85
Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada, Shimon Edelman Presented by Maksym Del ([email protected]) 2017

Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

  • Upload
    others

  • View
    22

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Similarity, kernels, and the fundamental constraints on

cognitionA survey paper

Authors: Reza Shahbazi, Rajeev Raizada, Shimon Edelman

Presented by Maksym Del

([email protected])

2017

Page 2: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Paper outline

• 4 fundamental constrains on any learning system

• Kernels

• Addressing constrains with kernels

• Application to neuroscience

* machine learning * neuroscience

Page 3: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

4 Fundamental Constraintson any learning system (animal, human, robot etc.)

Page 4: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

We expect learning system to

• 1) Measure external stimuli• Gather data from human eyes, ears, robot sensors etc.

Page 5: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

We expect learning system to

• 1) Measure external stimuli• Gather data from human eyes, ears, robot sensors etc.

• 2) Generalize from familiar to similar unfamiliar stimuli• The tiger is dangerous -> the lion is dangerous

Page 6: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

We expect learning system to

• 1) Measure external stimuli• Gather data from human eyes, ears, robot sensors etc.

• 2) Generalize from familiar to unfamiliar stimuli• The tiger is dangerous -> the lion is dangerous

• 3) Deal with high dimensional data• A lot of features from many sensors

Page 7: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

We expect learning system to

• 1) Measure external stimuli• Gather data from human eyes, ears, robot sensors etc.

• 2) Generalize from familiar to unfamiliar stimuli• The tiger is dangerous -> the lion is dangerous

• 3) Deal with high dimensional data• A lot of features from many sensors; sparse data

• 4) Solve complex tasks• Puzzles, regression / classification tasks, etc.

Page 8: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

• 1) Measurement:

• 2) Similarity:

• 3) Dimensionality:

• 4) Complexity:

Page 9: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

• 1) Measurement: dealing with raw data

• 2) Similarity:

• 3) Dimensionality:

• 4) Complexity:

Page 10: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

• 1) Measurement: dealing with raw data

• 2) Similarity: similarity estimation to generalize

• 3) Dimensionality:

• 4) Complexity:

Page 11: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

• 1) Measurement: dealing with raw data

• 2) Similarity: similarity estimation to generalize

• 3) Dimensionality: need to reduce it

• 4) Complexity:

Page 12: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

• 1) Measurement: dealing with raw data

• 2) Similarity: similarity estimation to generalize

• 3) Dimensionality: need to reduce it

• 4) Complexity: map problem into another space to make it simpler

Page 13: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

So far we defined 4 fundamental constraints on any learning system:

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

Page 14: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Paper outline

• 4 fundamental constrains on any learning system

• Kernels

• Addressing constrains with kernels

• Application to neuroscience

* machine learning * neuroscience

Page 15: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernelsin feature space mapping and as a similarity measure

Page 16: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets say we have a linear tool

Page 17: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets say we have a linear tool

Page 18: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets say we have a linear tool

Page 19: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

And linearly separable features

Image: https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 20: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

We are fine

Image: https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 21: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Non-linear data?

Image: https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 22: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Make it linearly separable

• Map the feature space to higher dimension

• Separate with linear tool

Page 23: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Non-linear data?

Image: https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 24: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets formalize

• x, y are n dimensional inputs

Page 25: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets formalize

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

Page 26: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets formalize

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

Page 27: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets formalize

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

• To map x and y to higher dimension we have to compute <f(x), f(y)>

Page 28: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernel

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

• To map x and y to higher dimension we have to compute <f(x), f(y)>

• However, computing f(x) and f(y) might be hard

Page 29: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernel

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

• To map x and y to higher dimension we have to compute <f(x), f(y)>

• However, computing f(x) and f(y) might be hard

• Note, that <f(x), f(y)> = r

Page 30: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernel

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

• To map x and y to higher dimension we have to compute <f(x), f(y)>

• However, computing f(x) and f(y) might be hard

• Note, that <f(x), f(y)> = r

• So lets introduce K(x, y) = r

Page 31: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernel

• x, y are n dimensional inputs

• f(.) is a map from n-dimension to m-dimension space (m > n)

• x and y often occur in the form of dot product: <x, y>

• To map x and y to higher dimension we have to compute <f(x), f(y)>

• However, computing f(x) and f(y) might be hard

• Note, that <f(x), f(y)> = r

• So lets introduce K(x, y) = r

• K(x, y) is called kernel

Page 32: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 33: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 34: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

• x = (1, 2, 3); y = (4, 5, 6)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 35: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

• x = (1, 2, 3); y = (4, 5, 6)

• f(x) = (1, 2, 3, 2, 4, 6, 3, 6, 9)

• f(y) = (16, 20, 24, 20, 25, 30, 24, 30, 36)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 36: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

• x = (1, 2, 3); y = (4, 5, 6)

• f(x) = (1, 2, 3, 2, 4, 6, 3, 6, 9)

• f(y) = (16, 20, 24, 20, 25, 30, 24, 30, 36)

• <f(x), f(y)> = 16 + 40 + 72 + 40 + 100+ 180 + 72 + 180 + 324 = 1024

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 37: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

• x = (1, 2, 3); y = (4, 5, 6)

• f(x) = (1, 2, 3, 2, 4, 6, 3, 6, 9)

• f(y) = (16, 20, 24, 20, 25, 30, 24, 30, 36)

• <f(x), f(y)> = 16 + 40 + 72 + 40 + 100+ 180 + 72 + 180 + 324 = 1024

• K(x, y ) = (<x, y>)^2

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 38: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Simple example

• x = (x1, x2, x3); y = (y1, y2, y3)

• f(x) = (x1x1, x1x2, x1x3, x2x1, x2x2, x2x3, x3x1, x3x2, x3x3)

• x = (1, 2, 3); y = (4, 5, 6)

• f(x) = (1, 2, 3, 2, 4, 6, 3, 6, 9)

• f(y) = (16, 20, 24, 20, 25, 30, 24, 30, 36)

• <f(x), f(y)> = 16 + 40 + 72 + 40 + 100+ 180 + 72 + 180 + 324 = 1024

• K(x, y ) = (<x, y>)^2

• K(x, y) = (4 + 10 + 18 ) ^2 = 32^2 = 1024

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 39: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

2 views on kernels

• Kernel as an implicit feature map (aka Kernel trick)

• ?

Page 40: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

Page 41: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

• But what is dot product <x,y>?

Page 42: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

• But what is dot product <x,y>?

• It is a measure of overlap between 2 vectors

Page 43: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

• But what is dot product <x,y>?

• It is a measure of overlap between 2 vectors

• Measure of overlap = measure of similarity

Page 44: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

• But what is dot product <x,y>?

• It is a measure of overlap between 2 vectors

• Measure of overlap = measure of similarity

• So what is kernel K = <f(x),f(y)>?

Page 45: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Another view on kernel

• We work with a dot product: <x,y>

• But what is dot product <x,y>?

• It is a measure of overlap between 2 vectors

• Measure of overlap = measure of similarity

• So what is kernel K = <f(x),f(y)>?

• It is a more high-level measure of similarity!

Page 46: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

2 views on kernels

• Kernel as an implicit feature map (aka Kernel trick)

• Kernel as a measure of similarity

Page 47: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Paper outline

• 4 fundamental constrains on any learning system

• Kernels

• Addressing constrains with kernels

• Application to neuroscience

* machine learning * neuroscience

Page 48: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Addressing constrains with the kernels

Page 49: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

So far we defined 4 fundamental constraints on any learning system:

Page 50: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Constraints

So far we defined 4 fundamental constraints on any learning system:

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

Page 51: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernels

And inferred 2 definitions of kernels:

Page 52: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Kernels

And inferred 2 definitions of kernels:

• Implicit feature map (kernel trick)

• A kernel is a measure of similarity

Page 53: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets map constrains with views on kernels

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 54: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets map constrains with views on kernels

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

• Direct access to information: K(x,y)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 55: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets map constrains with views on kernels

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

• Direct access to information: K(x,y)

• Kernel as similarity measure

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 56: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets map constrains with views on kernels

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

• Direct access to information: K(x,y)

• Kernel as similarity measure

• Kernel trick

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 57: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Lets map constrains with views on kernels

• Measurement constraint

• Similarity constraint

• Dimensionality constraint

• Complexity constraint

• Direct access to information: K(x,y)

• Kernel as similarity measure

• Kernel trick

• Kernel trick

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 58: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Paper outline

• 4 fundamental constrains on any learning system

• Kernels

• Addressing constrains the kernels

• Application to neuroscience

* machine learning * neuroscience

Page 59: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Application to neurosciencesimilarity and kernels in animal behavior

Page 60: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Example learning system

Page 61: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Tasks that can help an animal survive

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Example

• ‘‘Is this a dangerous animal?’’

• “Judging the similarity of a red apple to a green apple”

• “Detecting a lion’s roar from a distance on a windy day”

• “Learning tree climbing can help rock climbing”

Page 62: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Tasks that can help an animal survive

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Example

• ‘‘Is this a dangerous animal?’’

• “Judging the similarity of a red apple to a green apple”

• “Detecting a lion’s roar from a distance on a windy day”

• “Learning tree climbing can help rock climbing”

Page 63: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Tasks that can help an animal survive

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Example

• ‘‘Is this a dangerous animal?’’

• “Judging the similarity of a red apple to a green apple”

• “Detecting a lion’s roar from a distance on a windy day”

• “Learning tree climbing can help rock climbing”

Page 64: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Tasks that can help an animal survive

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Example

• ‘‘Is this a dangerous animal?’’

• “Judging the similarity of a red apple to a green apple”

• “Detecting a lion’s roar from a distance on a windy day”

• “Learning tree climbing can help rock climbing”

Page 65: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Tasks that can help an animal survive

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Example

• ‘‘Is this a dangerous animal?’’

• “Judging the similarity of a red apple to a green apple”

• “Detecting a lion’s roar from a distance on a windy day”

• “Learning tree climbing can help rock climbing”

Page 66: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Strategies to solve a task

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Page 67: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Strategies to solve a task

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Page 68: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Strategies to solve a task

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Page 69: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Strategies to solve a task

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Page 70: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Strategies to solve a task

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Page 71: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Corresponding ML technics

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Machine learning technique

• ???k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 72: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Corresponding ML technics

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Machine learning technique

• k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 73: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Corresponding ML technics

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Machine learning technique

• k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 74: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Corresponding ML technics

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Machine learning technique

• k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 75: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Corresponding ML technics

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Possible strategy

• Judge similarity to familiar examplesFind a decision boundary based on previous examplesDiscover and exploit structure within collected examplesQuantify output in terms of input

• Preserve pairwise distances

• Allow for variance

• Domain adaptation and transfer of learning

Machine learning technique

• k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 76: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Final mapping

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Task

• Deciding on an appropriate response to a novel stimulus

• Veridical representation

• Dealing with noise and confounding factors

• Generalizing learned skills to new tasks

Machine learning technique

• k-NN with kernel metricSVM, RBF networks

kPCA

Linear Regression, Gaussian processes

• MDS with kernel metric

• Regularization

• Hierarchical mixture models — not kernel based, deep convolutional networks — Implicitly kernel based

Page 77: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Paper outline

• 4 fundamental constrains on any learning system

• Kernels

• Addressing constrains the kernels

• Application to neuroscience

* machine learning * neuroscience

Page 78: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

What we (authors) did?

• We defined 4 foundational constrains on any learning system

• We investigated a notion of kernel

• We mapped learning constraints to the kernel characteristics

• We showed that the kernel-based and similarity-based concepts are used in every-day cognitive agent (learning system) tasks (with example of animal survival)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 79: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

What we (authors) did?

• We defined 4 foundational constrains on any learning system

• We investigated a notion of kernel

• We mapped learning constraints to the kernel characteristics

• We showed that the kernel-based and similarity-based concepts are used in every-day cognitive agent (learning system) tasks (with example of animal survival)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 80: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

What we (authors) did?

• We defined 4 foundational constrains on any learning system

• We investigated a notion of kernel

• We mapped learning constraints to the kernel characteristics

• We showed that the kernel-based and similarity-based concepts are used in every-day cognitive agent (learning system) tasks (with example of animal survival)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 81: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

What we (authors) did?

• We defined 4 foundational constrains on any learning system

• We investigated a notion of kernel

• We mapped learning constraints to the kernel characteristics

• We showed that the kernel-based and similarity-based concepts are used in every-day cognitive agent (learning system) tasks (with example of animal survival)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 82: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

What we (authors) did?

• We defined 4 foundational constrains on any learning system

• We investigated a notion of kernel

• We mapped learning constraints to the kernel characteristics

• We showed that the kernel-based and similarity-based concepts are used in every-day cognitive agent (learning system) tasks (with example of animal survival)

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 83: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Take home neuroscience message

• The concept of kernel is that natural for the animal (human) cognitive behavior so we might assume that the brain uses something similar to kernel.

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 84: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

P.S.:

• https://www.youtube.com/watch?v=3liCbRZPrZA

https://www.quora.com/What-are-Kernels-in-Machine-Learning-and-SVM

Page 85: Similarity, kernels, and the fundamental constraints on ... · Similarity, kernels, and the fundamental constraints on cognition A survey paper Authors: Reza Shahbazi, Rajeev Raizada,

Questions?Thank you for your attention