACHINE LEARNING DATA OUTPUT PROGRAM
15
11/27/2010
MACHINE LEARNING
HUMAN INTERFACE
Some machine learning systems attempt to eliminate the need for human intuition in data analysis Human intuition cannot, however, be entirely eliminated, since the system's designer must specify how the data is to be represented
16
11/27/2010
MACHINE LEARNING
APPLICATIONS
MACHINE PERCEPTION NLP ROBOTICS CHEMINFORMATICS CREDIT CARD FRAUD SEARCH ENGINES HANDWRITTING RECOGNITION
11/27/2010 MACHINE LEARNING
17
Why Learning Programs?
Differences in learning styles Saves human time Helps make advances in research
18
11/27/2010
MACHINE LEARNING
Why Not Learning Programs?
Difficult to program May require advance knowledge Supervision required in some cases Not suited for all tasks Human interaction cannot be eliminated
19
11/27/2010
MACHINE LEARNING
REFFRENCES :
http://en.wikipedia.org/wiki/Machine_learning http://www.lisa.org/globalizationinsider/2004/09/teac hing_comput.html http://teach-computers.org/ http://hunch.net/?p=290 http://www.machinelearning.net/
20
11/27/2010
MACHINE LEARNING
MESSAGE:
´ A life dedicated chiefly towards the fulfillment of personal desires sooner or later always leads to bitter disappointmentµ ALBERT EINSTIEN
21
11/27/2010
MACHINE LEARNING
