Brief intro : Invariance and Equivariance
CovNet are translational Equivalent
This demonstrates LeNet-5's invariance to small rotations (+/-40 degrees).
How about Rotation ?
Limitation of Conventional CovNet
2D convolution is equivariant under translation, but not under rotation
Limitation of Conventional CovNet
Invariance
Φ
Image(X)
Feature(Z) Z1 = Z = Z2
𝑇𝑔
1
Mapping
ft’n(Φ(·))
Φ
Transformation
X1 X2
Z = Z1 = Φ(X1) = Z2 = Φ(X2) = Φ(𝑻 𝒈
𝟏
X1 )
: Mapping independent of transformation, 𝑇𝑔, for all 𝑇𝑔
X2 = 𝑇𝑔
1
X1
To make a Convolutional Neural Networks (CNN) transformation-
invariant, data augmentation with training samples is generally used
Invariance
Equivariance
Φ
Image(X)
Feature(Z) Z1 Z2
𝑇𝑔
2
𝑇𝑔
1
Φ
Transformation
X1 X2
Z2 = 𝑻 𝒈
𝟐
Z1 = 𝑻 𝒈
𝟐
Φ(X1) = Φ(𝑻 𝒈
𝟏
X1 )
: Invariance is special case of equivariance where 𝑇𝑔
2 is the identity.
X2 = 𝑇𝑔
1
X1
Z2 = 𝑇𝑔
2
Z1
: Mapping preserves algebraic structure of transformation
Z1 ≠ Z2 but keeps the relationship
Mapping
ft’n(Φ(·))
Equivariance : Group CovNet
To understand the rotation or proportion change of a given entity, a
group of filters(a combination of rotated and mirror reflected versions of
filter) is adopted.
For example, the group p4 which contains translations and rotations by
multiples of ninety degrees, or, which additionally contains mirror
reflections.
: Rotation
: Mirror reflections
A filter in a G-CNN detects co-occurrences of features that have the
preferred relative pose, and can match such a feature constellation in
every global pose through an operation called the G-convolution.
Equivariance : Group CovNet
Filter group 1
Filter group 2
Filter group N
Visualization of classic 2D convolution
Visualization of the G-Conv for the roto-translation group
G-Convolution
Equivariance : Group CovNet
G-convolution is equivariant under rotation
G-Convolution
Equivariance : Group CovNet
Equivariance : Group CovNet
Latent representations learnt by a CNN and a G-CNN.
- The left part is the result of a typical CNN while the right one is that of a G-
CNN.
- In both parts, the outer cycles consist of the rotated images while the inner
cycles consist of the learnt representations.
- Features produced by a G-CNN is equivariant to rotation while that produced
by a typical CNN is not.
What we need : EQUIVARIANCE (not invariance)
“Equivariance makes a CNN understand the rotation or proportion change”
Equivariance : Capsule Net
“A capsule is a group of neurons whose activity vector represents
the instantiation parameters of a specific type of entity such as an
object or an object part.”
Equivariance : Capsule Net
Equivariance of Capsules
“A capsule is a group of neurons whose activity vector represents the
instantiation parameters of a specific type of entity such as an object or
an object part.”
Activity vector map Object
Equivariance : Capsule Net

Brief intro : Invariance and Equivariance

  • 1.
    Brief intro :Invariance and Equivariance
  • 2.
    CovNet are translationalEquivalent This demonstrates LeNet-5's invariance to small rotations (+/-40 degrees). How about Rotation ? Limitation of Conventional CovNet
  • 3.
    2D convolution isequivariant under translation, but not under rotation Limitation of Conventional CovNet
  • 4.
    Invariance Φ Image(X) Feature(Z) Z1 =Z = Z2 𝑇𝑔 1 Mapping ft’n(Φ(·)) Φ Transformation X1 X2 Z = Z1 = Φ(X1) = Z2 = Φ(X2) = Φ(𝑻 𝒈 𝟏 X1 ) : Mapping independent of transformation, 𝑇𝑔, for all 𝑇𝑔 X2 = 𝑇𝑔 1 X1
  • 5.
    To make aConvolutional Neural Networks (CNN) transformation- invariant, data augmentation with training samples is generally used Invariance
  • 6.
    Equivariance Φ Image(X) Feature(Z) Z1 Z2 𝑇𝑔 2 𝑇𝑔 1 Φ Transformation X1X2 Z2 = 𝑻 𝒈 𝟐 Z1 = 𝑻 𝒈 𝟐 Φ(X1) = Φ(𝑻 𝒈 𝟏 X1 ) : Invariance is special case of equivariance where 𝑇𝑔 2 is the identity. X2 = 𝑇𝑔 1 X1 Z2 = 𝑇𝑔 2 Z1 : Mapping preserves algebraic structure of transformation Z1 ≠ Z2 but keeps the relationship Mapping ft’n(Φ(·))
  • 7.
    Equivariance : GroupCovNet To understand the rotation or proportion change of a given entity, a group of filters(a combination of rotated and mirror reflected versions of filter) is adopted. For example, the group p4 which contains translations and rotations by multiples of ninety degrees, or, which additionally contains mirror reflections. : Rotation : Mirror reflections
  • 8.
    A filter ina G-CNN detects co-occurrences of features that have the preferred relative pose, and can match such a feature constellation in every global pose through an operation called the G-convolution. Equivariance : Group CovNet Filter group 1 Filter group 2 Filter group N
  • 9.
    Visualization of classic2D convolution Visualization of the G-Conv for the roto-translation group G-Convolution Equivariance : Group CovNet
  • 10.
    G-convolution is equivariantunder rotation G-Convolution Equivariance : Group CovNet
  • 11.
    Equivariance : GroupCovNet Latent representations learnt by a CNN and a G-CNN. - The left part is the result of a typical CNN while the right one is that of a G- CNN. - In both parts, the outer cycles consist of the rotated images while the inner cycles consist of the learnt representations. - Features produced by a G-CNN is equivariant to rotation while that produced by a typical CNN is not.
  • 12.
    What we need: EQUIVARIANCE (not invariance) “Equivariance makes a CNN understand the rotation or proportion change” Equivariance : Capsule Net
  • 13.
    “A capsule isa group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part.” Equivariance : Capsule Net
  • 14.
    Equivariance of Capsules “Acapsule is a group of neurons whose activity vector represents the instantiation parameters of a specific type of entity such as an object or an object part.” Activity vector map Object Equivariance : Capsule Net