SlideShare ist ein Scribd-Unternehmen logo
1 von 22
Downloaden Sie, um offline zu lesen
M A C H I N E L E A R N I N G T O D AY: C U R R E N T
R E S E A R C H A N D A D VA N C E S F R O M A M L A B , U V A
D A N I E L W O R R A L L
WHO ARE WE?
~30 researchers working under Max Welling and Joris Mooij
- 4 industrially funded ‘labs’
- Everyone works in deep learning
- We do fundamental research in machine learning
WHAT IS MACHINE LEARNING?
Mauna Loa is one of five volcanoes that form the Island of Hawaii in the U.S.
state of Hawaii in the Pacific Ocean. The largest subaerial volcano in both
mass and volume, Mauna Loa has historically been considered the largest
volcano on Earth, dwarfed only by Tamu Massif.
WHAT IS MACHINE LEARNING?
In machine learning we use past data to make predictions about the future.
p(y⇤|x⇤, D) = N y⇤|µ(x⇤), 2
(x⇤)
DataTest inputTest output Gaussian
WHAT IS MACHINE LEARNING?
Predictions are probability distributions.
Our main tool, conditional distributions:
Data
p(x|✓)
Parameters/models/unknowns
Symmetry constraints
Domain choice
Flexibility
Approximations
Computation
Memory
How do we choose p?
How do we learn θ?
WHAT IS MACHINE LEARNING?
Probability
p(x|✓)
✓
{x1, x2, ...} ✓
{x1, x2, ...}
Statistics
Machine
Learning
✓
{x1, x2, ...}
{x⇤}
Some terminology
WHAT WE DO
- Variational methods
- Normalizing flows
- Graphs
- Symmetry
- Reinforcement learning
- Transfer learning
- Medical imaging
- Generative modelling
- Compression
- Low precision neural networks
- Spiking neural networks
- Semi-supervised learning
VARIATIONAL METHODS
Approximate inference
Intractable
p(✓|D) =
p(D|✓)p(✓)
p(D)
=
p(D|✓)p(✓)
R
p(D|✓)p(✓) d✓
Log-likelihood Regulariser
ELBO
q (✓) = arg min DKL [q (✓)kp(✓|D)]
= arg min
Z
q (✓) log
q (✓)
p(✓|D)
d✓
= arg min
Z
q (✓) log
q (✓)p(D)
p(D|✓)p(✓)
d✓
= arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)]
q (✓) = arg min DKL [q (✓)kp(✓|D)]
= arg min
Z
q (✓) log
q (✓)
p(✓|D)
d✓
= arg min
Z
q (✓) log
q (✓)p(D)
p(D|✓)p(✓)
d✓
= arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)]
q (✓) = arg min DKL [q (✓)kp(✓|D)]
= arg min
Z
q (✓) log
q (✓)
p(✓|D)
d✓
= arg min
Z
q (✓) log
q (✓)p(D)
p(D|✓)p(✓)
d✓
= arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)]
‘Distance’ between
distributions
q (✓) = arg min DKL [q (✓)kp(✓|D)]
= arg min
Z
q (✓) log
q (✓)
p(✓|D)
d✓
= arg min
Z
q (✓) log
q (✓)p(D)
p(D|✓)p(✓)
d✓
= arg max Eq (✓)[p(D|✓)] DKL [q (✓)kp(✓)]
p(✓|D) =
p(D|✓)p(✓)
p(D)
=
p(D|✓)p(✓)
R
p(D|✓)p(✓) d✓
VARIATIONAL METHODS
Approximate inference
If we use latents (each x has a z) then we have a variational auto-encoder
arg max Ep(x)
⇥
Eq (z|x)[p(x|z)] DKL [q (z|x)kp(z)]
⇤
= arg min q (✓) log
q (✓)
p(✓|D)
d✓
= arg min
Z
q (✓) log
q (✓)p(D)
p(D|✓)p(✓)
d✓
= arg max Eq (✓)[p(D|✓)] DKL [q (✓)kp(✓)]
Neural network
Kingma and Welling (2013)
VARIATIONAL METHODS
NORMALIZING FLOWS
What is a flexible probability distribution?
e.g. p(x) = N(x|µ, 2
)
e.g. p(x) =
X
i
⇡iN(x|µi, 2
i )
x = f✓(z), z ⇠ p(z)
Implicitly define a distribution via a change
of variables
=) p(x) = p(z) det
@z
@x
= p(z) det
@f
@z
1
Rather expensive
Goal: design flexible f with cheap determinants
Target Flow
Rezende & Mohamed (2016)
NORMALIZING FLOWS
x = f✓(z), z ⇠ p(z)
Kingma & Dhariwal (2018)
NORMALIZING FLOWS: INVERTIBLE LAYERS
y = g(Wx + b)
Typical layer
Householder flow: volume-preservation
zt =
✓
I 2
vtv>
t
kvtk2
◆
zt 1 = Htzt
Predicted by NN
Tomczak and Welling (2016)
Sylvester normalising flows
zt = zt 1 + Ah (Bzt 1 + b)
det(I + AB) = det(I + BA)
det
@zt+1
@zt
= det (I + diag(h0
(Bz + b))BA)
van den Berg et al. (2018)
Emerging convolutions
Hoogeboom et al. (2019)
GRAPHS
A lot of data is graph-based: social networks, particle interactions, human
skeletal data, molecular structures, 3D graphics meshes
GRAPHS
Structure Weights
Kipf and Welling (2017)
GRAPHS: MOTION PREDICTION
Kipf et al. (2018)
SYMMETRY?
f(I) = f(T✓[I])
Notational aside:
T✓[I](x) = I(x ✓)
T✓[I](x) = I(R 1
✓ x)
T [I] = (I µ)/. 1
function/
feature mapping
image
transformation
Symmetry is a property of functions/tasks, e.g.
Classification
Disentangling
(cocktail party)
Signal discovery/detection
e.g. Geometric translation
e.g. Geometric rotation
e.g. Pixel normalisation
Set of input transformations leaving invariantS✓[f](I) = f(T✓[I])
EQUIVARIANCE
S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I])
transformation in feature space
Mapping preserves algebraic
structure of transformation
Different representations of same
transformation
https://github.com/vdumoulin/conv_arithmetic
Convolution (and correlation)
[I ⇤ W](x ✓) = T✓[I] ⇤ W(x)
S✓ = Id
Invariance
Convolutions Symmetry
GROUP EXAMPLES
*Current research direction: Scalings are probably better modelled as semigroups, i.e. groups without
the invertibility condition.
Scalings*
Translation
Reflections
Roto-translationRotation
Occlusions
Non-example
GROUP CONVOLUTIONS
“Convolution” examples [I ⇤ W](✓) =
X
x2Z2
I(x)W(R 1
✓ x)
[I ⇤ W](y) =
X
x2Z2
I(x)W(x y)
[I ⇤ W](✓, y) =
X
x2Z2
I(x)W(R 1
✓ x y)
[I ⇤ W](✓) =
X
x2Z2
I(x)T✓[W](x)Group convolution
[I ⇤ W](✓) =
X
x2Z2
T✓[I](x)W(x)Semigroup convolution
DenseNet Rotation
equivariant
DenseNet
DenseNet Rotation
equivariant
DenseNet
Input
Mean prediction Standard deviation
EXAMPLES
Machine Learning Today: Current Research And Advances From AMLAB, UvA

Weitere ähnliche Inhalte

Was ist angesagt?

Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
Создание картограмм на принципах грамматики графики. С помощью R-расширения g...Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
Matrunich Consulting
 

Was ist angesagt? (10)

Python Lecture 12
Python Lecture 12Python Lecture 12
Python Lecture 12
 
イケてる人生の描き方
イケてる人生の描き方イケてる人生の描き方
イケてる人生の描き方
 
Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
Создание картограмм на принципах грамматики графики. С помощью R-расширения g...Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
Создание картограмм на принципах грамматики графики. С помощью R-расширения g...
 
Combined Functions
Combined FunctionsCombined Functions
Combined Functions
 
Poster
PosterPoster
Poster
 
A Note on ZINB-VAE
A Note on ZINB-VAEA Note on ZINB-VAE
A Note on ZINB-VAE
 
Day 5 examples
Day 5 examplesDay 5 examples
Day 5 examples
 
Mate tarea - 2º
Mate   tarea - 2ºMate   tarea - 2º
Mate tarea - 2º
 
DeepLearning 6_5 ~ 6_5_3
DeepLearning 6_5 ~ 6_5_3DeepLearning 6_5 ~ 6_5_3
DeepLearning 6_5 ~ 6_5_3
 
Megadata With Python and Hadoop
Megadata With Python and HadoopMegadata With Python and Hadoop
Megadata With Python and Hadoop
 

Ähnlich wie Machine Learning Today: Current Research And Advances From AMLAB, UvA

Bayesian hybrid variable selection under generalized linear models
Bayesian hybrid variable selection under generalized linear modelsBayesian hybrid variable selection under generalized linear models
Bayesian hybrid variable selection under generalized linear models
Caleb (Shiqiang) Jin
 
lec2_CS540_handouts.pdf
lec2_CS540_handouts.pdflec2_CS540_handouts.pdf
lec2_CS540_handouts.pdf
ZineddineALICHE1
 

Ähnlich wie Machine Learning Today: Current Research And Advances From AMLAB, UvA (20)

ベイジアンディープニューラルネット
ベイジアンディープニューラルネットベイジアンディープニューラルネット
ベイジアンディープニューラルネット
 
Deep Learning Opening Workshop - Horseshoe Regularization for Machine Learnin...
Deep Learning Opening Workshop - Horseshoe Regularization for Machine Learnin...Deep Learning Opening Workshop - Horseshoe Regularization for Machine Learnin...
Deep Learning Opening Workshop - Horseshoe Regularization for Machine Learnin...
 
Workshop in honour of Don Poskitt and Gael Martin
Workshop in honour of Don Poskitt and Gael MartinWorkshop in honour of Don Poskitt and Gael Martin
Workshop in honour of Don Poskitt and Gael Martin
 
Pycon2017
Pycon2017Pycon2017
Pycon2017
 
Approximate Bayesian model choice via random forests
Approximate Bayesian model choice via random forestsApproximate Bayesian model choice via random forests
Approximate Bayesian model choice via random forests
 
Slides: A glance at information-geometric signal processing
Slides: A glance at information-geometric signal processingSlides: A glance at information-geometric signal processing
Slides: A glance at information-geometric signal processing
 
NBBC15, Reyjavik, June 08, 2015
NBBC15, Reyjavik, June 08, 2015NBBC15, Reyjavik, June 08, 2015
NBBC15, Reyjavik, June 08, 2015
 
Pattern learning and recognition on statistical manifolds: An information-geo...
Pattern learning and recognition on statistical manifolds: An information-geo...Pattern learning and recognition on statistical manifolds: An information-geo...
Pattern learning and recognition on statistical manifolds: An information-geo...
 
[DL輪読会]GANとエネルギーベースモデル
[DL輪読会]GANとエネルギーベースモデル[DL輪読会]GANとエネルギーベースモデル
[DL輪読会]GANとエネルギーベースモデル
 
Accelerating Pseudo-Marginal MCMC using Gaussian Processes
Accelerating Pseudo-Marginal MCMC using Gaussian ProcessesAccelerating Pseudo-Marginal MCMC using Gaussian Processes
Accelerating Pseudo-Marginal MCMC using Gaussian Processes
 
Bayesian Deep Learning
Bayesian Deep LearningBayesian Deep Learning
Bayesian Deep Learning
 
Dictionary Learning for Massive Matrix Factorization
Dictionary Learning for Massive Matrix FactorizationDictionary Learning for Massive Matrix Factorization
Dictionary Learning for Massive Matrix Factorization
 
Lecture12 xing
Lecture12 xingLecture12 xing
Lecture12 xing
 
Murphy: Machine learning A probabilistic perspective: Ch.9
Murphy: Machine learning A probabilistic perspective: Ch.9Murphy: Machine learning A probabilistic perspective: Ch.9
Murphy: Machine learning A probabilistic perspective: Ch.9
 
Bayesian hybrid variable selection under generalized linear models
Bayesian hybrid variable selection under generalized linear modelsBayesian hybrid variable selection under generalized linear models
Bayesian hybrid variable selection under generalized linear models
 
lec2_CS540_handouts.pdf
lec2_CS540_handouts.pdflec2_CS540_handouts.pdf
lec2_CS540_handouts.pdf
 
Laplace's Demon: seminar #1
Laplace's Demon: seminar #1Laplace's Demon: seminar #1
Laplace's Demon: seminar #1
 
Surface-related multiple elimination through orthogonal encoding in the laten...
Surface-related multiple elimination through orthogonal encoding in the laten...Surface-related multiple elimination through orthogonal encoding in the laten...
Surface-related multiple elimination through orthogonal encoding in the laten...
 
Continuous and Discrete-Time Analysis of SGD
Continuous and Discrete-Time Analysis of SGDContinuous and Discrete-Time Analysis of SGD
Continuous and Discrete-Time Analysis of SGD
 
Pattern-based classification of demographic sequences
Pattern-based classification of demographic sequencesPattern-based classification of demographic sequences
Pattern-based classification of demographic sequences
 

Mehr von Advanced-Concepts-Team

Pablo Gomez - Solving Large-scale Challenges with ESA Datalabs
Pablo Gomez - Solving Large-scale Challenges with ESA DatalabsPablo Gomez - Solving Large-scale Challenges with ESA Datalabs
Pablo Gomez - Solving Large-scale Challenges with ESA Datalabs
Advanced-Concepts-Team
 
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
Advanced-Concepts-Team
 
EDEN ISS - A space greenhouse analogue in Antarctica
EDEN ISS - A space greenhouse analogue in AntarcticaEDEN ISS - A space greenhouse analogue in Antarctica
EDEN ISS - A space greenhouse analogue in Antarctica
Advanced-Concepts-Team
 
How to give a robot a soul
How to give a robot a soulHow to give a robot a soul
How to give a robot a soul
Advanced-Concepts-Team
 
Vernal pools a new ecosystem for astrobiology studies
Vernal pools a new ecosystem for astrobiology studiesVernal pools a new ecosystem for astrobiology studies
Vernal pools a new ecosystem for astrobiology studies
Advanced-Concepts-Team
 

Mehr von Advanced-Concepts-Team (20)

2024.03.22 - Mike Heddes - Introduction to Hyperdimensional Computing.pdf
2024.03.22 - Mike Heddes - Introduction to Hyperdimensional Computing.pdf2024.03.22 - Mike Heddes - Introduction to Hyperdimensional Computing.pdf
2024.03.22 - Mike Heddes - Introduction to Hyperdimensional Computing.pdf
 
Isabelle Diacaire - From Ariadnas to Industry R&D in optics and photonics
Isabelle Diacaire - From Ariadnas to Industry R&D in optics and photonicsIsabelle Diacaire - From Ariadnas to Industry R&D in optics and photonics
Isabelle Diacaire - From Ariadnas to Industry R&D in optics and photonics
 
The ExoGRAVITY project - observations of exoplanets from the ground with opti...
The ExoGRAVITY project - observations of exoplanets from the ground with opti...The ExoGRAVITY project - observations of exoplanets from the ground with opti...
The ExoGRAVITY project - observations of exoplanets from the ground with opti...
 
MOND_famaey.pdf
MOND_famaey.pdfMOND_famaey.pdf
MOND_famaey.pdf
 
Pablo Gomez - Solving Large-scale Challenges with ESA Datalabs
Pablo Gomez - Solving Large-scale Challenges with ESA DatalabsPablo Gomez - Solving Large-scale Challenges with ESA Datalabs
Pablo Gomez - Solving Large-scale Challenges with ESA Datalabs
 
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
Jonathan Sauder - Miniaturizing Mechanical Systems for CubeSats: Design Princ...
 
Towards an Artificial Muse for new Ideas in Quantum Physics
Towards an Artificial Muse for new Ideas in Quantum PhysicsTowards an Artificial Muse for new Ideas in Quantum Physics
Towards an Artificial Muse for new Ideas in Quantum Physics
 
EDEN ISS - A space greenhouse analogue in Antarctica
EDEN ISS - A space greenhouse analogue in AntarcticaEDEN ISS - A space greenhouse analogue in Antarctica
EDEN ISS - A space greenhouse analogue in Antarctica
 
How to give a robot a soul
How to give a robot a soulHow to give a robot a soul
How to give a robot a soul
 
Information processing with artificial spiking neural networks
Information processing with artificial spiking neural networksInformation processing with artificial spiking neural networks
Information processing with artificial spiking neural networks
 
Exploring Architected Materials Using Machine Learning
Exploring Architected Materials Using Machine LearningExploring Architected Materials Using Machine Learning
Exploring Architected Materials Using Machine Learning
 
Electromagnetically Actuated Systems for Modular, Self-Assembling and Self-Re...
Electromagnetically Actuated Systems for Modular, Self-Assembling and Self-Re...Electromagnetically Actuated Systems for Modular, Self-Assembling and Self-Re...
Electromagnetically Actuated Systems for Modular, Self-Assembling and Self-Re...
 
HORUS: Peering into Lunar Shadowed Regions with AI
HORUS: Peering into Lunar Shadowed Regions with AIHORUS: Peering into Lunar Shadowed Regions with AI
HORUS: Peering into Lunar Shadowed Regions with AI
 
META-SPACE: Psycho-physiologically Adaptive and Personalized Virtual Reality ...
META-SPACE: Psycho-physiologically Adaptive and Personalized Virtual Reality ...META-SPACE: Psycho-physiologically Adaptive and Personalized Virtual Reality ...
META-SPACE: Psycho-physiologically Adaptive and Personalized Virtual Reality ...
 
The Large Interferometer For Exoplanets (LIFE) II: Key Methods and Technologies
The Large Interferometer For Exoplanets (LIFE) II: Key Methods and TechnologiesThe Large Interferometer For Exoplanets (LIFE) II: Key Methods and Technologies
The Large Interferometer For Exoplanets (LIFE) II: Key Methods and Technologies
 
Black Holes and Bright Quasars
Black Holes and Bright QuasarsBlack Holes and Bright Quasars
Black Holes and Bright Quasars
 
In vitro simulation of spaceflight environment to elucidate combined effect o...
In vitro simulation of spaceflight environment to elucidate combined effect o...In vitro simulation of spaceflight environment to elucidate combined effect o...
In vitro simulation of spaceflight environment to elucidate combined effect o...
 
The Large Interferometer For Exoplanets (LIFE): the science of characterising...
The Large Interferometer For Exoplanets (LIFE): the science of characterising...The Large Interferometer For Exoplanets (LIFE): the science of characterising...
The Large Interferometer For Exoplanets (LIFE): the science of characterising...
 
Vernal pools a new ecosystem for astrobiology studies
Vernal pools a new ecosystem for astrobiology studiesVernal pools a new ecosystem for astrobiology studies
Vernal pools a new ecosystem for astrobiology studies
 
Keeping a Sentinel Eye on the Volcanoes – from Space!
Keeping a Sentinel Eye on the Volcanoes – from Space!Keeping a Sentinel Eye on the Volcanoes – from Space!
Keeping a Sentinel Eye on the Volcanoes – from Space!
 

Kürzlich hochgeladen

Pests of mustard_Identification_Management_Dr.UPR.pdf
Pests of mustard_Identification_Management_Dr.UPR.pdfPests of mustard_Identification_Management_Dr.UPR.pdf
Pests of mustard_Identification_Management_Dr.UPR.pdf
PirithiRaju
 
Bacterial Identification and Classifications
Bacterial Identification and ClassificationsBacterial Identification and Classifications
Bacterial Identification and Classifications
Areesha Ahmad
 
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
Lokesh Kothari
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...
RohitNehra6
 
Presentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptxPresentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptx
gindu3009
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Sérgio Sacani
 
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
Lokesh Kothari
 

Kürzlich hochgeladen (20)

Pests of mustard_Identification_Management_Dr.UPR.pdf
Pests of mustard_Identification_Management_Dr.UPR.pdfPests of mustard_Identification_Management_Dr.UPR.pdf
Pests of mustard_Identification_Management_Dr.UPR.pdf
 
Bacterial Identification and Classifications
Bacterial Identification and ClassificationsBacterial Identification and Classifications
Bacterial Identification and Classifications
 
TEST BANK For Radiologic Science for Technologists, 12th Edition by Stewart C...
TEST BANK For Radiologic Science for Technologists, 12th Edition by Stewart C...TEST BANK For Radiologic Science for Technologists, 12th Edition by Stewart C...
TEST BANK For Radiologic Science for Technologists, 12th Edition by Stewart C...
 
Botany 4th semester file By Sumit Kumar yadav.pdf
Botany 4th semester file By Sumit Kumar yadav.pdfBotany 4th semester file By Sumit Kumar yadav.pdf
Botany 4th semester file By Sumit Kumar yadav.pdf
 
Creating and Analyzing Definitive Screening Designs
Creating and Analyzing Definitive Screening DesignsCreating and Analyzing Definitive Screening Designs
Creating and Analyzing Definitive Screening Designs
 
Zoology 4th semester series (krishna).pdf
Zoology 4th semester series (krishna).pdfZoology 4th semester series (krishna).pdf
Zoology 4th semester series (krishna).pdf
 
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
All-domain Anomaly Resolution Office U.S. Department of Defense (U) Case: “Eg...
 
Green chemistry and Sustainable development.pptx
Green chemistry  and Sustainable development.pptxGreen chemistry  and Sustainable development.pptx
Green chemistry and Sustainable development.pptx
 
Botany 4th semester series (krishna).pdf
Botany 4th semester series (krishna).pdfBotany 4th semester series (krishna).pdf
Botany 4th semester series (krishna).pdf
 
GBSN - Biochemistry (Unit 1)
GBSN - Biochemistry (Unit 1)GBSN - Biochemistry (Unit 1)
GBSN - Biochemistry (Unit 1)
 
❤Jammu Kashmir Call Girls 8617697112 Personal Whatsapp Number 💦✅.
❤Jammu Kashmir Call Girls 8617697112 Personal Whatsapp Number 💦✅.❤Jammu Kashmir Call Girls 8617697112 Personal Whatsapp Number 💦✅.
❤Jammu Kashmir Call Girls 8617697112 Personal Whatsapp Number 💦✅.
 
Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )Recombination DNA Technology (Nucleic Acid Hybridization )
Recombination DNA Technology (Nucleic Acid Hybridization )
 
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
GUIDELINES ON SIMILAR BIOLOGICS Regulatory Requirements for Marketing Authori...
 
Biopesticide (2).pptx .This slides helps to know the different types of biop...
Biopesticide (2).pptx  .This slides helps to know the different types of biop...Biopesticide (2).pptx  .This slides helps to know the different types of biop...
Biopesticide (2).pptx .This slides helps to know the different types of biop...
 
Presentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptxPresentation Vikram Lander by Vedansh Gupta.pptx
Presentation Vikram Lander by Vedansh Gupta.pptx
 
Hire 💕 9907093804 Hooghly Call Girls Service Call Girls Agency
Hire 💕 9907093804 Hooghly Call Girls Service Call Girls AgencyHire 💕 9907093804 Hooghly Call Girls Service Call Girls Agency
Hire 💕 9907093804 Hooghly Call Girls Service Call Girls Agency
 
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
High Class Escorts in Hyderabad ₹7.5k Pick Up & Drop With Cash Payment 969456...
 
Biological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdfBiological Classification BioHack (3).pdf
Biological Classification BioHack (3).pdf
 
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroidsHubble Asteroid Hunter III. Physical properties of newly found asteroids
Hubble Asteroid Hunter III. Physical properties of newly found asteroids
 
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
Labelling Requirements and Label Claims for Dietary Supplements and Recommend...
 

Machine Learning Today: Current Research And Advances From AMLAB, UvA

  • 1. M A C H I N E L E A R N I N G T O D AY: C U R R E N T R E S E A R C H A N D A D VA N C E S F R O M A M L A B , U V A D A N I E L W O R R A L L
  • 2. WHO ARE WE? ~30 researchers working under Max Welling and Joris Mooij - 4 industrially funded ‘labs’ - Everyone works in deep learning - We do fundamental research in machine learning
  • 3. WHAT IS MACHINE LEARNING? Mauna Loa is one of five volcanoes that form the Island of Hawaii in the U.S. state of Hawaii in the Pacific Ocean. The largest subaerial volcano in both mass and volume, Mauna Loa has historically been considered the largest volcano on Earth, dwarfed only by Tamu Massif.
  • 4. WHAT IS MACHINE LEARNING? In machine learning we use past data to make predictions about the future. p(y⇤|x⇤, D) = N y⇤|µ(x⇤), 2 (x⇤) DataTest inputTest output Gaussian
  • 5. WHAT IS MACHINE LEARNING? Predictions are probability distributions. Our main tool, conditional distributions: Data p(x|✓) Parameters/models/unknowns Symmetry constraints Domain choice Flexibility Approximations Computation Memory How do we choose p? How do we learn θ?
  • 6. WHAT IS MACHINE LEARNING? Probability p(x|✓) ✓ {x1, x2, ...} ✓ {x1, x2, ...} Statistics Machine Learning ✓ {x1, x2, ...} {x⇤} Some terminology
  • 7. WHAT WE DO - Variational methods - Normalizing flows - Graphs - Symmetry - Reinforcement learning - Transfer learning - Medical imaging - Generative modelling - Compression - Low precision neural networks - Spiking neural networks - Semi-supervised learning
  • 8. VARIATIONAL METHODS Approximate inference Intractable p(✓|D) = p(D|✓)p(✓) p(D) = p(D|✓)p(✓) R p(D|✓)p(✓) d✓ Log-likelihood Regulariser ELBO q (✓) = arg min DKL [q (✓)kp(✓|D)] = arg min Z q (✓) log q (✓) p(✓|D) d✓ = arg min Z q (✓) log q (✓)p(D) p(D|✓)p(✓) d✓ = arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)] q (✓) = arg min DKL [q (✓)kp(✓|D)] = arg min Z q (✓) log q (✓) p(✓|D) d✓ = arg min Z q (✓) log q (✓)p(D) p(D|✓)p(✓) d✓ = arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)] q (✓) = arg min DKL [q (✓)kp(✓|D)] = arg min Z q (✓) log q (✓) p(✓|D) d✓ = arg min Z q (✓) log q (✓)p(D) p(D|✓)p(✓) d✓ = arg max Eq (✓)[p(D|✓)] + DKL [q (✓)kp(✓)] ‘Distance’ between distributions q (✓) = arg min DKL [q (✓)kp(✓|D)] = arg min Z q (✓) log q (✓) p(✓|D) d✓ = arg min Z q (✓) log q (✓)p(D) p(D|✓)p(✓) d✓ = arg max Eq (✓)[p(D|✓)] DKL [q (✓)kp(✓)] p(✓|D) = p(D|✓)p(✓) p(D) = p(D|✓)p(✓) R p(D|✓)p(✓) d✓
  • 9. VARIATIONAL METHODS Approximate inference If we use latents (each x has a z) then we have a variational auto-encoder arg max Ep(x) ⇥ Eq (z|x)[p(x|z)] DKL [q (z|x)kp(z)] ⇤ = arg min q (✓) log q (✓) p(✓|D) d✓ = arg min Z q (✓) log q (✓)p(D) p(D|✓)p(✓) d✓ = arg max Eq (✓)[p(D|✓)] DKL [q (✓)kp(✓)] Neural network Kingma and Welling (2013)
  • 11. NORMALIZING FLOWS What is a flexible probability distribution? e.g. p(x) = N(x|µ, 2 ) e.g. p(x) = X i ⇡iN(x|µi, 2 i ) x = f✓(z), z ⇠ p(z) Implicitly define a distribution via a change of variables =) p(x) = p(z) det @z @x = p(z) det @f @z 1 Rather expensive Goal: design flexible f with cheap determinants Target Flow Rezende & Mohamed (2016)
  • 12. NORMALIZING FLOWS x = f✓(z), z ⇠ p(z) Kingma & Dhariwal (2018)
  • 13. NORMALIZING FLOWS: INVERTIBLE LAYERS y = g(Wx + b) Typical layer Householder flow: volume-preservation zt = ✓ I 2 vtv> t kvtk2 ◆ zt 1 = Htzt Predicted by NN Tomczak and Welling (2016) Sylvester normalising flows zt = zt 1 + Ah (Bzt 1 + b) det(I + AB) = det(I + BA) det @zt+1 @zt = det (I + diag(h0 (Bz + b))BA) van den Berg et al. (2018) Emerging convolutions Hoogeboom et al. (2019)
  • 14. GRAPHS A lot of data is graph-based: social networks, particle interactions, human skeletal data, molecular structures, 3D graphics meshes
  • 17. SYMMETRY? f(I) = f(T✓[I]) Notational aside: T✓[I](x) = I(x ✓) T✓[I](x) = I(R 1 ✓ x) T [I] = (I µ)/. 1 function/ feature mapping image transformation Symmetry is a property of functions/tasks, e.g. Classification Disentangling (cocktail party) Signal discovery/detection e.g. Geometric translation e.g. Geometric rotation e.g. Pixel normalisation Set of input transformations leaving invariantS✓[f](I) = f(T✓[I])
  • 18. EQUIVARIANCE S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I])S✓[f](I) = f(T✓[I]) transformation in feature space Mapping preserves algebraic structure of transformation Different representations of same transformation https://github.com/vdumoulin/conv_arithmetic Convolution (and correlation) [I ⇤ W](x ✓) = T✓[I] ⇤ W(x) S✓ = Id Invariance Convolutions Symmetry
  • 19. GROUP EXAMPLES *Current research direction: Scalings are probably better modelled as semigroups, i.e. groups without the invertibility condition. Scalings* Translation Reflections Roto-translationRotation Occlusions Non-example
  • 20. GROUP CONVOLUTIONS “Convolution” examples [I ⇤ W](✓) = X x2Z2 I(x)W(R 1 ✓ x) [I ⇤ W](y) = X x2Z2 I(x)W(x y) [I ⇤ W](✓, y) = X x2Z2 I(x)W(R 1 ✓ x y) [I ⇤ W](✓) = X x2Z2 I(x)T✓[W](x)Group convolution [I ⇤ W](✓) = X x2Z2 T✓[I](x)W(x)Semigroup convolution