SlideShare a Scribd company logo
Julián Tachella
CNRS
Physics Laboratory
École Normale Supérieure de Lyon
Imaging with Equivariant Deep
Learning
Edinburgh ICMS Workshop 2023
2
based on 2023 IEEE Signal Processing Magazine paper,
with Dongdong Chen, Mike Davies, Matthias Ehrhardt, Carola Schonlieb and Ferdia Sherry
3
Inverse Problems
Goal: recover signal 𝑥 from 𝑦
𝑦 = 𝐴𝑥 + 𝜖
4
Examples
Magnetic resonance imaging
• 𝐴 = subset of Fourier modes
(𝑘 − space) of 2D/3D images
Image inpainting
• 𝐴 = diagonal matrix
with 1’s and 0s.
Computed tomography
• 𝐴 = 1D projections
(sinograms) of 2D image
𝑥
𝑦 𝑦 𝑥
𝑦 𝑥
5
Why It Is Hard to Invert?
Even in the absence of noise, infinitely many 𝑥 consistent with 𝑦:
𝑥 = 𝐴†𝑦 + 𝑣
where 𝐴†
is the pseudo-inverse of 𝐴 and 𝑣 is any vector in nullspace of 𝐴
reconstruct
6
Low-Dimensionality Prior
Idea: Most natural signals sets 𝒳 are low-dimensional
• Mathematically, we use the box-counting dimension
boxdim 𝒳 = 𝑘 ≪ 𝑛
Theorem: A signal 𝑥 belonging to a set 𝒳 ⊂ ℝ𝑛 with
boxdim 𝒳 = 𝑘 can be uniquely recovered from the
measurement 𝑦 = 𝐴𝑥 with almost every 𝐴 ∈ ℝ𝑚×𝑛 if
𝑚 > 2𝑘.
7
Symmetry Prior
Idea: Most natural signals sets 𝒳 are invariant to groups of transformations.
Example: natural images are translation invariant
• Mathematically, a set 𝒳 is invariant to a group
of transformations 𝑇𝑔 ∈ ℝ𝑛×𝑛
𝑔∈𝐺
if
∀𝑥 ∈ 𝒳, ∀𝑔 ∈ 𝐺, 𝑇𝑔𝑥 ∈ 𝒳
Other symmetries: rotations, permutation, amplitude
8
Group Actions
• A group action of 𝐺 acting on ℝ𝑛 is a mapping 𝑇: 𝐺 × ℝ𝑛 ↦ ℝ𝑛 which inherits the group
axioms.
(multiplication) 𝑇𝑔2
∘ 𝑇𝑔1
(𝑥) = 𝑇𝑔2∘𝑔1
𝑥
(inverse) 𝑇𝑔−1 𝑥 = 𝑇𝑔
−1(𝑥)
(identity) 𝑇𝑒 𝑥 = 𝑥
• Let 𝑇, 𝑇 be group actions on ℝ𝑛
and ℝ𝑚
. A mapping 𝑓: ℝ𝑛
↦ ℝ𝑚
is 𝑮-equivariant if
𝑓 ∘ 𝑇𝑔(𝑥) = 𝑇𝑔 ∘ 𝑓(𝑥) for all 𝑔 ∈ 𝐺, 𝑥 ∈ ℝ𝑛
• If 𝑓 and 𝑔 are equivariant, ℎ = 𝑔 ∘ 𝑓 is equivariant
9
Representations
• A linear representation of a compact 𝐺 on ℝ𝑛 is the group action 𝑇: 𝐺 ↦ ℝ𝑛×𝑛 represented
by invertible 𝑛 × 𝑛 matrices.
• A linear representation can be decomposed in irreps {𝜌𝑔
𝑘 ∈ ℝ𝑠𝑘×𝑠𝑘}𝑘=1
𝐾
of dimension 𝒔𝒋 with
multiplicity 𝒄𝒌 and arbitrary basis 𝐵 ∈ ℝ𝑛×𝑛:
𝑇𝑔 = 𝐵−1
𝜌𝑔
1 0
0 𝜌𝑔
1 ⋯ 0
⋮ ⋱ ⋮
0 ⋯
𝜌𝑔
𝐾
0
0 𝜌𝑔
𝐾
𝐵
Example: Shift matrices, 𝐵 is Fourier transform and 𝜌𝑔
𝑘 = 𝑒−
i2𝜋𝑘𝑔
𝑛 (𝑠𝑘 = 𝑐𝑘 = 1) for 𝑘 = 1, … , 𝑛
10
Regularised Reconstruction
Standard regularisation approach:
argmin 𝐴𝑥 − 𝑦
2
+ 𝐽(𝑥)
• 𝐽(𝑥) enforces low-dimensionality and is 𝐺-invariant: 𝐽 𝑇𝑔𝑥 = 𝐽(𝑥)
Examples: total-variation (shift invariant), sparsity (permutation invariant)
Disadvantages: hard to define a good 𝐽 𝑥 in real world problems, loose with
respect to the true 𝒳
𝑥
11
Learning approach
Idea: use training pairs of signals and measurements 𝑥𝑖, 𝑦𝑖 𝑖 to directly learn the
inversion function
argmin
𝑖
𝑥𝑖 − 𝑓𝜃(𝑦𝑖)
2
where 𝑓𝜃: ℝ𝑚 ↦ ℝ𝑛 is a deep neural network with parameters 𝜃.
𝜃
𝑓𝜃
12
Advantages:
• State-of-the-art reconstructions
• Once trained, 𝑓𝜃 is easy to evaluate
Learning Approach
x8 accelerated MRI [Zbontar et al., 2019]
Deep network
(34.5 dB)
Ground-truth
Total variation
(28.2 dB)
13
Tutorial Goals
1. Incorporate symmetry into design of 𝑓𝜃
2. Use symmetry to generalize to unseen poses and noise levels
3. How symmetry enables fully unsupervised learning
14
Equivariant Nets
Convolutional neural networks (CNNs) are translation equivariant
• Going beyond translations: make every block equivariant!
𝑊1 𝜙 𝑊𝐿
… 𝜙 𝑓𝜃(𝑦)
𝑦
15
Equivariant Nets
• Steerable linear layers [Cohen et al., 2016], [Serre, 1971]
𝑆 = {𝑊 ∈ ℝ𝑝×𝑛
𝑇𝑔𝑊 = 𝑊𝑇𝑔, ∀𝑔 ∈ 𝐺
𝑊𝜃 =
𝑖=1
𝑟
𝜃𝑖𝜓𝑖
• Parameter efficiency:
dim 𝑆
dim(ℝ𝑝×𝑛)
=
𝑟
𝑝𝑛
• Non-linearities: elementwise 𝜙 with 𝑇𝑔 represented by permutation matrices.
• In practice, choose 𝑇𝑔 (multiplicities=channels) at each layer and use existing libraries.
ℝ𝑝×𝑛
𝑆
= span(𝜓1, … , 𝜓𝑟)
16
𝐴
𝑓𝜃
Equivariance in Inverse Problems
Role of invariance of 𝒳 when solving inverse problems?
System Equivariance: Let 𝑓𝜃: 𝑦 ↦ 𝑥 be the reconstruction function
𝑓𝜃 ∘ 𝐴 ∘ 𝑇𝑔𝑥 = 𝑇𝑔 ∘ 𝑓𝜃 ∘ 𝐴𝑥 ∀𝑔 ∈ 𝐺, ∀𝑥 ∈ 𝒳
17
Equivariance by Design
𝐴
𝑓𝜃
Case 1:
𝐴 equivariant on ℝ𝑛 + equivariant architecture 𝑓𝜃 = equivariant system 𝑓 ∘ 𝐴
18
𝐴
𝑓𝜃
?
Equivariance by Design
Case 2:
𝐴 not equivariant on ℝ𝑛 + equivariant architecture 𝑓𝜃 ≠ equivariant system 𝑓 ∘ 𝐴
19
Examples
Translation Rotation Permutation Amplitude
Gaussian
Blur
Image
Inpainting
Sparse-view
CT
Accelerated
MRI
Downsampling
(no antialias)
?
?
?
?
?
? ?
?
?
?
?
?
?
?
? ?
? ?
20
Unrolled Networks
Standard regularisation approach:
argmin 𝐴𝑥 − 𝑦
2
+ 𝐽(𝑥)
Solved via proximal gradient descent,
𝑥𝑡+1 = prox𝜂𝐽 𝑥𝑡 − 𝜂𝐴⊤ 𝐴𝑥𝑡 − 𝑦
where prox𝐽 𝑥 = argmin 𝑥 − 𝑥
2
+ 𝐽 𝑥 .
𝑥
𝑥
Proposition: If the regularisation 𝐽(𝑥) is 𝐺-invariant, the proximal
operator is 𝐺-equivariant
prox𝐽 𝑇𝑔𝑥 = 𝑇𝑔prox𝐽 𝑥
E. Celledoni et al., “Equivariant neural networks for inverse problems,” 2021.
21
Unrolled Networks
Idea: Unroll optimisation and replace proximal operators by learnable networks
for 𝑡 = 1, … , 𝑅 iterations
𝑥𝑡+1 = 𝑓𝜃𝑡
𝑥𝑡 − 𝜂𝐴⊤ 𝐴𝑥𝑡 − 𝑦
where 𝑓𝜃𝑡
: ℝ𝑛 ↦ ℝ𝑛 are equivariant learnable networks
𝐴⊤ 𝑓𝜃1
𝜂𝐴⊤
𝑦
𝑓𝜃𝑅
(𝐼 − 𝜂𝐴⊤
𝐴)
…
𝑓𝜃(𝑦)
22
Pose Generalization
E. Celledoni et al., “Equivariant neural networks for inverse problems,” 2021.
Computed tomography
• LIDC-IDRI dataset
• 512 x 512 images
• 50 views
• Supervised learning
23
Noise Generalization
S. Mohan et al., “Robust and interpretable blind image denoising via bias-free convolutional neural networks,” 2019.
Image denoising: 𝑦 = 𝑥 + 𝜖 where 𝜖 ∼ 𝒩 0, 𝜎2
Amplitude invariance: for all 𝑥 ∈ 𝒳, 𝑔 ∈ ℝ+ we have 𝑔𝑥 ∈ 𝒳.
Equivariant relu network: 𝑓𝜃 𝑔𝑦 = 𝑔𝑓𝜃(𝑦) by removing all biases.
𝑓𝜃 𝑔𝑥 + 𝜖 = 𝑔𝑓𝜃 𝑥 +
𝜖
𝑔
= 𝑔𝑓𝜃 𝑥 + 𝜖′
where 𝜖′
∼ 𝒩 0,
𝜎
𝑔
2
is an unseen noise distribution!
24
Noise Generalization
25
Unsupervised Learning
with Equivariance
26
Limits to Supervised Learning
Main disadvantage: Obtaining training signals 𝑥𝑖 can be expensive or impossible.
• Medical and scientific imaging
• Only solves inverse problems which we already know what to expect
• Risk of training with signals from a different distribution
train test?
27
Learning from only measurements 𝒚?
argmin
𝑖
𝑦𝑖 − 𝐴𝑓𝜃(𝑦𝑖)
2
Learning Approach
𝒇𝜽
𝐴
𝜃
Proposition: 𝑓𝜃 𝑦 = 𝐴†
𝑦 + 𝑔𝜃(𝑦) where 𝑔𝜃: ℝ𝑚
↦ 𝒩
𝐴 is any function whose
image belongs to the nullspace of 𝐴 obtains zero training error.
28
Exploiting Invariance
How to learn from only 𝒚? We need some prior information
For all 𝑔 ∈ 𝐺 we have
𝑦 = 𝐴𝑥 = 𝐴𝑇𝑔𝑇𝑔
−1
𝑥
• Implicit access to multiple operators 𝐴𝑔
• Each operator with different nullspace
= 𝐴𝑔𝑥′
29
Model Identification
Can we uniquely identify the set of signals 𝒳 ⊂ ℝ𝑛 from
the observed measurement sets 𝒴𝑔 = 𝐴𝑇𝑔𝒳
𝑔∈𝐺
?
Theorem [T., Chen and Davies ’22]:
Identifying 𝒳 requires that 𝐴 is not equivariant: 𝐴𝑇𝑔 ≠ 𝑇𝑔𝐴
30
Necessary Conditions
Proposition [T., Chen and Davies ’22]: Identifying 𝒳 from {𝒴𝑔 = 𝐴𝑔𝒳}
possible only if
rank
𝐴𝑇1
⋮
𝐴𝑇|𝐺|
= 𝑛,
and thus, if 𝑚 ≥ max
𝑐𝑗
𝑠𝑗
≥
𝑛
𝐺
where 𝑠𝑗 and 𝑐𝑗 are dimension and
multiplicity of irreps.
31
Can We Learn Any Set?
• Necessary conditions are independent of 𝒳…
• For finite |𝐺|, some signal distributions cannot be identified if 𝐴 is rank-deficient:
𝔼𝑦{𝑒i𝑤⊤𝐴𝑔
†
𝑦
} = 𝔼𝑥 𝑒i𝑥⊤(𝐴𝑔
†
𝐴𝑔𝑤)
= 𝜑(𝐴𝑔
†
𝐴𝑔𝑤)
• We only observe projection of characteristic function 𝜑 to certain subspaces!
Theorem [T., Chen and Davies ’22]: Let 𝐺 be a compact cyclic group. Identifying
set 𝒳 with box-counting dimension 𝑘 from 𝒴𝑔 = 𝐴𝑇𝑔𝒳
𝑔
is possible by almost
every 𝐴 ∈ ℝ𝑚×𝑛 with
𝑚 > 2𝑘 + max
𝑗
𝑐𝑗 + 1
where 𝑐𝑗 are the multiplicities of the representation and max
𝑗
𝑐𝑗 ≥ 𝑛/ 𝐺 .
32
Sufficient Conditions
Additional assumption: The signal set is low-dimensional.
If group is big, then 𝑚 > 2𝑘 + 2 are sufficient = same condition as signal recovery!
33
Consequences
Magnetic resonance imaging
• 𝐴 = subset of Fourier modes
• Equivariant to translations
• Not equivariant to rotations,
which have max 𝑐𝑗 ≈ √𝑛
𝑚 > 2𝑘 + 𝑛 + 1
Image inpainting
• 𝐴 = diagonal matrix with 1’s
and 0s.
• Not equivariant to
translations, which have
max 𝑐𝑗 ≈ 1
𝑚 > 2𝑘 + 2
Computed tomography
• 𝐴 = 1D projections
(sinograms)
• Equivariant to translations
• Not equivariant to rotations,
which have max 𝑐𝑗 ≈ √𝑛
𝑚 > 2𝑘 + 𝑛 + 1
𝑥
𝑦 𝑦 𝑥
𝑦 𝑥
34
Equivariant Imaging
Enforcing equivariance in practice?
• Unrolled equivariant networks might not achieve equivariance of 𝑓𝜃 ∘ 𝐴
• Equivariant prox is not sufficient
Example: Learned prox𝐽 𝑥 = 𝑥 is equivariant, resulting 𝑓𝜃 𝑦 = 𝐴†𝑦 is
measurement consistent but not system equivariant
Idea: enforce equivariance during training!
35
Self-supervised training loss:
argmin ℒ𝑀𝐶 𝜃
• ℒ𝑀𝐶 𝜃 = 𝑖 𝑦𝑖 − 𝐴𝑓𝜃 𝑦𝑖
2
enforces measurement consistency
• ℒ𝐸𝐼 𝜃 = 𝑖,𝑔 𝑓𝜃 𝐴𝑇𝑔𝑓𝜃 𝑦𝑖 − 𝑇𝑔𝑓𝜃 𝑦𝑖
2
enforces equivariance of 𝐴 ∘ 𝑓𝜃
𝜃
• Network-agnostic: applicable to any 𝑓𝜃 (including unrolled)
+ℒ𝐸𝐼 𝜃
Equivariant Imaging
36
Robust EI: SURE+EI
Robust Equivariant Imaging
argmin ℒ𝑆𝑈𝑅𝐸 𝜃 + ℒ𝐸𝐼 𝜃
where ℒ𝑆𝑈𝑅𝐸 𝜃 = 𝑖 𝑦𝑖 − 𝐴𝑓𝜃 𝑦𝑖
2
− 𝜎2𝑚 + 2𝜎2div(𝐴 ∘ 𝑓𝜃)(𝑦𝑖)
𝜃
Theorem [Stein, 1981] Under mild differentiability conditions on the function 𝐴 ∘ 𝑓𝜃, the following
holds
𝔼𝑦 ℒ𝑀𝐶 𝜃 = 𝔼𝑦 ℒ𝑆𝑈𝑅𝐸 𝜃
• Similar expressions of ℒ𝑆𝑈𝑅𝐸 𝜃 for Poisson, Poisson-Gaussian, etc. exist
37
Experiments
Tasks:
• Magnetic resonance imaging
Network
• 𝑓𝜃 = 𝑔𝜃 ∘ 𝐴†
where 𝑔𝜃 is a U-Net
Comparison
• Pseudo-inverse 𝐴†
𝑦𝑖 (no training)
• Meas. consistency 𝐴𝑓𝜃 𝑦𝑖 = 𝑦𝑖
• Fully supervised loss: 𝑓𝜃 𝑦𝑖 = 𝑥𝑖
• Equivariant imaging (unsupervised)
𝐴𝑓𝜃 𝑦𝑖 = 𝑦𝑖 and equivariant 𝐴 ∘ 𝑓𝜃
Equivariant imaging Fully supervised
𝐴†𝑦 Meas. consistency
38
Magnetic resonance imaging
• Operator 𝐴 is a subset of Fourier measurements (x2 downsampling)
• Dataset is approximately rotation invariant
Signal 𝑥 Measurements 𝑦
39
Inpainting
Signal 𝑥
Measurements 𝑦
• Operator 𝐴 is an inpainting mask (30% pixels dropped)
• Poisson noise (rate=10)
• Dataset is approximately translation invariant
Supervised Meas. consistency Robust EI
40
Noisy
measurements 𝑦
Robust EI
Supervised
Clean signal 𝑥 Meas. consistency
Computed Tomography
• Operator 𝐴 is (non-linear variant) sparse radon transform (50 views)
• Mixed Poisson-Gaussian noise
• Dataset is approximately rotation invariant
41
Compressed Sensing
• 𝐴 is a random iid Gaussian matrix
• Shift invariant (max𝑗 𝑐𝑗 = 1) MNIST dataset 𝑘 ≈ 12
42
Single-pixel camera
43
Conclusions
• Learn signal set from data
• Equivariance by design
• Generalize unseen poses and
noise level
• No ground-truth references needed
• Necessary and sufficient conditions
45
Conclusions
Future work/open problems
• Non-linear inverse problems
• Semi-group actions
• Approx. low-dimensional models
• Beyond Euclidean domains
46
Extras
• Theoretical bounds and algorithms can be extended to the case where we observe
data through multiple operators
𝐴1𝑥 𝐴2𝑥 𝐴3𝑥
“Unsupervised Learning From Incomplete Measurements for Inverse Problems”, Tachella, Chen and
Davies, NeurIPS 2022.
47
Papers
[1] “Equivariant Imaging: Learning Beyond the Range Space”, Chen, Tachella and
Davies, ICCV 2021 (Oral)
[2] “Robust Equivariant Imaging: a fully unsupervised framework for learning to
image from noisy and partial measurements”, Chen, Tachella and Davies, CVPR
[3] “Unsupervised Learning From Incomplete Measurements for Inverse Problems”,
Tachella, Chen and Davies, NeurIPS 2022.
[4] “Sensing Theorems for Unsupervised Learning in Inverse Problems”, Tachella,
Chen and Davies, JMLR 2023.
[5] “Imaging with Equivariant Deep Learning”, Chen, Davies, Eerhardt, Schonlieb,
Ferry and Tachella, IEEE SPM 2023.
Thanks for your attention!
Tachella.github.io
 Codes
 Presentations
 … and more
48

More Related Content

What's hot

An overview of gradient descent optimization algorithms
An overview of gradient descent optimization algorithms An overview of gradient descent optimization algorithms
An overview of gradient descent optimization algorithms
Hakky St
 
Deep Learning Theory Seminar (Chap 1-2, part 1)
Deep Learning Theory Seminar (Chap 1-2, part 1)Deep Learning Theory Seminar (Chap 1-2, part 1)
Deep Learning Theory Seminar (Chap 1-2, part 1)
Sangwoo Mo
 
Wasserstein GAN 수학 이해하기 I
Wasserstein GAN 수학 이해하기 IWasserstein GAN 수학 이해하기 I
Wasserstein GAN 수학 이해하기 I
Sungbin Lim
 
Pixel RNN to Pixel CNN++
Pixel RNN to Pixel CNN++Pixel RNN to Pixel CNN++
Pixel RNN to Pixel CNN++
Dongheon Lee
 
Neural Networks: Support Vector machines
Neural Networks: Support Vector machinesNeural Networks: Support Vector machines
Neural Networks: Support Vector machines
Mostafa G. M. Mostafa
 
Regularization in deep learning
Regularization in deep learningRegularization in deep learning
Regularization in deep learning
Kien Le
 
PAC Bayesian for Deep Learning
PAC Bayesian for Deep LearningPAC Bayesian for Deep Learning
PAC Bayesian for Deep Learning
Mark Chang
 
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
Simplilearn
 
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
Ono Shigeru
 
Perceptron (neural network)
Perceptron (neural network)Perceptron (neural network)
Perceptron (neural network)
EdutechLearners
 
Optimization in Deep Learning
Optimization in Deep LearningOptimization in Deep Learning
Optimization in Deep Learning
Yan Xu
 
What is the Expectation Maximization (EM) Algorithm?
What is the Expectation Maximization (EM) Algorithm?What is the Expectation Maximization (EM) Algorithm?
What is the Expectation Maximization (EM) Algorithm?
Kazuki Yoshida
 
Naive Bayes
Naive BayesNaive Bayes
Naive Bayes
CloudxLab
 
Gaussian Process Regression
Gaussian Process Regression  Gaussian Process Regression
Gaussian Process Regression
SEMINARGROOT
 
Exploring Simple Siamese Representation Learning
Exploring Simple Siamese Representation LearningExploring Simple Siamese Representation Learning
Exploring Simple Siamese Representation Learning
Sungchul Kim
 
The neural tangent link between CNN denoisers and non-local filters
The neural tangent link between CNN denoisers and non-local filtersThe neural tangent link between CNN denoisers and non-local filters
The neural tangent link between CNN denoisers and non-local filters
Julián Tachella
 
Sharpness-Aware Minimization for Efficiently Improving Generalization
Sharpness-Aware Minimization for Efficiently Improving GeneralizationSharpness-Aware Minimization for Efficiently Improving Generalization
Sharpness-Aware Minimization for Efficiently Improving Generalization
taeseon ryu
 
Variational Autoencoder
Variational AutoencoderVariational Autoencoder
Variational Autoencoder
Mark Chang
 
EMアルゴリズム
EMアルゴリズムEMアルゴリズム
Gradient descent method
Gradient descent methodGradient descent method
Gradient descent method
Prof. Neeta Awasthy
 

What's hot (20)

An overview of gradient descent optimization algorithms
An overview of gradient descent optimization algorithms An overview of gradient descent optimization algorithms
An overview of gradient descent optimization algorithms
 
Deep Learning Theory Seminar (Chap 1-2, part 1)
Deep Learning Theory Seminar (Chap 1-2, part 1)Deep Learning Theory Seminar (Chap 1-2, part 1)
Deep Learning Theory Seminar (Chap 1-2, part 1)
 
Wasserstein GAN 수학 이해하기 I
Wasserstein GAN 수학 이해하기 IWasserstein GAN 수학 이해하기 I
Wasserstein GAN 수학 이해하기 I
 
Pixel RNN to Pixel CNN++
Pixel RNN to Pixel CNN++Pixel RNN to Pixel CNN++
Pixel RNN to Pixel CNN++
 
Neural Networks: Support Vector machines
Neural Networks: Support Vector machinesNeural Networks: Support Vector machines
Neural Networks: Support Vector machines
 
Regularization in deep learning
Regularization in deep learningRegularization in deep learning
Regularization in deep learning
 
PAC Bayesian for Deep Learning
PAC Bayesian for Deep LearningPAC Bayesian for Deep Learning
PAC Bayesian for Deep Learning
 
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
Backpropagation And Gradient Descent In Neural Networks | Neural Network Tuto...
 
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
Goodfellow, Bengio, Couville (2016) "Deep Learning", Chap. 7
 
Perceptron (neural network)
Perceptron (neural network)Perceptron (neural network)
Perceptron (neural network)
 
Optimization in Deep Learning
Optimization in Deep LearningOptimization in Deep Learning
Optimization in Deep Learning
 
What is the Expectation Maximization (EM) Algorithm?
What is the Expectation Maximization (EM) Algorithm?What is the Expectation Maximization (EM) Algorithm?
What is the Expectation Maximization (EM) Algorithm?
 
Naive Bayes
Naive BayesNaive Bayes
Naive Bayes
 
Gaussian Process Regression
Gaussian Process Regression  Gaussian Process Regression
Gaussian Process Regression
 
Exploring Simple Siamese Representation Learning
Exploring Simple Siamese Representation LearningExploring Simple Siamese Representation Learning
Exploring Simple Siamese Representation Learning
 
The neural tangent link between CNN denoisers and non-local filters
The neural tangent link between CNN denoisers and non-local filtersThe neural tangent link between CNN denoisers and non-local filters
The neural tangent link between CNN denoisers and non-local filters
 
Sharpness-Aware Minimization for Efficiently Improving Generalization
Sharpness-Aware Minimization for Efficiently Improving GeneralizationSharpness-Aware Minimization for Efficiently Improving Generalization
Sharpness-Aware Minimization for Efficiently Improving Generalization
 
Variational Autoencoder
Variational AutoencoderVariational Autoencoder
Variational Autoencoder
 
EMアルゴリズム
EMアルゴリズムEMアルゴリズム
EMアルゴリズム
 
Gradient descent method
Gradient descent methodGradient descent method
Gradient descent method
 

Similar to Tutorial Equivariance in Imaging ICMS 23.pptx

Equivariant Imaging SELW'22
Equivariant Imaging SELW'22Equivariant Imaging SELW'22
Equivariant Imaging SELW'22
Julián Tachella
 
NeurIPS22.pptx
NeurIPS22.pptxNeurIPS22.pptx
NeurIPS22.pptx
Julián Tachella
 
Deep learning paper review ppt sourece -Direct clr
Deep learning paper review ppt sourece -Direct clr Deep learning paper review ppt sourece -Direct clr
Deep learning paper review ppt sourece -Direct clr
taeseon ryu
 
Optimum Engineering Design - Day 2b. Classical Optimization methods
Optimum Engineering Design - Day 2b. Classical Optimization methodsOptimum Engineering Design - Day 2b. Classical Optimization methods
Optimum Engineering Design - Day 2b. Classical Optimization methods
SantiagoGarridoBulln
 
Neural Networks
Neural NetworksNeural Networks
Lecture_9_LA_Review.pptx
Lecture_9_LA_Review.pptxLecture_9_LA_Review.pptx
Lecture_9_LA_Review.pptx
Sunny432360
 
The world of loss function
The world of loss functionThe world of loss function
The world of loss function
홍배 김
 
SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15
Hao Zhuang
 
Learning a nonlinear embedding by preserving class neibourhood structure 최종
Learning a nonlinear embedding by preserving class neibourhood structure   최종Learning a nonlinear embedding by preserving class neibourhood structure   최종
Learning a nonlinear embedding by preserving class neibourhood structure 최종
WooSung Choi
 
Deep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defenseDeep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defense
Wei Yang
 
04 Multi-layer Feedforward Networks
04 Multi-layer Feedforward Networks04 Multi-layer Feedforward Networks
04 Multi-layer Feedforward Networks
Tamer Ahmed Farrag, PhD
 
Support Vector Machine.pptx
Support Vector Machine.pptxSupport Vector Machine.pptx
Support Vector Machine.pptx
HarishNayak44
 
Elements of Statistical Learning 読み会 第2章
Elements of Statistical Learning 読み会 第2章Elements of Statistical Learning 読み会 第2章
Elements of Statistical Learning 読み会 第2章
Tsuyoshi Sakama
 
20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx
ssuser7807522
 
BEGAN Boundary Equilibrium Generative Adversarial Networks
BEGAN Boundary Equilibrium Generative Adversarial NetworksBEGAN Boundary Equilibrium Generative Adversarial Networks
BEGAN Boundary Equilibrium Generative Adversarial Networks
Chung-Il Kim
 
Paper study: Learning to solve circuit sat
Paper study: Learning to solve circuit satPaper study: Learning to solve circuit sat
Paper study: Learning to solve circuit sat
ChenYiHuang5
 
Optimization of positive linear systems via geometric programming
Optimization of positive linear systems via geometric programmingOptimization of positive linear systems via geometric programming
Optimization of positive linear systems via geometric programming
Masaki Ogura
 
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Tatsunori Taniai
 
Super resolution in deep learning era - Jaejun Yoo
Super resolution in deep learning era - Jaejun YooSuper resolution in deep learning era - Jaejun Yoo
Super resolution in deep learning era - Jaejun Yoo
JaeJun Yoo
 
Lecture_10_SVD.pptx
Lecture_10_SVD.pptxLecture_10_SVD.pptx
Lecture_10_SVD.pptx
avulodttlrjhycmbmc
 

Similar to Tutorial Equivariance in Imaging ICMS 23.pptx (20)

Equivariant Imaging SELW'22
Equivariant Imaging SELW'22Equivariant Imaging SELW'22
Equivariant Imaging SELW'22
 
NeurIPS22.pptx
NeurIPS22.pptxNeurIPS22.pptx
NeurIPS22.pptx
 
Deep learning paper review ppt sourece -Direct clr
Deep learning paper review ppt sourece -Direct clr Deep learning paper review ppt sourece -Direct clr
Deep learning paper review ppt sourece -Direct clr
 
Optimum Engineering Design - Day 2b. Classical Optimization methods
Optimum Engineering Design - Day 2b. Classical Optimization methodsOptimum Engineering Design - Day 2b. Classical Optimization methods
Optimum Engineering Design - Day 2b. Classical Optimization methods
 
Neural Networks
Neural NetworksNeural Networks
Neural Networks
 
Lecture_9_LA_Review.pptx
Lecture_9_LA_Review.pptxLecture_9_LA_Review.pptx
Lecture_9_LA_Review.pptx
 
The world of loss function
The world of loss functionThe world of loss function
The world of loss function
 
SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15
 
Learning a nonlinear embedding by preserving class neibourhood structure 최종
Learning a nonlinear embedding by preserving class neibourhood structure   최종Learning a nonlinear embedding by preserving class neibourhood structure   최종
Learning a nonlinear embedding by preserving class neibourhood structure 최종
 
Deep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defenseDeep learning-for-pose-estimation-wyang-defense
Deep learning-for-pose-estimation-wyang-defense
 
04 Multi-layer Feedforward Networks
04 Multi-layer Feedforward Networks04 Multi-layer Feedforward Networks
04 Multi-layer Feedforward Networks
 
Support Vector Machine.pptx
Support Vector Machine.pptxSupport Vector Machine.pptx
Support Vector Machine.pptx
 
Elements of Statistical Learning 読み会 第2章
Elements of Statistical Learning 読み会 第2章Elements of Statistical Learning 読み会 第2章
Elements of Statistical Learning 読み会 第2章
 
20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx20230213_ComputerVision_연구.pptx
20230213_ComputerVision_연구.pptx
 
BEGAN Boundary Equilibrium Generative Adversarial Networks
BEGAN Boundary Equilibrium Generative Adversarial NetworksBEGAN Boundary Equilibrium Generative Adversarial Networks
BEGAN Boundary Equilibrium Generative Adversarial Networks
 
Paper study: Learning to solve circuit sat
Paper study: Learning to solve circuit satPaper study: Learning to solve circuit sat
Paper study: Learning to solve circuit sat
 
Optimization of positive linear systems via geometric programming
Optimization of positive linear systems via geometric programmingOptimization of positive linear systems via geometric programming
Optimization of positive linear systems via geometric programming
 
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
Neural Inverse Rendering for General Reflectance Photometric Stereo (ICML 2018)
 
Super resolution in deep learning era - Jaejun Yoo
Super resolution in deep learning era - Jaejun YooSuper resolution in deep learning era - Jaejun Yoo
Super resolution in deep learning era - Jaejun Yoo
 
Lecture_10_SVD.pptx
Lecture_10_SVD.pptxLecture_10_SVD.pptx
Lecture_10_SVD.pptx
 

More from Julián Tachella

The role of overparameterization and optimization in CNN denoisers
The role of overparameterization and optimization in CNN denoisersThe role of overparameterization and optimization in CNN denoisers
The role of overparameterization and optimization in CNN denoisers
Julián Tachella
 
Thesis presentation
Thesis presentationThesis presentation
Thesis presentation
Julián Tachella
 
CAMSAP19
CAMSAP19CAMSAP19
EUSIPCO19
EUSIPCO19EUSIPCO19
EUSIPCO19
Julián Tachella
 
ICASSP19
ICASSP19ICASSP19
Bayesian restoration of high-dimensional photon-starved images
Bayesian restoration of high-dimensional photon-starved imagesBayesian restoration of high-dimensional photon-starved images
Bayesian restoration of high-dimensional photon-starved images
Julián Tachella
 

More from Julián Tachella (6)

The role of overparameterization and optimization in CNN denoisers
The role of overparameterization and optimization in CNN denoisersThe role of overparameterization and optimization in CNN denoisers
The role of overparameterization and optimization in CNN denoisers
 
Thesis presentation
Thesis presentationThesis presentation
Thesis presentation
 
CAMSAP19
CAMSAP19CAMSAP19
CAMSAP19
 
EUSIPCO19
EUSIPCO19EUSIPCO19
EUSIPCO19
 
ICASSP19
ICASSP19ICASSP19
ICASSP19
 
Bayesian restoration of high-dimensional photon-starved images
Bayesian restoration of high-dimensional photon-starved imagesBayesian restoration of high-dimensional photon-starved images
Bayesian restoration of high-dimensional photon-starved images
 

Recently uploaded

CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptxCFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
R&R Consult
 
ASME IX(9) 2007 Full Version .pdf
ASME IX(9)  2007 Full Version       .pdfASME IX(9)  2007 Full Version       .pdf
ASME IX(9) 2007 Full Version .pdf
AhmedHussein950959
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
obonagu
 
WATER CRISIS and its solutions-pptx 1234
WATER CRISIS and its solutions-pptx 1234WATER CRISIS and its solutions-pptx 1234
WATER CRISIS and its solutions-pptx 1234
AafreenAbuthahir2
 
Architectural Portfolio Sean Lockwood
Architectural Portfolio Sean LockwoodArchitectural Portfolio Sean Lockwood
Architectural Portfolio Sean Lockwood
seandesed
 
H.Seo, ICLR 2024, MLILAB, KAIST AI.pdf
H.Seo,  ICLR 2024, MLILAB,  KAIST AI.pdfH.Seo,  ICLR 2024, MLILAB,  KAIST AI.pdf
H.Seo, ICLR 2024, MLILAB, KAIST AI.pdf
MLILAB
 
ML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptxML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptx
Vijay Dialani, PhD
 
Cosmetic shop management system project report.pdf
Cosmetic shop management system project report.pdfCosmetic shop management system project report.pdf
Cosmetic shop management system project report.pdf
Kamal Acharya
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
Kamal Acharya
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
gerogepatton
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
Kamal Acharya
 
block diagram and signal flow graph representation
block diagram and signal flow graph representationblock diagram and signal flow graph representation
block diagram and signal flow graph representation
Divya Somashekar
 
DESIGN A COTTON SEED SEPARATION MACHINE.docx
DESIGN A COTTON SEED SEPARATION MACHINE.docxDESIGN A COTTON SEED SEPARATION MACHINE.docx
DESIGN A COTTON SEED SEPARATION MACHINE.docx
FluxPrime1
 
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Dr.Costas Sachpazis
 
ethical hacking-mobile hacking methods.ppt
ethical hacking-mobile hacking methods.pptethical hacking-mobile hacking methods.ppt
ethical hacking-mobile hacking methods.ppt
Jayaprasanna4
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
WENKENLI1
 
ethical hacking in wireless-hacking1.ppt
ethical hacking in wireless-hacking1.pptethical hacking in wireless-hacking1.ppt
ethical hacking in wireless-hacking1.ppt
Jayaprasanna4
 
Hierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power SystemHierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power System
Kerry Sado
 
Investor-Presentation-Q1FY2024 investor presentation document.pptx
Investor-Presentation-Q1FY2024 investor presentation document.pptxInvestor-Presentation-Q1FY2024 investor presentation document.pptx
Investor-Presentation-Q1FY2024 investor presentation document.pptx
AmarGB2
 
Runway Orientation Based on the Wind Rose Diagram.pptx
Runway Orientation Based on the Wind Rose Diagram.pptxRunway Orientation Based on the Wind Rose Diagram.pptx
Runway Orientation Based on the Wind Rose Diagram.pptx
SupreethSP4
 

Recently uploaded (20)

CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptxCFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptx
 
ASME IX(9) 2007 Full Version .pdf
ASME IX(9)  2007 Full Version       .pdfASME IX(9)  2007 Full Version       .pdf
ASME IX(9) 2007 Full Version .pdf
 
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
在线办理(ANU毕业证书)澳洲国立大学毕业证录取通知书一模一样
 
WATER CRISIS and its solutions-pptx 1234
WATER CRISIS and its solutions-pptx 1234WATER CRISIS and its solutions-pptx 1234
WATER CRISIS and its solutions-pptx 1234
 
Architectural Portfolio Sean Lockwood
Architectural Portfolio Sean LockwoodArchitectural Portfolio Sean Lockwood
Architectural Portfolio Sean Lockwood
 
H.Seo, ICLR 2024, MLILAB, KAIST AI.pdf
H.Seo,  ICLR 2024, MLILAB,  KAIST AI.pdfH.Seo,  ICLR 2024, MLILAB,  KAIST AI.pdf
H.Seo, ICLR 2024, MLILAB, KAIST AI.pdf
 
ML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptxML for identifying fraud using open blockchain data.pptx
ML for identifying fraud using open blockchain data.pptx
 
Cosmetic shop management system project report.pdf
Cosmetic shop management system project report.pdfCosmetic shop management system project report.pdf
Cosmetic shop management system project report.pdf
 
Final project report on grocery store management system..pdf
Final project report on grocery store management system..pdfFinal project report on grocery store management system..pdf
Final project report on grocery store management system..pdf
 
Immunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary AttacksImmunizing Image Classifiers Against Localized Adversary Attacks
Immunizing Image Classifiers Against Localized Adversary Attacks
 
Student information management system project report ii.pdf
Student information management system project report ii.pdfStudent information management system project report ii.pdf
Student information management system project report ii.pdf
 
block diagram and signal flow graph representation
block diagram and signal flow graph representationblock diagram and signal flow graph representation
block diagram and signal flow graph representation
 
DESIGN A COTTON SEED SEPARATION MACHINE.docx
DESIGN A COTTON SEED SEPARATION MACHINE.docxDESIGN A COTTON SEED SEPARATION MACHINE.docx
DESIGN A COTTON SEED SEPARATION MACHINE.docx
 
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...
 
ethical hacking-mobile hacking methods.ppt
ethical hacking-mobile hacking methods.pptethical hacking-mobile hacking methods.ppt
ethical hacking-mobile hacking methods.ppt
 
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdfGoverning Equations for Fundamental Aerodynamics_Anderson2010.pdf
Governing Equations for Fundamental Aerodynamics_Anderson2010.pdf
 
ethical hacking in wireless-hacking1.ppt
ethical hacking in wireless-hacking1.pptethical hacking in wireless-hacking1.ppt
ethical hacking in wireless-hacking1.ppt
 
Hierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power SystemHierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power System
 
Investor-Presentation-Q1FY2024 investor presentation document.pptx
Investor-Presentation-Q1FY2024 investor presentation document.pptxInvestor-Presentation-Q1FY2024 investor presentation document.pptx
Investor-Presentation-Q1FY2024 investor presentation document.pptx
 
Runway Orientation Based on the Wind Rose Diagram.pptx
Runway Orientation Based on the Wind Rose Diagram.pptxRunway Orientation Based on the Wind Rose Diagram.pptx
Runway Orientation Based on the Wind Rose Diagram.pptx
 

Tutorial Equivariance in Imaging ICMS 23.pptx

  • 1. Julián Tachella CNRS Physics Laboratory École Normale Supérieure de Lyon Imaging with Equivariant Deep Learning Edinburgh ICMS Workshop 2023
  • 2. 2 based on 2023 IEEE Signal Processing Magazine paper, with Dongdong Chen, Mike Davies, Matthias Ehrhardt, Carola Schonlieb and Ferdia Sherry
  • 3. 3 Inverse Problems Goal: recover signal 𝑥 from 𝑦 𝑦 = 𝐴𝑥 + 𝜖
  • 4. 4 Examples Magnetic resonance imaging • 𝐴 = subset of Fourier modes (𝑘 − space) of 2D/3D images Image inpainting • 𝐴 = diagonal matrix with 1’s and 0s. Computed tomography • 𝐴 = 1D projections (sinograms) of 2D image 𝑥 𝑦 𝑦 𝑥 𝑦 𝑥
  • 5. 5 Why It Is Hard to Invert? Even in the absence of noise, infinitely many 𝑥 consistent with 𝑦: 𝑥 = 𝐴†𝑦 + 𝑣 where 𝐴† is the pseudo-inverse of 𝐴 and 𝑣 is any vector in nullspace of 𝐴 reconstruct
  • 6. 6 Low-Dimensionality Prior Idea: Most natural signals sets 𝒳 are low-dimensional • Mathematically, we use the box-counting dimension boxdim 𝒳 = 𝑘 ≪ 𝑛 Theorem: A signal 𝑥 belonging to a set 𝒳 ⊂ ℝ𝑛 with boxdim 𝒳 = 𝑘 can be uniquely recovered from the measurement 𝑦 = 𝐴𝑥 with almost every 𝐴 ∈ ℝ𝑚×𝑛 if 𝑚 > 2𝑘.
  • 7. 7 Symmetry Prior Idea: Most natural signals sets 𝒳 are invariant to groups of transformations. Example: natural images are translation invariant • Mathematically, a set 𝒳 is invariant to a group of transformations 𝑇𝑔 ∈ ℝ𝑛×𝑛 𝑔∈𝐺 if ∀𝑥 ∈ 𝒳, ∀𝑔 ∈ 𝐺, 𝑇𝑔𝑥 ∈ 𝒳 Other symmetries: rotations, permutation, amplitude
  • 8. 8 Group Actions • A group action of 𝐺 acting on ℝ𝑛 is a mapping 𝑇: 𝐺 × ℝ𝑛 ↦ ℝ𝑛 which inherits the group axioms. (multiplication) 𝑇𝑔2 ∘ 𝑇𝑔1 (𝑥) = 𝑇𝑔2∘𝑔1 𝑥 (inverse) 𝑇𝑔−1 𝑥 = 𝑇𝑔 −1(𝑥) (identity) 𝑇𝑒 𝑥 = 𝑥 • Let 𝑇, 𝑇 be group actions on ℝ𝑛 and ℝ𝑚 . A mapping 𝑓: ℝ𝑛 ↦ ℝ𝑚 is 𝑮-equivariant if 𝑓 ∘ 𝑇𝑔(𝑥) = 𝑇𝑔 ∘ 𝑓(𝑥) for all 𝑔 ∈ 𝐺, 𝑥 ∈ ℝ𝑛 • If 𝑓 and 𝑔 are equivariant, ℎ = 𝑔 ∘ 𝑓 is equivariant
  • 9. 9 Representations • A linear representation of a compact 𝐺 on ℝ𝑛 is the group action 𝑇: 𝐺 ↦ ℝ𝑛×𝑛 represented by invertible 𝑛 × 𝑛 matrices. • A linear representation can be decomposed in irreps {𝜌𝑔 𝑘 ∈ ℝ𝑠𝑘×𝑠𝑘}𝑘=1 𝐾 of dimension 𝒔𝒋 with multiplicity 𝒄𝒌 and arbitrary basis 𝐵 ∈ ℝ𝑛×𝑛: 𝑇𝑔 = 𝐵−1 𝜌𝑔 1 0 0 𝜌𝑔 1 ⋯ 0 ⋮ ⋱ ⋮ 0 ⋯ 𝜌𝑔 𝐾 0 0 𝜌𝑔 𝐾 𝐵 Example: Shift matrices, 𝐵 is Fourier transform and 𝜌𝑔 𝑘 = 𝑒− i2𝜋𝑘𝑔 𝑛 (𝑠𝑘 = 𝑐𝑘 = 1) for 𝑘 = 1, … , 𝑛
  • 10. 10 Regularised Reconstruction Standard regularisation approach: argmin 𝐴𝑥 − 𝑦 2 + 𝐽(𝑥) • 𝐽(𝑥) enforces low-dimensionality and is 𝐺-invariant: 𝐽 𝑇𝑔𝑥 = 𝐽(𝑥) Examples: total-variation (shift invariant), sparsity (permutation invariant) Disadvantages: hard to define a good 𝐽 𝑥 in real world problems, loose with respect to the true 𝒳 𝑥
  • 11. 11 Learning approach Idea: use training pairs of signals and measurements 𝑥𝑖, 𝑦𝑖 𝑖 to directly learn the inversion function argmin 𝑖 𝑥𝑖 − 𝑓𝜃(𝑦𝑖) 2 where 𝑓𝜃: ℝ𝑚 ↦ ℝ𝑛 is a deep neural network with parameters 𝜃. 𝜃 𝑓𝜃
  • 12. 12 Advantages: • State-of-the-art reconstructions • Once trained, 𝑓𝜃 is easy to evaluate Learning Approach x8 accelerated MRI [Zbontar et al., 2019] Deep network (34.5 dB) Ground-truth Total variation (28.2 dB)
  • 13. 13 Tutorial Goals 1. Incorporate symmetry into design of 𝑓𝜃 2. Use symmetry to generalize to unseen poses and noise levels 3. How symmetry enables fully unsupervised learning
  • 14. 14 Equivariant Nets Convolutional neural networks (CNNs) are translation equivariant • Going beyond translations: make every block equivariant! 𝑊1 𝜙 𝑊𝐿 … 𝜙 𝑓𝜃(𝑦) 𝑦
  • 15. 15 Equivariant Nets • Steerable linear layers [Cohen et al., 2016], [Serre, 1971] 𝑆 = {𝑊 ∈ ℝ𝑝×𝑛 𝑇𝑔𝑊 = 𝑊𝑇𝑔, ∀𝑔 ∈ 𝐺 𝑊𝜃 = 𝑖=1 𝑟 𝜃𝑖𝜓𝑖 • Parameter efficiency: dim 𝑆 dim(ℝ𝑝×𝑛) = 𝑟 𝑝𝑛 • Non-linearities: elementwise 𝜙 with 𝑇𝑔 represented by permutation matrices. • In practice, choose 𝑇𝑔 (multiplicities=channels) at each layer and use existing libraries. ℝ𝑝×𝑛 𝑆 = span(𝜓1, … , 𝜓𝑟)
  • 16. 16 𝐴 𝑓𝜃 Equivariance in Inverse Problems Role of invariance of 𝒳 when solving inverse problems? System Equivariance: Let 𝑓𝜃: 𝑦 ↦ 𝑥 be the reconstruction function 𝑓𝜃 ∘ 𝐴 ∘ 𝑇𝑔𝑥 = 𝑇𝑔 ∘ 𝑓𝜃 ∘ 𝐴𝑥 ∀𝑔 ∈ 𝐺, ∀𝑥 ∈ 𝒳
  • 17. 17 Equivariance by Design 𝐴 𝑓𝜃 Case 1: 𝐴 equivariant on ℝ𝑛 + equivariant architecture 𝑓𝜃 = equivariant system 𝑓 ∘ 𝐴
  • 18. 18 𝐴 𝑓𝜃 ? Equivariance by Design Case 2: 𝐴 not equivariant on ℝ𝑛 + equivariant architecture 𝑓𝜃 ≠ equivariant system 𝑓 ∘ 𝐴
  • 19. 19 Examples Translation Rotation Permutation Amplitude Gaussian Blur Image Inpainting Sparse-view CT Accelerated MRI Downsampling (no antialias) ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
  • 20. 20 Unrolled Networks Standard regularisation approach: argmin 𝐴𝑥 − 𝑦 2 + 𝐽(𝑥) Solved via proximal gradient descent, 𝑥𝑡+1 = prox𝜂𝐽 𝑥𝑡 − 𝜂𝐴⊤ 𝐴𝑥𝑡 − 𝑦 where prox𝐽 𝑥 = argmin 𝑥 − 𝑥 2 + 𝐽 𝑥 . 𝑥 𝑥 Proposition: If the regularisation 𝐽(𝑥) is 𝐺-invariant, the proximal operator is 𝐺-equivariant prox𝐽 𝑇𝑔𝑥 = 𝑇𝑔prox𝐽 𝑥 E. Celledoni et al., “Equivariant neural networks for inverse problems,” 2021.
  • 21. 21 Unrolled Networks Idea: Unroll optimisation and replace proximal operators by learnable networks for 𝑡 = 1, … , 𝑅 iterations 𝑥𝑡+1 = 𝑓𝜃𝑡 𝑥𝑡 − 𝜂𝐴⊤ 𝐴𝑥𝑡 − 𝑦 where 𝑓𝜃𝑡 : ℝ𝑛 ↦ ℝ𝑛 are equivariant learnable networks 𝐴⊤ 𝑓𝜃1 𝜂𝐴⊤ 𝑦 𝑓𝜃𝑅 (𝐼 − 𝜂𝐴⊤ 𝐴) … 𝑓𝜃(𝑦)
  • 22. 22 Pose Generalization E. Celledoni et al., “Equivariant neural networks for inverse problems,” 2021. Computed tomography • LIDC-IDRI dataset • 512 x 512 images • 50 views • Supervised learning
  • 23. 23 Noise Generalization S. Mohan et al., “Robust and interpretable blind image denoising via bias-free convolutional neural networks,” 2019. Image denoising: 𝑦 = 𝑥 + 𝜖 where 𝜖 ∼ 𝒩 0, 𝜎2 Amplitude invariance: for all 𝑥 ∈ 𝒳, 𝑔 ∈ ℝ+ we have 𝑔𝑥 ∈ 𝒳. Equivariant relu network: 𝑓𝜃 𝑔𝑦 = 𝑔𝑓𝜃(𝑦) by removing all biases. 𝑓𝜃 𝑔𝑥 + 𝜖 = 𝑔𝑓𝜃 𝑥 + 𝜖 𝑔 = 𝑔𝑓𝜃 𝑥 + 𝜖′ where 𝜖′ ∼ 𝒩 0, 𝜎 𝑔 2 is an unseen noise distribution!
  • 26. 26 Limits to Supervised Learning Main disadvantage: Obtaining training signals 𝑥𝑖 can be expensive or impossible. • Medical and scientific imaging • Only solves inverse problems which we already know what to expect • Risk of training with signals from a different distribution train test?
  • 27. 27 Learning from only measurements 𝒚? argmin 𝑖 𝑦𝑖 − 𝐴𝑓𝜃(𝑦𝑖) 2 Learning Approach 𝒇𝜽 𝐴 𝜃 Proposition: 𝑓𝜃 𝑦 = 𝐴† 𝑦 + 𝑔𝜃(𝑦) where 𝑔𝜃: ℝ𝑚 ↦ 𝒩 𝐴 is any function whose image belongs to the nullspace of 𝐴 obtains zero training error.
  • 28. 28 Exploiting Invariance How to learn from only 𝒚? We need some prior information For all 𝑔 ∈ 𝐺 we have 𝑦 = 𝐴𝑥 = 𝐴𝑇𝑔𝑇𝑔 −1 𝑥 • Implicit access to multiple operators 𝐴𝑔 • Each operator with different nullspace = 𝐴𝑔𝑥′
  • 29. 29 Model Identification Can we uniquely identify the set of signals 𝒳 ⊂ ℝ𝑛 from the observed measurement sets 𝒴𝑔 = 𝐴𝑇𝑔𝒳 𝑔∈𝐺 ?
  • 30. Theorem [T., Chen and Davies ’22]: Identifying 𝒳 requires that 𝐴 is not equivariant: 𝐴𝑇𝑔 ≠ 𝑇𝑔𝐴 30 Necessary Conditions Proposition [T., Chen and Davies ’22]: Identifying 𝒳 from {𝒴𝑔 = 𝐴𝑔𝒳} possible only if rank 𝐴𝑇1 ⋮ 𝐴𝑇|𝐺| = 𝑛, and thus, if 𝑚 ≥ max 𝑐𝑗 𝑠𝑗 ≥ 𝑛 𝐺 where 𝑠𝑗 and 𝑐𝑗 are dimension and multiplicity of irreps.
  • 31. 31 Can We Learn Any Set? • Necessary conditions are independent of 𝒳… • For finite |𝐺|, some signal distributions cannot be identified if 𝐴 is rank-deficient: 𝔼𝑦{𝑒i𝑤⊤𝐴𝑔 † 𝑦 } = 𝔼𝑥 𝑒i𝑥⊤(𝐴𝑔 † 𝐴𝑔𝑤) = 𝜑(𝐴𝑔 † 𝐴𝑔𝑤) • We only observe projection of characteristic function 𝜑 to certain subspaces!
  • 32. Theorem [T., Chen and Davies ’22]: Let 𝐺 be a compact cyclic group. Identifying set 𝒳 with box-counting dimension 𝑘 from 𝒴𝑔 = 𝐴𝑇𝑔𝒳 𝑔 is possible by almost every 𝐴 ∈ ℝ𝑚×𝑛 with 𝑚 > 2𝑘 + max 𝑗 𝑐𝑗 + 1 where 𝑐𝑗 are the multiplicities of the representation and max 𝑗 𝑐𝑗 ≥ 𝑛/ 𝐺 . 32 Sufficient Conditions Additional assumption: The signal set is low-dimensional. If group is big, then 𝑚 > 2𝑘 + 2 are sufficient = same condition as signal recovery!
  • 33. 33 Consequences Magnetic resonance imaging • 𝐴 = subset of Fourier modes • Equivariant to translations • Not equivariant to rotations, which have max 𝑐𝑗 ≈ √𝑛 𝑚 > 2𝑘 + 𝑛 + 1 Image inpainting • 𝐴 = diagonal matrix with 1’s and 0s. • Not equivariant to translations, which have max 𝑐𝑗 ≈ 1 𝑚 > 2𝑘 + 2 Computed tomography • 𝐴 = 1D projections (sinograms) • Equivariant to translations • Not equivariant to rotations, which have max 𝑐𝑗 ≈ √𝑛 𝑚 > 2𝑘 + 𝑛 + 1 𝑥 𝑦 𝑦 𝑥 𝑦 𝑥
  • 34. 34 Equivariant Imaging Enforcing equivariance in practice? • Unrolled equivariant networks might not achieve equivariance of 𝑓𝜃 ∘ 𝐴 • Equivariant prox is not sufficient Example: Learned prox𝐽 𝑥 = 𝑥 is equivariant, resulting 𝑓𝜃 𝑦 = 𝐴†𝑦 is measurement consistent but not system equivariant Idea: enforce equivariance during training!
  • 35. 35 Self-supervised training loss: argmin ℒ𝑀𝐶 𝜃 • ℒ𝑀𝐶 𝜃 = 𝑖 𝑦𝑖 − 𝐴𝑓𝜃 𝑦𝑖 2 enforces measurement consistency • ℒ𝐸𝐼 𝜃 = 𝑖,𝑔 𝑓𝜃 𝐴𝑇𝑔𝑓𝜃 𝑦𝑖 − 𝑇𝑔𝑓𝜃 𝑦𝑖 2 enforces equivariance of 𝐴 ∘ 𝑓𝜃 𝜃 • Network-agnostic: applicable to any 𝑓𝜃 (including unrolled) +ℒ𝐸𝐼 𝜃 Equivariant Imaging
  • 36. 36 Robust EI: SURE+EI Robust Equivariant Imaging argmin ℒ𝑆𝑈𝑅𝐸 𝜃 + ℒ𝐸𝐼 𝜃 where ℒ𝑆𝑈𝑅𝐸 𝜃 = 𝑖 𝑦𝑖 − 𝐴𝑓𝜃 𝑦𝑖 2 − 𝜎2𝑚 + 2𝜎2div(𝐴 ∘ 𝑓𝜃)(𝑦𝑖) 𝜃 Theorem [Stein, 1981] Under mild differentiability conditions on the function 𝐴 ∘ 𝑓𝜃, the following holds 𝔼𝑦 ℒ𝑀𝐶 𝜃 = 𝔼𝑦 ℒ𝑆𝑈𝑅𝐸 𝜃 • Similar expressions of ℒ𝑆𝑈𝑅𝐸 𝜃 for Poisson, Poisson-Gaussian, etc. exist
  • 37. 37 Experiments Tasks: • Magnetic resonance imaging Network • 𝑓𝜃 = 𝑔𝜃 ∘ 𝐴† where 𝑔𝜃 is a U-Net Comparison • Pseudo-inverse 𝐴† 𝑦𝑖 (no training) • Meas. consistency 𝐴𝑓𝜃 𝑦𝑖 = 𝑦𝑖 • Fully supervised loss: 𝑓𝜃 𝑦𝑖 = 𝑥𝑖 • Equivariant imaging (unsupervised) 𝐴𝑓𝜃 𝑦𝑖 = 𝑦𝑖 and equivariant 𝐴 ∘ 𝑓𝜃
  • 38. Equivariant imaging Fully supervised 𝐴†𝑦 Meas. consistency 38 Magnetic resonance imaging • Operator 𝐴 is a subset of Fourier measurements (x2 downsampling) • Dataset is approximately rotation invariant Signal 𝑥 Measurements 𝑦
  • 39. 39 Inpainting Signal 𝑥 Measurements 𝑦 • Operator 𝐴 is an inpainting mask (30% pixels dropped) • Poisson noise (rate=10) • Dataset is approximately translation invariant Supervised Meas. consistency Robust EI
  • 40. 40 Noisy measurements 𝑦 Robust EI Supervised Clean signal 𝑥 Meas. consistency Computed Tomography • Operator 𝐴 is (non-linear variant) sparse radon transform (50 views) • Mixed Poisson-Gaussian noise • Dataset is approximately rotation invariant
  • 41. 41 Compressed Sensing • 𝐴 is a random iid Gaussian matrix • Shift invariant (max𝑗 𝑐𝑗 = 1) MNIST dataset 𝑘 ≈ 12
  • 43. 43 Conclusions • Learn signal set from data • Equivariance by design • Generalize unseen poses and noise level • No ground-truth references needed • Necessary and sufficient conditions
  • 44. 45 Conclusions Future work/open problems • Non-linear inverse problems • Semi-group actions • Approx. low-dimensional models • Beyond Euclidean domains
  • 45. 46 Extras • Theoretical bounds and algorithms can be extended to the case where we observe data through multiple operators 𝐴1𝑥 𝐴2𝑥 𝐴3𝑥 “Unsupervised Learning From Incomplete Measurements for Inverse Problems”, Tachella, Chen and Davies, NeurIPS 2022.
  • 46. 47 Papers [1] “Equivariant Imaging: Learning Beyond the Range Space”, Chen, Tachella and Davies, ICCV 2021 (Oral) [2] “Robust Equivariant Imaging: a fully unsupervised framework for learning to image from noisy and partial measurements”, Chen, Tachella and Davies, CVPR [3] “Unsupervised Learning From Incomplete Measurements for Inverse Problems”, Tachella, Chen and Davies, NeurIPS 2022. [4] “Sensing Theorems for Unsupervised Learning in Inverse Problems”, Tachella, Chen and Davies, JMLR 2023. [5] “Imaging with Equivariant Deep Learning”, Chen, Davies, Eerhardt, Schonlieb, Ferry and Tachella, IEEE SPM 2023.
  • 47. Thanks for your attention! Tachella.github.io  Codes  Presentations  … and more 48

Editor's Notes

  1. Ask people to answer these ones