SlideShare a Scribd company logo
1 of 117
Download to read offline
Mathematics for Artiļ¬cial Intelligence
Vector Spaces
Andres Mendez-Vazquez
March 14, 2020
1 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
2 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
3 / 59
Introduction
What is this class about?
It is clear that the use of mathematics is essential for the data mining and
machine learning ļ¬elds.
Therefore...
The understanding of Mathematical Modeling is part of the deal...
If you want to be
A Good Data Scientist!!!
4 / 59
Introduction
What is this class about?
It is clear that the use of mathematics is essential for the data mining and
machine learning ļ¬elds.
Therefore...
The understanding of Mathematical Modeling is part of the deal...
If you want to be
A Good Data Scientist!!!
4 / 59
Introduction
What is this class about?
It is clear that the use of mathematics is essential for the data mining and
machine learning ļ¬elds.
Therefore...
The understanding of Mathematical Modeling is part of the deal...
If you want to be
A Good Data Scientist!!!
4 / 59
Example
Imagine
A web surfer moves from a web page to another web page...
Question: How do you model this?
You can use a graph!!!
5 / 59
Example
Imagine
A web surfer moves from a web page to another web page...
Question: How do you model this?
You can use a graph!!!
1
8
2
9
10
4
7
14
11
12
6
3
13
5
5 / 59
Now
Add Some Probabilities
1
8
2
9
10
4
7
14
11
12
6
3
13
5
6 / 59
Thus
We can build a matrix
M =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
P11 P12 Ā· Ā· Ā· P1N
P21 P22 Ā· Ā· Ā· P2N
...
...
...
...
PN1 PN2 Ā· Ā· Ā· PNN
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
(1)
Thus, it is possible to obtain certain information by looking at the
eigenvector and eigenvalues
These vectors vĪ»s and values Ī» s have the property that
MvĪ» = Ī»vĪ» (2)
7 / 59
Thus
We can build a matrix
M =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
P11 P12 Ā· Ā· Ā· P1N
P21 P22 Ā· Ā· Ā· P2N
...
...
...
...
PN1 PN2 Ā· Ā· Ā· PNN
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
(1)
Thus, it is possible to obtain certain information by looking at the
eigenvector and eigenvalues
These vectors vĪ»s and values Ī» s have the property that
MvĪ» = Ī»vĪ» (2)
7 / 59
This is the Basis of Page Rank in Google
For example
Look at this example...
8 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
9 / 59
About 4000 years ago
Babylonians knew how to solve the following kind of systems
ax + by = c
dx + ey = f
As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow
It is only after the death of Plato and Aristotle, that the Chinese (Nine
Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3
system.
By working an ā€œelimination methodā€
Similar to the one devised by Gauss 2000 years later for general systems.
10 / 59
About 4000 years ago
Babylonians knew how to solve the following kind of systems
ax + by = c
dx + ey = f
As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow
It is only after the death of Plato and Aristotle, that the Chinese (Nine
Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3
system.
By working an ā€œelimination methodā€
Similar to the one devised by Gauss 2000 years later for general systems.
10 / 59
About 4000 years ago
Babylonians knew how to solve the following kind of systems
ax + by = c
dx + ey = f
As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow
It is only after the death of Plato and Aristotle, that the Chinese (Nine
Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3
system.
By working an ā€œelimination methodā€
Similar to the one devised by Gauss 2000 years later for general systems.
10 / 59
Not only that
The Matrix
Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations
in his book ā€œDisquisitions.ā€
The Final Deļ¬nition of Matrix
It was introduced by Cayley in two papers in 1850 and 1858 respectively,
which allowed him to prove the important Cayley-Hamilton Theorem.
There is quite a lot
Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007).
11 / 59
Not only that
The Matrix
Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations
in his book ā€œDisquisitions.ā€
The Final Deļ¬nition of Matrix
It was introduced by Cayley in two papers in 1850 and 1858 respectively,
which allowed him to prove the important Cayley-Hamilton Theorem.
There is quite a lot
Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007).
11 / 59
Not only that
The Matrix
Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations
in his book ā€œDisquisitions.ā€
The Final Deļ¬nition of Matrix
It was introduced by Cayley in two papers in 1850 and 1858 respectively,
which allowed him to prove the important Cayley-Hamilton Theorem.
There is quite a lot
Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007).
11 / 59
Matrix can help to represent many things
They are important for many calculations as
a11x1 + a12x2 + ... + a1nxn =b1,
a21x1 + a22x2 + ... + a2nxn =b2,
Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā·
am1x1 + am2x2 + ... + amnxn =b2.
It is clear
We would like to collect those linear equations in a compact structure that
allows for simpler manipulation.
12 / 59
Matrix can help to represent many things
They are important for many calculations as
a11x1 + a12x2 + ... + a1nxn =b1,
a21x1 + a22x2 + ... + a2nxn =b2,
Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā·
am1x1 + am2x2 + ... + amnxn =b2.
It is clear
We would like to collect those linear equations in a compact structure that
allows for simpler manipulation.
12 / 59
Therefore, we have
For example
x =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
x1
x2
...
xn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
, b =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
b1
b2
...
bn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
and A =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
a11 a12 Ā· Ā· Ā· a1n
a21 a22 Ā· Ā· Ā· a2n
...
...
...
...
am1 am2 Ā· Ā· Ā· amn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
Using a little of notation
Ax = b
13 / 59
Therefore, we have
For example
x =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
x1
x2
...
xn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
, b =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
b1
b2
...
bn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
and A =
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
a11 a12 Ā· Ā· Ā· a1n
a21 a22 Ā· Ā· Ā· a2n
...
...
...
...
am1 am2 Ā· Ā· Ā· amn
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
Using a little of notation
Ax = b
13 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
14 / 59
Introduction
As always, we star with a simple fact
Everything is an element in a set.
For example
The set of Real Numbers R.
The set of n-tuples in Rn.
The set of Complex Number C.
15 / 59
Introduction
As always, we star with a simple fact
Everything is an element in a set.
For example
The set of Real Numbers R.
The set of n-tuples in Rn.
The set of Complex Number C.
15 / 59
Introduction
As always, we star with a simple fact
Everything is an element in a set.
For example
The set of Real Numbers R.
The set of n-tuples in Rn.
The set of Complex Number C.
15 / 59
Introduction
As always, we star with a simple fact
Everything is an element in a set.
For example
The set of Real Numbers R.
The set of n-tuples in Rn.
The set of Complex Number C.
15 / 59
Deļ¬nition
We shall say that K is a ļ¬eld if it satisļ¬es the following conditions for
the addition
Property Formalism
Addition is Commutative x + y = y + x for all x, y āˆˆ K
Addition is associative x + (y + z) = (x + y) + z for all x, y, z āˆˆ K
Existence of 0 x + 0 = x, for every x āˆˆ K
Existence of the inverse āˆ€x there is āˆƒ āˆ’ x =ā‡’ x + (āˆ’x) = 0
16 / 59
Furthermore
We have the following properties for the product
Property Formalism
Product is Commutative xy = yx for all x, y āˆˆ K
Product is associative x (yz) = (xy) z for all x, y, z āˆˆ K
Existence of 1 1x = x1 = x, for every x āˆˆ K.
Existence of the inverse xāˆ’1
or 1
x
in K such that xxāˆ’1
= 1.
Multiplication is Distributive over addition x (y + z) = xy + xz, for all x, y, z āˆˆ K
17 / 59
Therefore
Examples
1 For example the reals R and the C.
2 In addition, we have the rationals Q too.
The elements of the ļ¬eld will be also called numbers
Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K.
18 / 59
Therefore
Examples
1 For example the reals R and the C.
2 In addition, we have the rationals Q too.
The elements of the ļ¬eld will be also called numbers
Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K.
18 / 59
Therefore
Examples
1 For example the reals R and the C.
2 In addition, we have the rationals Q too.
The elements of the ļ¬eld will be also called numbers
Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K.
18 / 59
Then, we get a crazy moment
How do we relate these numbers to obtain certain properties
We have then the vector and matrix structures for this...
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
a11 Ā· Ā· Ā· Ā· Ā· Ā· a1n
a21 Ā· Ā· Ā· Ā· Ā· Ā· a2n
...
...
...
...
an1 Ā· Ā· Ā· Ā· Ā· Ā· ann
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
and
ļ£«
ļ£¬
ļ£¬
ļ£¬
ļ£¬
ļ£­
a11
a21
...
an1
ļ£¶
ļ£·
ļ£·
ļ£·
ļ£·
ļ£ø
19 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
20 / 59
Vector Space V
Deļ¬nition
A vector space V over the ļ¬eld K is a set of objects which can be added
and multiplied by elements of K.
Where
The sum of two elements of V is again an element of V .
The product of an element of V by an element of K is an element of
V .
21 / 59
Vector Space V
Deļ¬nition
A vector space V over the ļ¬eld K is a set of objects which can be added
and multiplied by elements of K.
Where
The sum of two elements of V is again an element of V .
The product of an element of V by an element of K is an element of
V .
21 / 59
Vector Space V
Deļ¬nition
A vector space V over the ļ¬eld K is a set of objects which can be added
and multiplied by elements of K.
Where
The sum of two elements of V is again an element of V .
The product of an element of V by an element of K is an element of
V .
21 / 59
Vector Space V
Deļ¬nition
A vector space V over the ļ¬eld K is a set of objects which can be added
and multiplied by elements of K.
Where
The sum of two elements of V is again an element of V .
The product of an element of V by an element of K is an element of
V .
21 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Properties
We have then
1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w).
2 There is an element of V , denoted by O, such that
O + u = u + O = u for all elements u of V .
3 Given an element u of V , there exists an element āˆ’u in V such that
u + (āˆ’u) = O.
4 For all elements u, v of V , we have u + v = v + u.
5 For all elements u of V , we have 1 Ā· u = u.
6 If c is a number, then c (u + v) = cu + cv.
7 if a, b are two numbers, then (ab) v = a (bv).
8 If a, b are two numbers, then (a + b) v = av + bv.
22 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
23 / 59
Notation
First, u + (āˆ’v)
As u āˆ’ v.
For O
We will write sometimes 0.
The elements in the ļ¬eld K
They can receive the name of number or scalar.
24 / 59
Notation
First, u + (āˆ’v)
As u āˆ’ v.
For O
We will write sometimes 0.
The elements in the ļ¬eld K
They can receive the name of number or scalar.
24 / 59
Notation
First, u + (āˆ’v)
As u āˆ’ v.
For O
We will write sometimes 0.
The elements in the ļ¬eld K
They can receive the name of number or scalar.
24 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
25 / 59
Many Times
We have this kind of data sets (House Prices)
26 / 59
Therefore
We can represent these relations as vectors
Squared Feet
Price
=
2104
400
,
1800
460
,
1600
300
, ...
Thus, we can start using
All the tools that Linear Algebra can provide!!!
27 / 59
Therefore
We can represent these relations as vectors
Squared Feet
Price
=
2104
400
,
1800
460
,
1600
300
, ...
Thus, we can start using
All the tools that Linear Algebra can provide!!!
27 / 59
Thus
We can adjust a line/hyper-plane to be able to forecast prices
28 / 59
Thus, Our Objective
To ļ¬nd such hyper-plane
To do forecasting on the prices of a house given its surface size!!!
Here, where ā€œLearningā€ comes around
Basically, the process deļ¬ned in Machine Learning!!!
29 / 59
Thus, Our Objective
To ļ¬nd such hyper-plane
To do forecasting on the prices of a house given its surface size!!!
Here, where ā€œLearningā€ comes around
Basically, the process deļ¬ned in Machine Learning!!!
29 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
30 / 59
Sub-spaces
Deļ¬nition
Let V a vector space and W āŠ† V , thus W is a subspace if:
1 If v, w āˆˆ W, then v + w āˆˆ W.
2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W.
3 The element 0 āˆˆ V is also an element of W.
31 / 59
Sub-spaces
Deļ¬nition
Let V a vector space and W āŠ† V , thus W is a subspace if:
1 If v, w āˆˆ W, then v + w āˆˆ W.
2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W.
3 The element 0 āˆˆ V is also an element of W.
31 / 59
Sub-spaces
Deļ¬nition
Let V a vector space and W āŠ† V , thus W is a subspace if:
1 If v, w āˆˆ W, then v + w āˆˆ W.
2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W.
3 The element 0 āˆˆ V is also an element of W.
31 / 59
Sub-spaces
Deļ¬nition
Let V a vector space and W āŠ† V , thus W is a subspace if:
1 If v, w āˆˆ W, then v + w āˆˆ W.
2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W.
3 The element 0 āˆˆ V is also an element of W.
31 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
32 / 59
Some ways of recognizing Sub-spaces
Theorem
A non-empty subset W of V is a subspace of V if and only if for each pair
of vectors v, w āˆˆ W and each scalar c āˆˆ K the vector cv + w āˆˆ W.
33 / 59
Example
For R2
0
1 2 3
1
2
4
0
34 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
35 / 59
Linear Combinations
Deļ¬nition
Let V an arbitrary vector space, and let v1, v2, ..., vn āˆˆ V and
x1, x2, ..., xn āˆˆ K. Then, an expression like
x1v1 + x2v2 + ... + xnvn (3)
is called a linear combination of v1, v2, ..., vn.
36 / 59
Classic Examples
Endmember Representation in Hyperspectral Images
Look at the board
Geometric Representation of addition of forces in Physics
Look at the board!!
37 / 59
Classic Examples
Endmember Representation in Hyperspectral Images
Look at the board
Geometric Representation of addition of forces in Physics
Look at the board!!
37 / 59
Properties and Deļ¬nitions
Theorem
Let V be a vector space over the ļ¬eld K. The intersection of any
collection of sub-spaces of V is a subspace of V .
Deļ¬nition
Let S be a set of vectors in a vector space V .
The sub-space spanned by S is deļ¬ned as the intersection W of all
sub-spaces of V which contains S.
When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall
simply call W the sub-space spanned by the vectors v1, v2, . . . , vn.
38 / 59
Properties and Deļ¬nitions
Theorem
Let V be a vector space over the ļ¬eld K. The intersection of any
collection of sub-spaces of V is a subspace of V .
Deļ¬nition
Let S be a set of vectors in a vector space V .
The sub-space spanned by S is deļ¬ned as the intersection W of all
sub-spaces of V which contains S.
When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall
simply call W the sub-space spanned by the vectors v1, v2, . . . , vn.
38 / 59
Properties and Deļ¬nitions
Theorem
Let V be a vector space over the ļ¬eld K. The intersection of any
collection of sub-spaces of V is a subspace of V .
Deļ¬nition
Let S be a set of vectors in a vector space V .
The sub-space spanned by S is deļ¬ned as the intersection W of all
sub-spaces of V which contains S.
When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall
simply call W the sub-space spanned by the vectors v1, v2, . . . , vn.
38 / 59
Properties and Deļ¬nitions
Theorem
Let V be a vector space over the ļ¬eld K. The intersection of any
collection of sub-spaces of V is a subspace of V .
Deļ¬nition
Let S be a set of vectors in a vector space V .
The sub-space spanned by S is deļ¬ned as the intersection W of all
sub-spaces of V which contains S.
When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall
simply call W the sub-space spanned by the vectors v1, v2, . . . , vn.
38 / 59
We get the following Theorem
Theorem
The subspace spanned by S = āˆ… is the set of all linear combinations of
vectors in S.
39 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
40 / 59
Linear Independence
Deļ¬nition
Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We
have that v1, v2, ..., vn are linearly dependent over K if there are elements
a1, a2, ..., an āˆˆ K not all equal to 0 such that
a1v1 + a2v2 + ... + anvn = O
Thus
Therefore, if there are not such numbers, then we say that v1, v2, ..., vn
are linearly independent.
We have the following
Example!!!
41 / 59
Linear Independence
Deļ¬nition
Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We
have that v1, v2, ..., vn are linearly dependent over K if there are elements
a1, a2, ..., an āˆˆ K not all equal to 0 such that
a1v1 + a2v2 + ... + anvn = O
Thus
Therefore, if there are not such numbers, then we say that v1, v2, ..., vn
are linearly independent.
We have the following
Example!!!
41 / 59
Linear Independence
Deļ¬nition
Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We
have that v1, v2, ..., vn are linearly dependent over K if there are elements
a1, a2, ..., an āˆˆ K not all equal to 0 such that
a1v1 + a2v2 + ... + anvn = O
Thus
Therefore, if there are not such numbers, then we say that v1, v2, ..., vn
are linearly independent.
We have the following
Example!!!
41 / 59
Basis
Deļ¬nition
If elements v1, v2, ..., vn generate V and in addition are linearly
independent, then {v1, v2, ..., vn} is called a basis of V . In other words
the elements v1, v2, ..., vn form a basis of V .
Examples
The Classic Ones!!!
42 / 59
Basis
Deļ¬nition
If elements v1, v2, ..., vn generate V and in addition are linearly
independent, then {v1, v2, ..., vn} is called a basis of V . In other words
the elements v1, v2, ..., vn form a basis of V .
Examples
The Classic Ones!!!
42 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
43 / 59
Coordinates
Theorem
Let V be a vector space. Let v1, v2, ..., vn be linearly independent elements
of V. Let x1, . . . , xn and y1, . . . , yn be numbers. Suppose that we have
x1v1 + x2v2 + Ā· Ā· Ā· + xnvn = y1v1 + y2v2 + Ā· Ā· Ā· + ynvn (4)
Then, xi = yi for all i = 1, . . . , n.
44 / 59
Coordinates
Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V
For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn.
Thus, this n-tuple is uniquely determined by v
We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the
basis.
The nāˆ’tuple X = (x1, x2, . . . , xn)
It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} .
45 / 59
Coordinates
Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V
For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn.
Thus, this n-tuple is uniquely determined by v
We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the
basis.
The nāˆ’tuple X = (x1, x2, . . . , xn)
It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} .
45 / 59
Coordinates
Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V
For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn.
Thus, this n-tuple is uniquely determined by v
We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the
basis.
The nāˆ’tuple X = (x1, x2, . . . , xn)
It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} .
45 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
46 / 59
Properties of a Basis
Theorem - (Limit in the size of the basis)
Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let
w1, w2, ..., wn be elements of V , and assume that n > m. Then
w1, w2, ..., wn are linearly dependent.
Examples
Matrix Space
Canonical Space vectors
etc
47 / 59
Properties of a Basis
Theorem - (Limit in the size of the basis)
Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let
w1, w2, ..., wn be elements of V , and assume that n > m. Then
w1, w2, ..., wn are linearly dependent.
Examples
Matrix Space
Canonical Space vectors
etc
47 / 59
Properties of a Basis
Theorem - (Limit in the size of the basis)
Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let
w1, w2, ..., wn be elements of V , and assume that n > m. Then
w1, w2, ..., wn are linearly dependent.
Examples
Matrix Space
Canonical Space vectors
etc
47 / 59
Some Basic Deļ¬nitions
We will deļ¬ne the dimension of a vector space V over K
As the number of elements in the basis.
Denoted by dimK V , or simply dim V
Therefore
A vector space with a basis consisting of a ļ¬nite number of elements, or
the zero vector space, is called a ļ¬nite dimensional.
Now
Is this number unique?
48 / 59
Some Basic Deļ¬nitions
We will deļ¬ne the dimension of a vector space V over K
As the number of elements in the basis.
Denoted by dimK V , or simply dim V
Therefore
A vector space with a basis consisting of a ļ¬nite number of elements, or
the zero vector space, is called a ļ¬nite dimensional.
Now
Is this number unique?
48 / 59
Some Basic Deļ¬nitions
We will deļ¬ne the dimension of a vector space V over K
As the number of elements in the basis.
Denoted by dimK V , or simply dim V
Therefore
A vector space with a basis consisting of a ļ¬nite number of elements, or
the zero vector space, is called a ļ¬nite dimensional.
Now
Is this number unique?
48 / 59
Maximal Set of Linearly Independent Elements
Theorem
Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly
independent elements of V . Then, {v1, v2, ..., vn} is a basis of V .
Theorem
Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly
independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V .
49 / 59
Maximal Set of Linearly Independent Elements
Theorem
Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly
independent elements of V . Then, {v1, v2, ..., vn} is a basis of V .
Theorem
Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly
independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V .
49 / 59
Maximal Set of Linearly Independent Elements
Theorem
Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly
independent elements of V . Then, {v1, v2, ..., vn} is a basis of V .
Theorem
Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly
independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V .
49 / 59
Equality between Basis
Corollary
Let V be a vector space and let W be a subspace. If dim W = dim V
then V = W.
Proof
At the Board...
Corollary
Let V be a vector space of dimension n. Let r be a positive integer with
r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then
one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a
basis of V .
Proof
At the Board...
50 / 59
Equality between Basis
Corollary
Let V be a vector space and let W be a subspace. If dim W = dim V
then V = W.
Proof
At the Board...
Corollary
Let V be a vector space of dimension n. Let r be a positive integer with
r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then
one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a
basis of V .
Proof
At the Board...
50 / 59
Equality between Basis
Corollary
Let V be a vector space and let W be a subspace. If dim W = dim V
then V = W.
Proof
At the Board...
Corollary
Let V be a vector space of dimension n. Let r be a positive integer with
r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then
one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a
basis of V .
Proof
At the Board...
50 / 59
Equality between Basis
Corollary
Let V be a vector space and let W be a subspace. If dim W = dim V
then V = W.
Proof
At the Board...
Corollary
Let V be a vector space of dimension n. Let r be a positive integer with
r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then
one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a
basis of V .
Proof
At the Board...
50 / 59
Finally
Theorem
Let V be a vector space having a basis consisting of n elements. Let W be
a subspace which does not consist of O alone. Then W has a basis, and
the dimension of W is ā‰¤ n.
Proof
At the Board...
51 / 59
Finally
Theorem
Let V be a vector space having a basis consisting of n elements. Let W be
a subspace which does not consist of O alone. Then W has a basis, and
the dimension of W is ā‰¤ n.
Proof
At the Board...
51 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
52 / 59
Feature Vector
Deļ¬nition
A feature vector is a n-dimensional vector of numerical features that
represent an object.
Why is this important?
This allows to use linear algebra to represent basic classiļ¬cation algorithms
because
The tuples {(x, y) |x āˆˆ Kn, y āˆˆ K} can be easily used to design
speciļ¬c algorithms.
53 / 59
Feature Vector
Deļ¬nition
A feature vector is a n-dimensional vector of numerical features that
represent an object.
Why is this important?
This allows to use linear algebra to represent basic classiļ¬cation algorithms
because
The tuples {(x, y) |x āˆˆ Kn, y āˆˆ K} can be easily used to design
speciļ¬c algorithms.
53 / 59
Outline
1 Why Liner Algebra
Why and What?
A Little Bit of History
2 The Beginning
Fields
3 Vector Space
Introduction
Some Notes in Notation
Use of Linear Algebra in Regression...
Sub-spaces and Linear Combinations
Recognizing Sub-spaces
Combinations
4 Basis and Dimensions
Basis
Coordinates
Basis and Dimensions
5 Application in Machine Learning
Feature Vector
Least Squared Error
54 / 59
Least Squared Error
We need to ļ¬t a series of points against a certain function
We want
The general problem is given a set of functions f1, f2, ..., fK ļ¬nd values of
coeļ¬ƒcients a1, a2, ..., ak such that the linear combination:
y = a1f1 (x) + Ā· Ā· Ā· + aKfK (x) (5)
55 / 59
Least Squared Error
We need to ļ¬t a series of points against a certain function
We want
The general problem is given a set of functions f1, f2, ..., fK ļ¬nd values of
coeļ¬ƒcients a1, a2, ..., ak such that the linear combination:
y = a1f1 (x) + Ā· Ā· Ā· + aKfK (x) (5)
55 / 59
Thus
We have that given the datasets {(x1, y1) , ..., (xN , yN )}
x =
1
N
N
i=1
xi. (6)
Thus, we have the following problem
A Possible High Variance on the Data itself
Variance
Ļƒ2
x =
1
N
N
i=1
(xi āˆ’ x) (7)
56 / 59
Thus
We have that given the datasets {(x1, y1) , ..., (xN , yN )}
x =
1
N
N
i=1
xi. (6)
Thus, we have the following problem
A Possible High Variance on the Data itself
Variance
Ļƒ2
x =
1
N
N
i=1
(xi āˆ’ x) (7)
56 / 59
Thus
We have that given the datasets {(x1, y1) , ..., (xN , yN )}
x =
1
N
N
i=1
xi. (6)
Thus, we have the following problem
A Possible High Variance on the Data itself
Variance
Ļƒ2
x =
1
N
N
i=1
(xi āˆ’ x) (7)
56 / 59
Now
Assume
A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0.
We get a series of errors given the following observations
{(x1, y1) , ..., (xN , yN )}
{y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} .
Then, the mean should be really small (If it is a good ļ¬t)
Ļƒ2
yāˆ’(ax+b) =
1
N
N
i=1
(yi āˆ’ (axi + b))2
(8)
57 / 59
Now
Assume
A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0.
We get a series of errors given the following observations
{(x1, y1) , ..., (xN , yN )}
{y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} .
Then, the mean should be really small (If it is a good ļ¬t)
Ļƒ2
yāˆ’(ax+b) =
1
N
N
i=1
(yi āˆ’ (axi + b))2
(8)
57 / 59
Now
Assume
A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0.
We get a series of errors given the following observations
{(x1, y1) , ..., (xN , yN )}
{y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} .
Then, the mean should be really small (If it is a good ļ¬t)
Ļƒ2
yāˆ’(ax+b) =
1
N
N
i=1
(yi āˆ’ (axi + b))2
(8)
57 / 59
Thus
We can deļ¬ne the following error Ei (a, b) = y āˆ’ (ax + b)
E (a, b) =
N
i=1
Ei (a, b) =
N
i=1
(yi āˆ’ (axi + b)) (9)
We want to minimize the previous equation
āˆ‚E
āˆ‚a
= 0,
āˆ‚E
āˆ‚b
= 0.
58 / 59
Thus
We can deļ¬ne the following error Ei (a, b) = y āˆ’ (ax + b)
E (a, b) =
N
i=1
Ei (a, b) =
N
i=1
(yi āˆ’ (axi + b)) (9)
We want to minimize the previous equation
āˆ‚E
āˆ‚a
= 0,
āˆ‚E
āˆ‚b
= 0.
58 / 59
Finally
Look at the Board
We need to obtain the necessary equations.
59 / 59

More Related Content

What's hot

Polya recurrence
Polya recurrencePolya recurrence
Polya recurrenceBrian Burns
Ā 
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeks
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeksBeginning direct3d gameprogrammingmath04_calculus_20160324_jintaeks
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeksJinTaek Seo
Ā 
Lecture 8 nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6
Lecture 8   nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6Lecture 8   nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6
Lecture 8 nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6njit-ronbrown
Ā 
Lecture 9 dim & rank - 4-5 & 4-6
Lecture 9   dim & rank -  4-5 & 4-6Lecture 9   dim & rank -  4-5 & 4-6
Lecture 9 dim & rank - 4-5 & 4-6njit-ronbrown
Ā 
Linear transformations and matrices
Linear transformations and matricesLinear transformations and matrices
Linear transformations and matricesEasyStudy3
Ā 
My presentation all shortestpath
My presentation all shortestpathMy presentation all shortestpath
My presentation all shortestpathCarlostheran
Ā 
Linear transformation and application
Linear transformation and applicationLinear transformation and application
Linear transformation and applicationshreyansp
Ā 
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeks
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeksBeginning direct3d gameprogrammingmath03_vectors_20160328_jintaeks
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeksJinTaek Seo
Ā 
Metric space
Metric spaceMetric space
Metric spaceNaliniSPatil
Ā 
Null space, Rank and nullity theorem
Null space, Rank and nullity theoremNull space, Rank and nullity theorem
Null space, Rank and nullity theoremRonak Machhi
Ā 
08 decrease and conquer spring 15
08 decrease and conquer spring 1508 decrease and conquer spring 15
08 decrease and conquer spring 15Hira Gul
Ā 
Numerical Methods 1
Numerical Methods 1Numerical Methods 1
Numerical Methods 1Dr. Nirav Vyas
Ā 
THE CALCULUS INTEGRAL (Beta Version 2009)
THE CALCULUS INTEGRAL (Beta Version 2009)THE CALCULUS INTEGRAL (Beta Version 2009)
THE CALCULUS INTEGRAL (Beta Version 2009)briansthomson
Ā 
Lecture 7 determinants cramers spaces - section 3-2 3-3 and 4-1
Lecture 7   determinants cramers spaces - section 3-2 3-3 and 4-1Lecture 7   determinants cramers spaces - section 3-2 3-3 and 4-1
Lecture 7 determinants cramers spaces - section 3-2 3-3 and 4-1njit-ronbrown
Ā 
FPDE presentation
FPDE presentationFPDE presentation
FPDE presentationDivyansh Verma
Ā 

What's hot (20)

Vector spaces
Vector spacesVector spaces
Vector spaces
Ā 
Polya recurrence
Polya recurrencePolya recurrence
Polya recurrence
Ā 
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeks
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeksBeginning direct3d gameprogrammingmath04_calculus_20160324_jintaeks
Beginning direct3d gameprogrammingmath04_calculus_20160324_jintaeks
Ā 
Lecture 8 nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6
Lecture 8   nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6Lecture 8   nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6
Lecture 8 nul col bases dim & rank - section 4-2, 4-3, 4-5 & 4-6
Ā 
Lecture 9 dim & rank - 4-5 & 4-6
Lecture 9   dim & rank -  4-5 & 4-6Lecture 9   dim & rank -  4-5 & 4-6
Lecture 9 dim & rank - 4-5 & 4-6
Ā 
Linear transformations and matrices
Linear transformations and matricesLinear transformations and matrices
Linear transformations and matrices
Ā 
My presentation all shortestpath
My presentation all shortestpathMy presentation all shortestpath
My presentation all shortestpath
Ā 
Linear transformation and application
Linear transformation and applicationLinear transformation and application
Linear transformation and application
Ā 
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeks
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeksBeginning direct3d gameprogrammingmath03_vectors_20160328_jintaeks
Beginning direct3d gameprogrammingmath03_vectors_20160328_jintaeks
Ā 
Divide and Conquer
Divide and ConquerDivide and Conquer
Divide and Conquer
Ā 
2 vectors notes
2 vectors notes2 vectors notes
2 vectors notes
Ā 
Metric space
Metric spaceMetric space
Metric space
Ā 
Null space, Rank and nullity theorem
Null space, Rank and nullity theoremNull space, Rank and nullity theorem
Null space, Rank and nullity theorem
Ā 
08 decrease and conquer spring 15
08 decrease and conquer spring 1508 decrease and conquer spring 15
08 decrease and conquer spring 15
Ā 
Numerical Methods 1
Numerical Methods 1Numerical Methods 1
Numerical Methods 1
Ā 
Inner product spaces
Inner product spacesInner product spaces
Inner product spaces
Ā 
THE CALCULUS INTEGRAL (Beta Version 2009)
THE CALCULUS INTEGRAL (Beta Version 2009)THE CALCULUS INTEGRAL (Beta Version 2009)
THE CALCULUS INTEGRAL (Beta Version 2009)
Ā 
Divide and conquer
Divide and conquerDivide and conquer
Divide and conquer
Ā 
Lecture 7 determinants cramers spaces - section 3-2 3-3 and 4-1
Lecture 7   determinants cramers spaces - section 3-2 3-3 and 4-1Lecture 7   determinants cramers spaces - section 3-2 3-3 and 4-1
Lecture 7 determinants cramers spaces - section 3-2 3-3 and 4-1
Ā 
FPDE presentation
FPDE presentationFPDE presentation
FPDE presentation
Ā 

Similar to 01.01 vector spaces

matrices-1.pdf
matrices-1.pdfmatrices-1.pdf
matrices-1.pdfWunnamAlabani
Ā 
Setting linear algebra problems
Setting linear algebra problemsSetting linear algebra problems
Setting linear algebra problemsJB Online
Ā 
Why recursion is impotant_new_my.docx
Why recursion is impotant_new_my.docxWhy recursion is impotant_new_my.docx
Why recursion is impotant_new_my.docxMiracule D Gavor
Ā 
integral calculus.pdf
integral calculus.pdfintegral calculus.pdf
integral calculus.pdfAudreyMendoza6
Ā 
1560 mathematics for economists
1560 mathematics for economists1560 mathematics for economists
1560 mathematics for economistsDr Fereidoun Dejahang
Ā 
elliptic-curves-modern
elliptic-curves-modernelliptic-curves-modern
elliptic-curves-modernEric Seifert
Ā 
Study on a class of recursive functions
Study on a class of recursive functionsStudy on a class of recursive functions
Study on a class of recursive functionsLuca Polidori
Ā 
1 1 number theory
1 1 number theory1 1 number theory
1 1 number theorysmillertx281
Ā 
Mathematics Ist year pdf.pdf
Mathematics Ist year pdf.pdfMathematics Ist year pdf.pdf
Mathematics Ist year pdf.pdfproacademyhub
Ā 
MMath Paper, Canlin Zhang
MMath Paper, Canlin ZhangMMath Paper, Canlin Zhang
MMath Paper, Canlin Zhangcanlin zhang
Ā 
Maths 11
Maths 11Maths 11
Maths 11Mehtab Rai
Ā 
Matlab intro notes
Matlab intro notesMatlab intro notes
Matlab intro notespawanss
Ā 
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...ijscmcj
Ā 
Business mathematics presentation
Business mathematics presentationBusiness mathematics presentation
Business mathematics presentationSourov Shaha Suvo
Ā 
Proof in Mathematics
Proof in MathematicsProof in Mathematics
Proof in Mathematicsmscartersmaths
Ā 
signalsandsystemsnotes.pdf
signalsandsystemsnotes.pdfsignalsandsystemsnotes.pdf
signalsandsystemsnotes.pdfSubbuSiva1
Ā 

Similar to 01.01 vector spaces (20)

matrices-1.pdf
matrices-1.pdfmatrices-1.pdf
matrices-1.pdf
Ā 
Setting linear algebra problems
Setting linear algebra problemsSetting linear algebra problems
Setting linear algebra problems
Ā 
Why recursion is impotant_new_my.docx
Why recursion is impotant_new_my.docxWhy recursion is impotant_new_my.docx
Why recursion is impotant_new_my.docx
Ā 
integral calculus.pdf
integral calculus.pdfintegral calculus.pdf
integral calculus.pdf
Ā 
1560 mathematics for economists
1560 mathematics for economists1560 mathematics for economists
1560 mathematics for economists
Ā 
Problemas de Smale
Problemas de SmaleProblemas de Smale
Problemas de Smale
Ā 
Free221
Free221Free221
Free221
Ā 
10.1.1.96.9176
10.1.1.96.917610.1.1.96.9176
10.1.1.96.9176
Ā 
elliptic-curves-modern
elliptic-curves-modernelliptic-curves-modern
elliptic-curves-modern
Ā 
Study on a class of recursive functions
Study on a class of recursive functionsStudy on a class of recursive functions
Study on a class of recursive functions
Ā 
1 1 number theory
1 1 number theory1 1 number theory
1 1 number theory
Ā 
Mathematics Ist year pdf.pdf
Mathematics Ist year pdf.pdfMathematics Ist year pdf.pdf
Mathematics Ist year pdf.pdf
Ā 
MMath Paper, Canlin Zhang
MMath Paper, Canlin ZhangMMath Paper, Canlin Zhang
MMath Paper, Canlin Zhang
Ā 
Maths 11
Maths 11Maths 11
Maths 11
Ā 
Matlab intro notes
Matlab intro notesMatlab intro notes
Matlab intro notes
Ā 
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...
A PROBABILISTIC ALGORITHM OF COMPUTING THE POLYNOMIAL GREATEST COMMON DIVISOR...
Ā 
Business mathematics presentation
Business mathematics presentationBusiness mathematics presentation
Business mathematics presentation
Ā 
02 Notes Divide and Conquer
02 Notes Divide and Conquer02 Notes Divide and Conquer
02 Notes Divide and Conquer
Ā 
Proof in Mathematics
Proof in MathematicsProof in Mathematics
Proof in Mathematics
Ā 
signalsandsystemsnotes.pdf
signalsandsystemsnotes.pdfsignalsandsystemsnotes.pdf
signalsandsystemsnotes.pdf
Ā 

More from Andres Mendez-Vazquez

05 backpropagation automatic_differentiation
05 backpropagation automatic_differentiation05 backpropagation automatic_differentiation
05 backpropagation automatic_differentiationAndres Mendez-Vazquez
Ā 
01 Introduction to Neural Networks and Deep Learning
01 Introduction to Neural Networks and Deep Learning01 Introduction to Neural Networks and Deep Learning
01 Introduction to Neural Networks and Deep LearningAndres Mendez-Vazquez
Ā 
25 introduction reinforcement_learning
25 introduction reinforcement_learning25 introduction reinforcement_learning
25 introduction reinforcement_learningAndres Mendez-Vazquez
Ā 
Neural Networks and Deep Learning Syllabus
Neural Networks and Deep Learning SyllabusNeural Networks and Deep Learning Syllabus
Neural Networks and Deep Learning SyllabusAndres Mendez-Vazquez
Ā 
Introduction to artificial_intelligence_syllabus
Introduction to artificial_intelligence_syllabusIntroduction to artificial_intelligence_syllabus
Introduction to artificial_intelligence_syllabusAndres Mendez-Vazquez
Ā 
Ideas about a Bachelor in Machine Learning/Data Sciences
Ideas about a Bachelor in Machine Learning/Data SciencesIdeas about a Bachelor in Machine Learning/Data Sciences
Ideas about a Bachelor in Machine Learning/Data SciencesAndres Mendez-Vazquez
Ā 
Analysis of Algorithms Syllabus
Analysis of Algorithms  SyllabusAnalysis of Algorithms  Syllabus
Analysis of Algorithms SyllabusAndres Mendez-Vazquez
Ā 
20 k-means, k-center, k-meoids and variations
20 k-means, k-center, k-meoids and variations20 k-means, k-center, k-meoids and variations
20 k-means, k-center, k-meoids and variationsAndres Mendez-Vazquez
Ā 
17 vapnik chervonenkis dimension
17 vapnik chervonenkis dimension17 vapnik chervonenkis dimension
17 vapnik chervonenkis dimensionAndres Mendez-Vazquez
Ā 
A basic introduction to learning
A basic introduction to learningA basic introduction to learning
A basic introduction to learningAndres Mendez-Vazquez
Ā 
Introduction Mathematics Intelligent Systems Syllabus
Introduction Mathematics Intelligent Systems SyllabusIntroduction Mathematics Intelligent Systems Syllabus
Introduction Mathematics Intelligent Systems SyllabusAndres Mendez-Vazquez
Ā 
Introduction Machine Learning Syllabus
Introduction Machine Learning SyllabusIntroduction Machine Learning Syllabus
Introduction Machine Learning SyllabusAndres Mendez-Vazquez
Ā 
Schedule for Data Lab Community Path in Machine Learning
Schedule for Data Lab Community Path in Machine LearningSchedule for Data Lab Community Path in Machine Learning
Schedule for Data Lab Community Path in Machine LearningAndres Mendez-Vazquez
Ā 
Schedule Classes DataLab Community
Schedule Classes DataLab CommunitySchedule Classes DataLab Community
Schedule Classes DataLab CommunityAndres Mendez-Vazquez
Ā 
Introduction to logistic regression
Introduction to logistic regressionIntroduction to logistic regression
Introduction to logistic regressionAndres Mendez-Vazquez
Ā 

More from Andres Mendez-Vazquez (20)

2.03 bayesian estimation
2.03 bayesian estimation2.03 bayesian estimation
2.03 bayesian estimation
Ā 
06 recurrent neural_networks
06 recurrent neural_networks06 recurrent neural_networks
06 recurrent neural_networks
Ā 
05 backpropagation automatic_differentiation
05 backpropagation automatic_differentiation05 backpropagation automatic_differentiation
05 backpropagation automatic_differentiation
Ā 
Zetta global
Zetta globalZetta global
Zetta global
Ā 
01 Introduction to Neural Networks and Deep Learning
01 Introduction to Neural Networks and Deep Learning01 Introduction to Neural Networks and Deep Learning
01 Introduction to Neural Networks and Deep Learning
Ā 
25 introduction reinforcement_learning
25 introduction reinforcement_learning25 introduction reinforcement_learning
25 introduction reinforcement_learning
Ā 
Neural Networks and Deep Learning Syllabus
Neural Networks and Deep Learning SyllabusNeural Networks and Deep Learning Syllabus
Neural Networks and Deep Learning Syllabus
Ā 
Introduction to artificial_intelligence_syllabus
Introduction to artificial_intelligence_syllabusIntroduction to artificial_intelligence_syllabus
Introduction to artificial_intelligence_syllabus
Ā 
Ideas 09 22_2018
Ideas 09 22_2018Ideas 09 22_2018
Ideas 09 22_2018
Ā 
Ideas about a Bachelor in Machine Learning/Data Sciences
Ideas about a Bachelor in Machine Learning/Data SciencesIdeas about a Bachelor in Machine Learning/Data Sciences
Ideas about a Bachelor in Machine Learning/Data Sciences
Ā 
Analysis of Algorithms Syllabus
Analysis of Algorithms  SyllabusAnalysis of Algorithms  Syllabus
Analysis of Algorithms Syllabus
Ā 
20 k-means, k-center, k-meoids and variations
20 k-means, k-center, k-meoids and variations20 k-means, k-center, k-meoids and variations
20 k-means, k-center, k-meoids and variations
Ā 
18.1 combining models
18.1 combining models18.1 combining models
18.1 combining models
Ā 
17 vapnik chervonenkis dimension
17 vapnik chervonenkis dimension17 vapnik chervonenkis dimension
17 vapnik chervonenkis dimension
Ā 
A basic introduction to learning
A basic introduction to learningA basic introduction to learning
A basic introduction to learning
Ā 
Introduction Mathematics Intelligent Systems Syllabus
Introduction Mathematics Intelligent Systems SyllabusIntroduction Mathematics Intelligent Systems Syllabus
Introduction Mathematics Intelligent Systems Syllabus
Ā 
Introduction Machine Learning Syllabus
Introduction Machine Learning SyllabusIntroduction Machine Learning Syllabus
Introduction Machine Learning Syllabus
Ā 
Schedule for Data Lab Community Path in Machine Learning
Schedule for Data Lab Community Path in Machine LearningSchedule for Data Lab Community Path in Machine Learning
Schedule for Data Lab Community Path in Machine Learning
Ā 
Schedule Classes DataLab Community
Schedule Classes DataLab CommunitySchedule Classes DataLab Community
Schedule Classes DataLab Community
Ā 
Introduction to logistic regression
Introduction to logistic regressionIntroduction to logistic regression
Introduction to logistic regression
Ā 

Recently uploaded

Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Stand
Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night StandCall Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Stand
Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Standamitlee9823
Ā 
Employee leave management system project.
Employee leave management system project.Employee leave management system project.
Employee leave management system project.Kamal Acharya
Ā 
Work-Permit-Receiver-in-Saudi-Aramco.pptx
Work-Permit-Receiver-in-Saudi-Aramco.pptxWork-Permit-Receiver-in-Saudi-Aramco.pptx
Work-Permit-Receiver-in-Saudi-Aramco.pptxJuliansyahHarahap1
Ā 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...SUHANI PANDEY
Ā 
University management System project report..pdf
University management System project report..pdfUniversity management System project report..pdf
University management System project report..pdfKamal Acharya
Ā 
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...roncy bisnoi
Ā 
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
Ā 
Hostel management system project report..pdf
Hostel management system project report..pdfHostel management system project report..pdf
Hostel management system project report..pdfKamal Acharya
Ā 
Unleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leapUnleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leapRishantSharmaFr
Ā 
Unit 2- Effective stress & Permeability.pdf
Unit 2- Effective stress & Permeability.pdfUnit 2- Effective stress & Permeability.pdf
Unit 2- Effective stress & Permeability.pdfRagavanV2
Ā 
Design For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the startDesign For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the startQuintin Balsdon
Ā 
COST-EFFETIVE and Energy Efficient BUILDINGS ptx
COST-EFFETIVE  and Energy Efficient BUILDINGS ptxCOST-EFFETIVE  and Energy Efficient BUILDINGS ptx
COST-EFFETIVE and Energy Efficient BUILDINGS ptxJIT KUMAR GUPTA
Ā 
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar ā‰¼šŸ” Delhi door step de...
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar  ā‰¼šŸ” Delhi door step de...Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar  ā‰¼šŸ” Delhi door step de...
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar ā‰¼šŸ” Delhi door step de...9953056974 Low Rate Call Girls In Saket, Delhi NCR
Ā 
2016EF22_0 solar project report rooftop projects
2016EF22_0 solar project report rooftop projects2016EF22_0 solar project report rooftop projects
2016EF22_0 solar project report rooftop projectssmsksolar
Ā 
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoor
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoorTop Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoor
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoordharasingh5698
Ā 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueBhangaleSonal
Ā 
chapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineeringchapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineeringmulugeta48
Ā 

Recently uploaded (20)

Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Stand
Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night StandCall Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Stand
Call Girls In Bangalore ā˜Ž 7737669865 šŸ„µ Book Your One night Stand
Ā 
Employee leave management system project.
Employee leave management system project.Employee leave management system project.
Employee leave management system project.
Ā 
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
(INDIRA) Call Girl Meerut Call Now 8617697112 Meerut Escorts 24x7
Ā 
Work-Permit-Receiver-in-Saudi-Aramco.pptx
Work-Permit-Receiver-in-Saudi-Aramco.pptxWork-Permit-Receiver-in-Saudi-Aramco.pptx
Work-Permit-Receiver-in-Saudi-Aramco.pptx
Ā 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
Ā 
University management System project report..pdf
University management System project report..pdfUniversity management System project report..pdf
University management System project report..pdf
Ā 
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Call Girls Pimpri Chinchwad Call Me 7737669865 Budget Friendly No Advance Boo...
Ā 
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Wakad Call Me 7737669865 Budget Friendly No Advance Booking
Ā 
Hostel management system project report..pdf
Hostel management system project report..pdfHostel management system project report..pdf
Hostel management system project report..pdf
Ā 
Unleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leapUnleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leap
Ā 
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
Ā 
Unit 2- Effective stress & Permeability.pdf
Unit 2- Effective stress & Permeability.pdfUnit 2- Effective stress & Permeability.pdf
Unit 2- Effective stress & Permeability.pdf
Ā 
Design For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the startDesign For Accessibility: Getting it right from the start
Design For Accessibility: Getting it right from the start
Ā 
COST-EFFETIVE and Energy Efficient BUILDINGS ptx
COST-EFFETIVE  and Energy Efficient BUILDINGS ptxCOST-EFFETIVE  and Energy Efficient BUILDINGS ptx
COST-EFFETIVE and Energy Efficient BUILDINGS ptx
Ā 
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar ā‰¼šŸ” Delhi door step de...
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar  ā‰¼šŸ” Delhi door step de...Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar  ā‰¼šŸ” Delhi door step de...
Call Now ā‰½ 9953056974 ā‰¼šŸ” Call Girls In New Ashok Nagar ā‰¼šŸ” Delhi door step de...
Ā 
2016EF22_0 solar project report rooftop projects
2016EF22_0 solar project report rooftop projects2016EF22_0 solar project report rooftop projects
2016EF22_0 solar project report rooftop projects
Ā 
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoor
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoorTop Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoor
Top Rated Call Girls In chittoor šŸ“± {7001035870} VIP Escorts chittoor
Ā 
(INDIRA) Call Girl Aurangabad Call Now 8617697112 Aurangabad Escorts 24x7
(INDIRA) Call Girl Aurangabad Call Now 8617697112 Aurangabad Escorts 24x7(INDIRA) Call Girl Aurangabad Call Now 8617697112 Aurangabad Escorts 24x7
(INDIRA) Call Girl Aurangabad Call Now 8617697112 Aurangabad Escorts 24x7
Ā 
Double Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torqueDouble Revolving field theory-how the rotor develops torque
Double Revolving field theory-how the rotor develops torque
Ā 
chapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineeringchapter 5.pptx: drainage and irrigation engineering
chapter 5.pptx: drainage and irrigation engineering
Ā 

01.01 vector spaces

  • 1. Mathematics for Artiļ¬cial Intelligence Vector Spaces Andres Mendez-Vazquez March 14, 2020 1 / 59
  • 2. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 2 / 59
  • 3. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 3 / 59
  • 4. Introduction What is this class about? It is clear that the use of mathematics is essential for the data mining and machine learning ļ¬elds. Therefore... The understanding of Mathematical Modeling is part of the deal... If you want to be A Good Data Scientist!!! 4 / 59
  • 5. Introduction What is this class about? It is clear that the use of mathematics is essential for the data mining and machine learning ļ¬elds. Therefore... The understanding of Mathematical Modeling is part of the deal... If you want to be A Good Data Scientist!!! 4 / 59
  • 6. Introduction What is this class about? It is clear that the use of mathematics is essential for the data mining and machine learning ļ¬elds. Therefore... The understanding of Mathematical Modeling is part of the deal... If you want to be A Good Data Scientist!!! 4 / 59
  • 7. Example Imagine A web surfer moves from a web page to another web page... Question: How do you model this? You can use a graph!!! 5 / 59
  • 8. Example Imagine A web surfer moves from a web page to another web page... Question: How do you model this? You can use a graph!!! 1 8 2 9 10 4 7 14 11 12 6 3 13 5 5 / 59
  • 10. Thus We can build a matrix M = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ P11 P12 Ā· Ā· Ā· P1N P21 P22 Ā· Ā· Ā· P2N ... ... ... ... PN1 PN2 Ā· Ā· Ā· PNN ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø (1) Thus, it is possible to obtain certain information by looking at the eigenvector and eigenvalues These vectors vĪ»s and values Ī» s have the property that MvĪ» = Ī»vĪ» (2) 7 / 59
  • 11. Thus We can build a matrix M = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ P11 P12 Ā· Ā· Ā· P1N P21 P22 Ā· Ā· Ā· P2N ... ... ... ... PN1 PN2 Ā· Ā· Ā· PNN ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø (1) Thus, it is possible to obtain certain information by looking at the eigenvector and eigenvalues These vectors vĪ»s and values Ī» s have the property that MvĪ» = Ī»vĪ» (2) 7 / 59
  • 12. This is the Basis of Page Rank in Google For example Look at this example... 8 / 59
  • 13. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 9 / 59
  • 14. About 4000 years ago Babylonians knew how to solve the following kind of systems ax + by = c dx + ey = f As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow It is only after the death of Plato and Aristotle, that the Chinese (Nine Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3 system. By working an ā€œelimination methodā€ Similar to the one devised by Gauss 2000 years later for general systems. 10 / 59
  • 15. About 4000 years ago Babylonians knew how to solve the following kind of systems ax + by = c dx + ey = f As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow It is only after the death of Plato and Aristotle, that the Chinese (Nine Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3 system. By working an ā€œelimination methodā€ Similar to the one devised by Gauss 2000 years later for general systems. 10 / 59
  • 16. About 4000 years ago Babylonians knew how to solve the following kind of systems ax + by = c dx + ey = f As always the ļ¬rst steps in any ļ¬eld of knowledge tend to be slow It is only after the death of Plato and Aristotle, that the Chinese (Nine Chapters of the Mathematical Art 200 B.C.) were able to solve 3 Ɨ 3 system. By working an ā€œelimination methodā€ Similar to the one devised by Gauss 2000 years later for general systems. 10 / 59
  • 17. Not only that The Matrix Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations in his book ā€œDisquisitions.ā€ The Final Deļ¬nition of Matrix It was introduced by Cayley in two papers in 1850 and 1858 respectively, which allowed him to prove the important Cayley-Hamilton Theorem. There is quite a lot Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007). 11 / 59
  • 18. Not only that The Matrix Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations in his book ā€œDisquisitions.ā€ The Final Deļ¬nition of Matrix It was introduced by Cayley in two papers in 1850 and 1858 respectively, which allowed him to prove the important Cayley-Hamilton Theorem. There is quite a lot Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007). 11 / 59
  • 19. Not only that The Matrix Gauss deļ¬ned implicitly the concept of a Matrix as linear transformations in his book ā€œDisquisitions.ā€ The Final Deļ¬nition of Matrix It was introduced by Cayley in two papers in 1850 and 1858 respectively, which allowed him to prove the important Cayley-Hamilton Theorem. There is quite a lot Kleiner, I., A History of Abstract Algebra (BirkhƤuser Boston, 2007). 11 / 59
  • 20. Matrix can help to represent many things They are important for many calculations as a11x1 + a12x2 + ... + a1nxn =b1, a21x1 + a22x2 + ... + a2nxn =b2, Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· am1x1 + am2x2 + ... + amnxn =b2. It is clear We would like to collect those linear equations in a compact structure that allows for simpler manipulation. 12 / 59
  • 21. Matrix can help to represent many things They are important for many calculations as a11x1 + a12x2 + ... + a1nxn =b1, a21x1 + a22x2 + ... + a2nxn =b2, Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· Ā· am1x1 + am2x2 + ... + amnxn =b2. It is clear We would like to collect those linear equations in a compact structure that allows for simpler manipulation. 12 / 59
  • 22. Therefore, we have For example x = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ x1 x2 ... xn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø , b = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ b1 b2 ... bn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø and A = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ a11 a12 Ā· Ā· Ā· a1n a21 a22 Ā· Ā· Ā· a2n ... ... ... ... am1 am2 Ā· Ā· Ā· amn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø Using a little of notation Ax = b 13 / 59
  • 23. Therefore, we have For example x = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ x1 x2 ... xn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø , b = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ b1 b2 ... bn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø and A = ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ a11 a12 Ā· Ā· Ā· a1n a21 a22 Ā· Ā· Ā· a2n ... ... ... ... am1 am2 Ā· Ā· Ā· amn ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø Using a little of notation Ax = b 13 / 59
  • 24. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 14 / 59
  • 25. Introduction As always, we star with a simple fact Everything is an element in a set. For example The set of Real Numbers R. The set of n-tuples in Rn. The set of Complex Number C. 15 / 59
  • 26. Introduction As always, we star with a simple fact Everything is an element in a set. For example The set of Real Numbers R. The set of n-tuples in Rn. The set of Complex Number C. 15 / 59
  • 27. Introduction As always, we star with a simple fact Everything is an element in a set. For example The set of Real Numbers R. The set of n-tuples in Rn. The set of Complex Number C. 15 / 59
  • 28. Introduction As always, we star with a simple fact Everything is an element in a set. For example The set of Real Numbers R. The set of n-tuples in Rn. The set of Complex Number C. 15 / 59
  • 29. Deļ¬nition We shall say that K is a ļ¬eld if it satisļ¬es the following conditions for the addition Property Formalism Addition is Commutative x + y = y + x for all x, y āˆˆ K Addition is associative x + (y + z) = (x + y) + z for all x, y, z āˆˆ K Existence of 0 x + 0 = x, for every x āˆˆ K Existence of the inverse āˆ€x there is āˆƒ āˆ’ x =ā‡’ x + (āˆ’x) = 0 16 / 59
  • 30. Furthermore We have the following properties for the product Property Formalism Product is Commutative xy = yx for all x, y āˆˆ K Product is associative x (yz) = (xy) z for all x, y, z āˆˆ K Existence of 1 1x = x1 = x, for every x āˆˆ K. Existence of the inverse xāˆ’1 or 1 x in K such that xxāˆ’1 = 1. Multiplication is Distributive over addition x (y + z) = xy + xz, for all x, y, z āˆˆ K 17 / 59
  • 31. Therefore Examples 1 For example the reals R and the C. 2 In addition, we have the rationals Q too. The elements of the ļ¬eld will be also called numbers Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K. 18 / 59
  • 32. Therefore Examples 1 For example the reals R and the C. 2 In addition, we have the rationals Q too. The elements of the ļ¬eld will be also called numbers Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K. 18 / 59
  • 33. Therefore Examples 1 For example the reals R and the C. 2 In addition, we have the rationals Q too. The elements of the ļ¬eld will be also called numbers Thus, we will use this ideas to deļ¬ne the Vector Space V over a ļ¬eld K. 18 / 59
  • 34. Then, we get a crazy moment How do we relate these numbers to obtain certain properties We have then the vector and matrix structures for this... ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ a11 Ā· Ā· Ā· Ā· Ā· Ā· a1n a21 Ā· Ā· Ā· Ā· Ā· Ā· a2n ... ... ... ... an1 Ā· Ā· Ā· Ā· Ā· Ā· ann ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø and ļ£« ļ£¬ ļ£¬ ļ£¬ ļ£¬ ļ£­ a11 a21 ... an1 ļ£¶ ļ£· ļ£· ļ£· ļ£· ļ£ø 19 / 59
  • 35. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 20 / 59
  • 36. Vector Space V Deļ¬nition A vector space V over the ļ¬eld K is a set of objects which can be added and multiplied by elements of K. Where The sum of two elements of V is again an element of V . The product of an element of V by an element of K is an element of V . 21 / 59
  • 37. Vector Space V Deļ¬nition A vector space V over the ļ¬eld K is a set of objects which can be added and multiplied by elements of K. Where The sum of two elements of V is again an element of V . The product of an element of V by an element of K is an element of V . 21 / 59
  • 38. Vector Space V Deļ¬nition A vector space V over the ļ¬eld K is a set of objects which can be added and multiplied by elements of K. Where The sum of two elements of V is again an element of V . The product of an element of V by an element of K is an element of V . 21 / 59
  • 39. Vector Space V Deļ¬nition A vector space V over the ļ¬eld K is a set of objects which can be added and multiplied by elements of K. Where The sum of two elements of V is again an element of V . The product of an element of V by an element of K is an element of V . 21 / 59
  • 40. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 41. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 42. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 43. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 44. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 45. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 46. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 47. Properties We have then 1 Given elements u, v, w of V , we have (u + v) + w = u + (v + w). 2 There is an element of V , denoted by O, such that O + u = u + O = u for all elements u of V . 3 Given an element u of V , there exists an element āˆ’u in V such that u + (āˆ’u) = O. 4 For all elements u, v of V , we have u + v = v + u. 5 For all elements u of V , we have 1 Ā· u = u. 6 If c is a number, then c (u + v) = cu + cv. 7 if a, b are two numbers, then (ab) v = a (bv). 8 If a, b are two numbers, then (a + b) v = av + bv. 22 / 59
  • 48. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 23 / 59
  • 49. Notation First, u + (āˆ’v) As u āˆ’ v. For O We will write sometimes 0. The elements in the ļ¬eld K They can receive the name of number or scalar. 24 / 59
  • 50. Notation First, u + (āˆ’v) As u āˆ’ v. For O We will write sometimes 0. The elements in the ļ¬eld K They can receive the name of number or scalar. 24 / 59
  • 51. Notation First, u + (āˆ’v) As u āˆ’ v. For O We will write sometimes 0. The elements in the ļ¬eld K They can receive the name of number or scalar. 24 / 59
  • 52. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 25 / 59
  • 53. Many Times We have this kind of data sets (House Prices) 26 / 59
  • 54. Therefore We can represent these relations as vectors Squared Feet Price = 2104 400 , 1800 460 , 1600 300 , ... Thus, we can start using All the tools that Linear Algebra can provide!!! 27 / 59
  • 55. Therefore We can represent these relations as vectors Squared Feet Price = 2104 400 , 1800 460 , 1600 300 , ... Thus, we can start using All the tools that Linear Algebra can provide!!! 27 / 59
  • 56. Thus We can adjust a line/hyper-plane to be able to forecast prices 28 / 59
  • 57. Thus, Our Objective To ļ¬nd such hyper-plane To do forecasting on the prices of a house given its surface size!!! Here, where ā€œLearningā€ comes around Basically, the process deļ¬ned in Machine Learning!!! 29 / 59
  • 58. Thus, Our Objective To ļ¬nd such hyper-plane To do forecasting on the prices of a house given its surface size!!! Here, where ā€œLearningā€ comes around Basically, the process deļ¬ned in Machine Learning!!! 29 / 59
  • 59. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 30 / 59
  • 60. Sub-spaces Deļ¬nition Let V a vector space and W āŠ† V , thus W is a subspace if: 1 If v, w āˆˆ W, then v + w āˆˆ W. 2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W. 3 The element 0 āˆˆ V is also an element of W. 31 / 59
  • 61. Sub-spaces Deļ¬nition Let V a vector space and W āŠ† V , thus W is a subspace if: 1 If v, w āˆˆ W, then v + w āˆˆ W. 2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W. 3 The element 0 āˆˆ V is also an element of W. 31 / 59
  • 62. Sub-spaces Deļ¬nition Let V a vector space and W āŠ† V , thus W is a subspace if: 1 If v, w āˆˆ W, then v + w āˆˆ W. 2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W. 3 The element 0 āˆˆ V is also an element of W. 31 / 59
  • 63. Sub-spaces Deļ¬nition Let V a vector space and W āŠ† V , thus W is a subspace if: 1 If v, w āˆˆ W, then v + w āˆˆ W. 2 If v āˆˆ W and c āˆˆ K, then cv āˆˆ W. 3 The element 0 āˆˆ V is also an element of W. 31 / 59
  • 64. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 32 / 59
  • 65. Some ways of recognizing Sub-spaces Theorem A non-empty subset W of V is a subspace of V if and only if for each pair of vectors v, w āˆˆ W and each scalar c āˆˆ K the vector cv + w āˆˆ W. 33 / 59
  • 66. Example For R2 0 1 2 3 1 2 4 0 34 / 59
  • 67. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 35 / 59
  • 68. Linear Combinations Deļ¬nition Let V an arbitrary vector space, and let v1, v2, ..., vn āˆˆ V and x1, x2, ..., xn āˆˆ K. Then, an expression like x1v1 + x2v2 + ... + xnvn (3) is called a linear combination of v1, v2, ..., vn. 36 / 59
  • 69. Classic Examples Endmember Representation in Hyperspectral Images Look at the board Geometric Representation of addition of forces in Physics Look at the board!! 37 / 59
  • 70. Classic Examples Endmember Representation in Hyperspectral Images Look at the board Geometric Representation of addition of forces in Physics Look at the board!! 37 / 59
  • 71. Properties and Deļ¬nitions Theorem Let V be a vector space over the ļ¬eld K. The intersection of any collection of sub-spaces of V is a subspace of V . Deļ¬nition Let S be a set of vectors in a vector space V . The sub-space spanned by S is deļ¬ned as the intersection W of all sub-spaces of V which contains S. When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall simply call W the sub-space spanned by the vectors v1, v2, . . . , vn. 38 / 59
  • 72. Properties and Deļ¬nitions Theorem Let V be a vector space over the ļ¬eld K. The intersection of any collection of sub-spaces of V is a subspace of V . Deļ¬nition Let S be a set of vectors in a vector space V . The sub-space spanned by S is deļ¬ned as the intersection W of all sub-spaces of V which contains S. When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall simply call W the sub-space spanned by the vectors v1, v2, . . . , vn. 38 / 59
  • 73. Properties and Deļ¬nitions Theorem Let V be a vector space over the ļ¬eld K. The intersection of any collection of sub-spaces of V is a subspace of V . Deļ¬nition Let S be a set of vectors in a vector space V . The sub-space spanned by S is deļ¬ned as the intersection W of all sub-spaces of V which contains S. When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall simply call W the sub-space spanned by the vectors v1, v2, . . . , vn. 38 / 59
  • 74. Properties and Deļ¬nitions Theorem Let V be a vector space over the ļ¬eld K. The intersection of any collection of sub-spaces of V is a subspace of V . Deļ¬nition Let S be a set of vectors in a vector space V . The sub-space spanned by S is deļ¬ned as the intersection W of all sub-spaces of V which contains S. When S is a ļ¬nite set of vectors, S = {v1, v2, . . . , vn}, we shall simply call W the sub-space spanned by the vectors v1, v2, . . . , vn. 38 / 59
  • 75. We get the following Theorem Theorem The subspace spanned by S = āˆ… is the set of all linear combinations of vectors in S. 39 / 59
  • 76. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 40 / 59
  • 77. Linear Independence Deļ¬nition Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We have that v1, v2, ..., vn are linearly dependent over K if there are elements a1, a2, ..., an āˆˆ K not all equal to 0 such that a1v1 + a2v2 + ... + anvn = O Thus Therefore, if there are not such numbers, then we say that v1, v2, ..., vn are linearly independent. We have the following Example!!! 41 / 59
  • 78. Linear Independence Deļ¬nition Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We have that v1, v2, ..., vn are linearly dependent over K if there are elements a1, a2, ..., an āˆˆ K not all equal to 0 such that a1v1 + a2v2 + ... + anvn = O Thus Therefore, if there are not such numbers, then we say that v1, v2, ..., vn are linearly independent. We have the following Example!!! 41 / 59
  • 79. Linear Independence Deļ¬nition Let V be a vector space over a ļ¬eld K, and let v1, v2, ..., vn āˆˆ V . We have that v1, v2, ..., vn are linearly dependent over K if there are elements a1, a2, ..., an āˆˆ K not all equal to 0 such that a1v1 + a2v2 + ... + anvn = O Thus Therefore, if there are not such numbers, then we say that v1, v2, ..., vn are linearly independent. We have the following Example!!! 41 / 59
  • 80. Basis Deļ¬nition If elements v1, v2, ..., vn generate V and in addition are linearly independent, then {v1, v2, ..., vn} is called a basis of V . In other words the elements v1, v2, ..., vn form a basis of V . Examples The Classic Ones!!! 42 / 59
  • 81. Basis Deļ¬nition If elements v1, v2, ..., vn generate V and in addition are linearly independent, then {v1, v2, ..., vn} is called a basis of V . In other words the elements v1, v2, ..., vn form a basis of V . Examples The Classic Ones!!! 42 / 59
  • 82. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 43 / 59
  • 83. Coordinates Theorem Let V be a vector space. Let v1, v2, ..., vn be linearly independent elements of V. Let x1, . . . , xn and y1, . . . , yn be numbers. Suppose that we have x1v1 + x2v2 + Ā· Ā· Ā· + xnvn = y1v1 + y2v2 + Ā· Ā· Ā· + ynvn (4) Then, xi = yi for all i = 1, . . . , n. 44 / 59
  • 84. Coordinates Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn. Thus, this n-tuple is uniquely determined by v We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the basis. The nāˆ’tuple X = (x1, x2, . . . , xn) It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} . 45 / 59
  • 85. Coordinates Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn. Thus, this n-tuple is uniquely determined by v We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the basis. The nāˆ’tuple X = (x1, x2, . . . , xn) It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} . 45 / 59
  • 86. Coordinates Let V be a vector space, and let {v1, v2, ..., vn} be a basis of V For all v āˆˆ V , v = x1v1 + x2v2 + Ā· Ā· Ā· + xnvn. Thus, this n-tuple is uniquely determined by v We will call (x1, x2, . . . , xn) as the coordinates of v with respect to the basis. The nāˆ’tuple X = (x1, x2, . . . , xn) It is the coordinate vector of v with respect to the basis {v1, v2, ..., vn} . 45 / 59
  • 87. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 46 / 59
  • 88. Properties of a Basis Theorem - (Limit in the size of the basis) Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let w1, w2, ..., wn be elements of V , and assume that n > m. Then w1, w2, ..., wn are linearly dependent. Examples Matrix Space Canonical Space vectors etc 47 / 59
  • 89. Properties of a Basis Theorem - (Limit in the size of the basis) Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let w1, w2, ..., wn be elements of V , and assume that n > m. Then w1, w2, ..., wn are linearly dependent. Examples Matrix Space Canonical Space vectors etc 47 / 59
  • 90. Properties of a Basis Theorem - (Limit in the size of the basis) Let V be a vector space over a ļ¬eld K with a basis {v1, v2, ..., vm}. Let w1, w2, ..., wn be elements of V , and assume that n > m. Then w1, w2, ..., wn are linearly dependent. Examples Matrix Space Canonical Space vectors etc 47 / 59
  • 91. Some Basic Deļ¬nitions We will deļ¬ne the dimension of a vector space V over K As the number of elements in the basis. Denoted by dimK V , or simply dim V Therefore A vector space with a basis consisting of a ļ¬nite number of elements, or the zero vector space, is called a ļ¬nite dimensional. Now Is this number unique? 48 / 59
  • 92. Some Basic Deļ¬nitions We will deļ¬ne the dimension of a vector space V over K As the number of elements in the basis. Denoted by dimK V , or simply dim V Therefore A vector space with a basis consisting of a ļ¬nite number of elements, or the zero vector space, is called a ļ¬nite dimensional. Now Is this number unique? 48 / 59
  • 93. Some Basic Deļ¬nitions We will deļ¬ne the dimension of a vector space V over K As the number of elements in the basis. Denoted by dimK V , or simply dim V Therefore A vector space with a basis consisting of a ļ¬nite number of elements, or the zero vector space, is called a ļ¬nite dimensional. Now Is this number unique? 48 / 59
  • 94. Maximal Set of Linearly Independent Elements Theorem Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly independent elements of V . Then, {v1, v2, ..., vn} is a basis of V . Theorem Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V . 49 / 59
  • 95. Maximal Set of Linearly Independent Elements Theorem Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly independent elements of V . Then, {v1, v2, ..., vn} is a basis of V . Theorem Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V . 49 / 59
  • 96. Maximal Set of Linearly Independent Elements Theorem Let V be a vector space, and {v1, v2, ..., vn} a maximal set of linearly independent elements of V . Then, {v1, v2, ..., vn} is a basis of V . Theorem Let V be a vector space of dimension n, and let v1, v2, ..., vn be linearly independent elements of V . Then, v1, v2, ..., vn constitutes a basis of V . 49 / 59
  • 97. Equality between Basis Corollary Let V be a vector space and let W be a subspace. If dim W = dim V then V = W. Proof At the Board... Corollary Let V be a vector space of dimension n. Let r be a positive integer with r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a basis of V . Proof At the Board... 50 / 59
  • 98. Equality between Basis Corollary Let V be a vector space and let W be a subspace. If dim W = dim V then V = W. Proof At the Board... Corollary Let V be a vector space of dimension n. Let r be a positive integer with r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a basis of V . Proof At the Board... 50 / 59
  • 99. Equality between Basis Corollary Let V be a vector space and let W be a subspace. If dim W = dim V then V = W. Proof At the Board... Corollary Let V be a vector space of dimension n. Let r be a positive integer with r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a basis of V . Proof At the Board... 50 / 59
  • 100. Equality between Basis Corollary Let V be a vector space and let W be a subspace. If dim W = dim V then V = W. Proof At the Board... Corollary Let V be a vector space of dimension n. Let r be a positive integer with r < n, and let v1, v2, ..., vr be linearly independent elements of V. Then one can ļ¬nd elements vr+1, vr+2, ..., vn such that {v1, v2, ..., vn} is a basis of V . Proof At the Board... 50 / 59
  • 101. Finally Theorem Let V be a vector space having a basis consisting of n elements. Let W be a subspace which does not consist of O alone. Then W has a basis, and the dimension of W is ā‰¤ n. Proof At the Board... 51 / 59
  • 102. Finally Theorem Let V be a vector space having a basis consisting of n elements. Let W be a subspace which does not consist of O alone. Then W has a basis, and the dimension of W is ā‰¤ n. Proof At the Board... 51 / 59
  • 103. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 52 / 59
  • 104. Feature Vector Deļ¬nition A feature vector is a n-dimensional vector of numerical features that represent an object. Why is this important? This allows to use linear algebra to represent basic classiļ¬cation algorithms because The tuples {(x, y) |x āˆˆ Kn, y āˆˆ K} can be easily used to design speciļ¬c algorithms. 53 / 59
  • 105. Feature Vector Deļ¬nition A feature vector is a n-dimensional vector of numerical features that represent an object. Why is this important? This allows to use linear algebra to represent basic classiļ¬cation algorithms because The tuples {(x, y) |x āˆˆ Kn, y āˆˆ K} can be easily used to design speciļ¬c algorithms. 53 / 59
  • 106. Outline 1 Why Liner Algebra Why and What? A Little Bit of History 2 The Beginning Fields 3 Vector Space Introduction Some Notes in Notation Use of Linear Algebra in Regression... Sub-spaces and Linear Combinations Recognizing Sub-spaces Combinations 4 Basis and Dimensions Basis Coordinates Basis and Dimensions 5 Application in Machine Learning Feature Vector Least Squared Error 54 / 59
  • 107. Least Squared Error We need to ļ¬t a series of points against a certain function We want The general problem is given a set of functions f1, f2, ..., fK ļ¬nd values of coeļ¬ƒcients a1, a2, ..., ak such that the linear combination: y = a1f1 (x) + Ā· Ā· Ā· + aKfK (x) (5) 55 / 59
  • 108. Least Squared Error We need to ļ¬t a series of points against a certain function We want The general problem is given a set of functions f1, f2, ..., fK ļ¬nd values of coeļ¬ƒcients a1, a2, ..., ak such that the linear combination: y = a1f1 (x) + Ā· Ā· Ā· + aKfK (x) (5) 55 / 59
  • 109. Thus We have that given the datasets {(x1, y1) , ..., (xN , yN )} x = 1 N N i=1 xi. (6) Thus, we have the following problem A Possible High Variance on the Data itself Variance Ļƒ2 x = 1 N N i=1 (xi āˆ’ x) (7) 56 / 59
  • 110. Thus We have that given the datasets {(x1, y1) , ..., (xN , yN )} x = 1 N N i=1 xi. (6) Thus, we have the following problem A Possible High Variance on the Data itself Variance Ļƒ2 x = 1 N N i=1 (xi āˆ’ x) (7) 56 / 59
  • 111. Thus We have that given the datasets {(x1, y1) , ..., (xN , yN )} x = 1 N N i=1 xi. (6) Thus, we have the following problem A Possible High Variance on the Data itself Variance Ļƒ2 x = 1 N N i=1 (xi āˆ’ x) (7) 56 / 59
  • 112. Now Assume A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0. We get a series of errors given the following observations {(x1, y1) , ..., (xN , yN )} {y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} . Then, the mean should be really small (If it is a good ļ¬t) Ļƒ2 yāˆ’(ax+b) = 1 N N i=1 (yi āˆ’ (axi + b))2 (8) 57 / 59
  • 113. Now Assume A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0. We get a series of errors given the following observations {(x1, y1) , ..., (xN , yN )} {y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} . Then, the mean should be really small (If it is a good ļ¬t) Ļƒ2 yāˆ’(ax+b) = 1 N N i=1 (yi āˆ’ (axi + b))2 (8) 57 / 59
  • 114. Now Assume A linear equation y = ax + b, then y āˆ’ (ax + b) ā‰ˆ 0. We get a series of errors given the following observations {(x1, y1) , ..., (xN , yN )} {y1 āˆ’ (ax1 + b) , ..., yN āˆ’ (axN + b)} . Then, the mean should be really small (If it is a good ļ¬t) Ļƒ2 yāˆ’(ax+b) = 1 N N i=1 (yi āˆ’ (axi + b))2 (8) 57 / 59
  • 115. Thus We can deļ¬ne the following error Ei (a, b) = y āˆ’ (ax + b) E (a, b) = N i=1 Ei (a, b) = N i=1 (yi āˆ’ (axi + b)) (9) We want to minimize the previous equation āˆ‚E āˆ‚a = 0, āˆ‚E āˆ‚b = 0. 58 / 59
  • 116. Thus We can deļ¬ne the following error Ei (a, b) = y āˆ’ (ax + b) E (a, b) = N i=1 Ei (a, b) = N i=1 (yi āˆ’ (axi + b)) (9) We want to minimize the previous equation āˆ‚E āˆ‚a = 0, āˆ‚E āˆ‚b = 0. 58 / 59
  • 117. Finally Look at the Board We need to obtain the necessary equations. 59 / 59