SlideShare a Scribd company logo
Variational Autoencoders for Speech
Dmytro Bielievtsov
2018
Log magnitude spectrograms, linear and mel
Interpreting spectrograms
from home.cc.umanitoba.ca/~robh/cursol.html
Generative models
Generative models model the distribution of high-dimensional
objects like images, audio, etc.
They model these distributions either by allowing you to
sample from the distribution or by computing likelihoods
Three families of generative model dominate the deep learning
landscape today: autoregressive, GANs, and VAEs
Generative model comparison
Autoregressive models can be very accurate (WaveNet) but
are slow to generate (but cf. Distilled WaveNet), and produce
no latent representation of the data
GANs produce a latent representation of the data but are
finicky to train and don’t give likelihoods (at least in their
pure form)
VAEs lack the disadvantages of autoregressive models and
GANs, but in their basic form are prone to producing blurry
samples
Disentangled representations
Each latent controls one “factor of variation”. You don’t have
one latent controlling, e.g., pitch + volume and another pitch
- volume
The factor of variation each latent controls is independent of
the setting of the other latents
VAE model definition
VAE model definition
Latent variable models, easy things to do
Drawing samples is easy: z ∼ pθ(z), x ∼ pθ(x | z)
As is computing joint likelihoods
Latent variable models, hard things to do
To train the model, we need to maximize pθ(x)
We also need pθ(x) if we want to find likelihoods of samples
To find latent codes of samples (e.g., to do morphing), we
need to know pθ(z | x)
Unfortunately, pθ(x) and pθ(z | x) are intractible
Latent variable models, intractable integrals
pθ(x) = pθ(z)pθ(x | z) dz
pθ(z | x) = pθ(z)pθ(x | z)/pθ(x)
Since z is high-dimensional, pθ(x) and pθ(z | x) are intractable
By the way, from now on we will drop θ from our notation for less
clutter
The idea of an encoder
For fixed x, for almost all values of z, p(z | x) will be very
nearly zero
If we could somehow train a network to approximate p(z | x),
maybe we could estimate p(x) much more efficiently by
sampling z values that are likely to generate something close
to x
The encoder network
The encoder network
Instead of using fixed variances, as with the decoder network,
the encoder network also produces variances
Intuitively, we want the network to find an efficient
representation of the data, so we penalize it for outputting
very small variances
The fundamental VAE equation
Start with the definition of KL-divergence and then use Bayes’ Law:
D (q(z | x) p(z | x))
= Ez∼q(z|x)[log q(z | x) − log p(z | x)]
= Ez∼q(z|x)[log q(z | x) − log p(x | z) − log p(z)] + log p(x)
Rearranging terms and using the definition of KL-divergence,
OELBO
:= log p(x) − D (q(z | x) p(z | x))
= Ez∼q(z|x)[log p(x | z)]
tractable to optimize
− D (q(z | x) p(z))
tractable to optimize
Recap
pθ(x | z) := N(x | fθ(z), σ2
I)
p(z) := N(z | 0, I)
qφ(z | x) := N(x | gφ(x), diag(hφ(x))
OELBO(x; θ, φ) := Ez∼qφ(z|x)[log pθ(x | z)] − D (qφ(z | x) p(z))
A picture, and the reparameterization trick
Interpretability of latent variables
Latent variables produced by vanilla VAE’s are usually meaningless
when considered individually.
Adapted from Higgins et al. 2017
Disentangling variables
Simple approaches to encourage disentangled latent variables:
β-VAE
factor-VAE
Disentangling variables: β-VAE
Original objective function (ELBO):
OELBO = Ez∼q(z|x)[log p(x | z)] − D (q(z | x) p(z))
Disentangling variables: β-VAE
Original objective function (ELBO):
OELBO = Ez∼q(z|x)[log p(x | z)] − D (q(z | x) p(z))
New objective function:
Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
Disentangling variables: β-VAE
Adapted from Higgins et al. 2017
Disentangling variables: factor-VAE
Upon closer inspection, β-VAE creates some tradeoff between
disentanglement and reconstruction:
β-VAE objective:
Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
Disentangling variables: factor-VAE
Upon closer inspection, β-VAE creates some tradeoff between
disentanglement and reconstruction:
β-VAE objective:
Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
Let’s look closer at that regularizer:
Ex∼pdata(x)[D (q(z | x) p(z))] = Ienc(x; z) + D (q(z) p(z))
Disentangling variables: factor-VAE
Upon closer inspection, β-VAE creates some tradeoff between
disentanglement and reconstruction:
β-VAE objective:
Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
Let’s look closer at that regularizer:
Ex∼pdata(x)[D (q(z | x) p(z))] = Ienc(x; z) + D (q(z) p(z))
We don’t want to hurt that mutual information term!
Disentangling variables: factor-VAE
Yet another variation on the ELBO objective:
Oγ = OELBO − γD (q(z) ¯q(z))
Disentangling variables: factor-VAE
Yet another variation on the ELBO objective:
Oγ = OELBO − γD (q(z) ¯q(z))
Adapted from Kim et al. 2018
Disentangling variables: factor-VAE
Yet another variation on the ELBO objective:
Oγ = OELBO − γD (q(z) ¯q(z))
Adapted from Kim et al. 2018
VQ-VAE, quick look
Adapted from Oord et al. 2017
References
Doersch, Carl. “Tutorial on Variational Autoencoders.” ArXiv:1606.05908 [Cs, Stat], June 19, 2016.
http://arxiv.org/abs/1606.05908.
Higgins, Irina, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir
Mohamed, and Alexander Lerchner. “β-VAE: LEARNING BASIC VISUAL CONCEPTS WITH A
CONSTRAINED VARIATIONAL FRAMEWORK,” 2017, 22.
Hsu, Wei-Ning, Yu Zhang, and James Glass. “Learning Latent Representations for Speech Generation and
Transformation.” ArXiv:1704.04222 [Cs, Stat], April 13, 2017. http://arxiv.org/abs/1704.04222.
———. “Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data.”
ArXiv:1709.07902 [Cs, Eess, Stat], September 22, 2017. http://arxiv.org/abs/1709.07902.
Kim, Hyunjik, and Andriy Mnih. “Disentangling by Factorising.” ArXiv:1802.05983 [Cs, Stat], February 16,
2018. http://arxiv.org/abs/1802.05983.
Kingma, Diederik P., and Max Welling. “Auto-Encoding Variational Bayes.” ArXiv:1312.6114 [Cs, Stat],
December 20, 2013. http://arxiv.org/abs/1312.6114.
Oord, Aaron van den, Oriol Vinyals, and Koray Kavukcuoglu. “Neural Discrete Representation Learning.”
ArXiv:1711.00937 [Cs], November 2, 2017. http://arxiv.org/abs/1711.00937.

More Related Content

What's hot

Variational Autoencoders For Image Generation
Variational Autoencoders For Image GenerationVariational Autoencoders For Image Generation
Variational Autoencoders For Image Generation
Jason Anderson
 
Submodularity slides
Submodularity slidesSubmodularity slides
Submodularity slides
dragonthu
 
Toward Disentanglement through Understand ELBO
Toward Disentanglement through Understand ELBOToward Disentanglement through Understand ELBO
Toward Disentanglement through Understand ELBO
Kai-Wen Zhao
 
Mathematics notes and formula for class 12 chapter 7. integrals
Mathematics notes and formula for class 12 chapter 7. integrals Mathematics notes and formula for class 12 chapter 7. integrals
Mathematics notes and formula for class 12 chapter 7. integrals
sakhi pathak
 
Efficient end-to-end learning for quantizable representations
Efficient end-to-end learning for quantizable representationsEfficient end-to-end learning for quantizable representations
Efficient end-to-end learning for quantizable representations
NAVER Engineering
 
RBM from Scratch
RBM from Scratch RBM from Scratch
RBM from Scratch
Hadi Sinaee
 
(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning
Masahiro Suzuki
 
Functionsandpigeonholeprinciple
FunctionsandpigeonholeprincipleFunctionsandpigeonholeprinciple
Functionsandpigeonholeprinciple
Shiwani Gupta
 
Lecture 06 marco aurelio ranzato - deep learning
Lecture 06   marco aurelio ranzato - deep learningLecture 06   marco aurelio ranzato - deep learning
Lecture 06 marco aurelio ranzato - deep learning
mustafa sarac
 
Linear Classifiers
Linear ClassifiersLinear Classifiers
Linear Classifiers
Alexander Jung
 
Shortest path search for real road networks and dynamic costs with pgRouting
Shortest path search for real road networks and dynamic costs with pgRoutingShortest path search for real road networks and dynamic costs with pgRouting
Shortest path search for real road networks and dynamic costs with pgRouting
antonpa
 
4 informed-search
4 informed-search4 informed-search
4 informed-search
Mhd Sb
 
Reading group gan - 20170417
Reading group   gan - 20170417Reading group   gan - 20170417
Reading group gan - 20170417
Shuai Zhang
 
Detecting Bugs in Binaries Using Decompilation and Data Flow Analysis
Detecting Bugs in Binaries Using Decompilation and Data Flow AnalysisDetecting Bugs in Binaries Using Decompilation and Data Flow Analysis
Detecting Bugs in Binaries Using Decompilation and Data Flow Analysis
Silvio Cesare
 
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
Rizwan Habib
 
20180722 pyro
20180722 pyro20180722 pyro
20180722 pyro
Taku Yoshioka
 
Graph Traversal Algorithms - Depth First Search Traversal
Graph Traversal Algorithms - Depth First Search TraversalGraph Traversal Algorithms - Depth First Search Traversal
Graph Traversal Algorithms - Depth First Search Traversal
Amrinder Arora
 
Boosted Tree-based Multinomial Logit Model for Aggregated Market Data
Boosted Tree-based Multinomial Logit Model for Aggregated Market DataBoosted Tree-based Multinomial Logit Model for Aggregated Market Data
Boosted Tree-based Multinomial Logit Model for Aggregated Market Data
Jay (Jianqiang) Wang
 
Lec07 aggregation-and-retrieval-system
Lec07 aggregation-and-retrieval-systemLec07 aggregation-and-retrieval-system
Lec07 aggregation-and-retrieval-system
United States Air Force Academy
 

What's hot (19)

Variational Autoencoders For Image Generation
Variational Autoencoders For Image GenerationVariational Autoencoders For Image Generation
Variational Autoencoders For Image Generation
 
Submodularity slides
Submodularity slidesSubmodularity slides
Submodularity slides
 
Toward Disentanglement through Understand ELBO
Toward Disentanglement through Understand ELBOToward Disentanglement through Understand ELBO
Toward Disentanglement through Understand ELBO
 
Mathematics notes and formula for class 12 chapter 7. integrals
Mathematics notes and formula for class 12 chapter 7. integrals Mathematics notes and formula for class 12 chapter 7. integrals
Mathematics notes and formula for class 12 chapter 7. integrals
 
Efficient end-to-end learning for quantizable representations
Efficient end-to-end learning for quantizable representationsEfficient end-to-end learning for quantizable representations
Efficient end-to-end learning for quantizable representations
 
RBM from Scratch
RBM from Scratch RBM from Scratch
RBM from Scratch
 
(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning(DL輪読)Matching Networks for One Shot Learning
(DL輪読)Matching Networks for One Shot Learning
 
Functionsandpigeonholeprinciple
FunctionsandpigeonholeprincipleFunctionsandpigeonholeprinciple
Functionsandpigeonholeprinciple
 
Lecture 06 marco aurelio ranzato - deep learning
Lecture 06   marco aurelio ranzato - deep learningLecture 06   marco aurelio ranzato - deep learning
Lecture 06 marco aurelio ranzato - deep learning
 
Linear Classifiers
Linear ClassifiersLinear Classifiers
Linear Classifiers
 
Shortest path search for real road networks and dynamic costs with pgRouting
Shortest path search for real road networks and dynamic costs with pgRoutingShortest path search for real road networks and dynamic costs with pgRouting
Shortest path search for real road networks and dynamic costs with pgRouting
 
4 informed-search
4 informed-search4 informed-search
4 informed-search
 
Reading group gan - 20170417
Reading group   gan - 20170417Reading group   gan - 20170417
Reading group gan - 20170417
 
Detecting Bugs in Binaries Using Decompilation and Data Flow Analysis
Detecting Bugs in Binaries Using Decompilation and Data Flow AnalysisDetecting Bugs in Binaries Using Decompilation and Data Flow Analysis
Detecting Bugs in Binaries Using Decompilation and Data Flow Analysis
 
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
NYAI - A Path To Unsupervised Learning Through Adversarial Networks by Soumit...
 
20180722 pyro
20180722 pyro20180722 pyro
20180722 pyro
 
Graph Traversal Algorithms - Depth First Search Traversal
Graph Traversal Algorithms - Depth First Search TraversalGraph Traversal Algorithms - Depth First Search Traversal
Graph Traversal Algorithms - Depth First Search Traversal
 
Boosted Tree-based Multinomial Logit Model for Aggregated Market Data
Boosted Tree-based Multinomial Logit Model for Aggregated Market DataBoosted Tree-based Multinomial Logit Model for Aggregated Market Data
Boosted Tree-based Multinomial Logit Model for Aggregated Market Data
 
Lec07 aggregation-and-retrieval-system
Lec07 aggregation-and-retrieval-systemLec07 aggregation-and-retrieval-system
Lec07 aggregation-and-retrieval-system
 

Similar to Variational autoencoders for speech processing d.bielievtsov dataconf 21 04 18

VAE-type Deep Generative Models
VAE-type Deep Generative ModelsVAE-type Deep Generative Models
VAE-type Deep Generative Models
Kenta Oono
 
從 VAE 走向深度學習新理論
從 VAE 走向深度學習新理論從 VAE 走向深度學習新理論
從 VAE 走向深度學習新理論
岳華 杜
 
Introduction to Deep Generative Models
Introduction to Deep Generative ModelsIntroduction to Deep Generative Models
Introduction to Deep Generative Models
Hao-Wen (Herman) Dong
 
A discussion on sampling graphs to approximate network classification functions
A discussion on sampling graphs to approximate network classification functionsA discussion on sampling graphs to approximate network classification functions
A discussion on sampling graphs to approximate network classification functions
LARCA UPC
 
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
Frank Nielsen
 
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
준식 최
 
Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...
Valentin De Bortoli
 
Auto encoding-variational-bayes
Auto encoding-variational-bayesAuto encoding-variational-bayes
Auto encoding-variational-bayes
mehdi Cherti
 
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
Kohei Hayashi
 
Reading revue of "Inferring Multiple Graphical Structures"
Reading revue of "Inferring Multiple Graphical Structures"Reading revue of "Inferring Multiple Graphical Structures"
Reading revue of "Inferring Multiple Graphical Structures"
tuxette
 
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
The Statistical and Applied Mathematical Sciences Institute
 
Deep generative model.pdf
Deep generative model.pdfDeep generative model.pdf
Deep generative model.pdf
Hyungjoo Cho
 
Presentation OCIP 2015
Presentation OCIP 2015Presentation OCIP 2015
Presentation OCIP 2015
Fabian Froehlich
 
17_monte_carlo.pdf
17_monte_carlo.pdf17_monte_carlo.pdf
17_monte_carlo.pdf
KSChidanandKumarJSSS
 
Paper Summary of Disentangling by Factorising (Factor-VAE)
Paper Summary of Disentangling by Factorising (Factor-VAE)Paper Summary of Disentangling by Factorising (Factor-VAE)
Paper Summary of Disentangling by Factorising (Factor-VAE)
준식 최
 
Signals and Systems Homework Help.pptx
Signals and Systems Homework Help.pptxSignals and Systems Homework Help.pptx
Signals and Systems Homework Help.pptx
Matlab Assignment Experts
 
Automatic bayesian cubature
Automatic bayesian cubatureAutomatic bayesian cubature
Automatic bayesian cubature
Jagadeeswaran Rathinavel
 
the ABC of ABC
the ABC of ABCthe ABC of ABC
the ABC of ABC
Christian Robert
 
GAN(と強化学習との関係)
GAN(と強化学習との関係)GAN(と強化学習との関係)
GAN(と強化学習との関係)
Masahiro Suzuki
 
Curve fitting and Optimization
Curve fitting and OptimizationCurve fitting and Optimization
Curve fitting and Optimization
Syahrul Senin
 

Similar to Variational autoencoders for speech processing d.bielievtsov dataconf 21 04 18 (20)

VAE-type Deep Generative Models
VAE-type Deep Generative ModelsVAE-type Deep Generative Models
VAE-type Deep Generative Models
 
從 VAE 走向深度學習新理論
從 VAE 走向深度學習新理論從 VAE 走向深度學習新理論
從 VAE 走向深度學習新理論
 
Introduction to Deep Generative Models
Introduction to Deep Generative ModelsIntroduction to Deep Generative Models
Introduction to Deep Generative Models
 
A discussion on sampling graphs to approximate network classification functions
A discussion on sampling graphs to approximate network classification functionsA discussion on sampling graphs to approximate network classification functions
A discussion on sampling graphs to approximate network classification functions
 
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
 
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
Paper Summary of Beta-VAE: Learning Basic Visual Concepts with a Constrained ...
 
Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...
 
Auto encoding-variational-bayes
Auto encoding-variational-bayesAuto encoding-variational-bayes
Auto encoding-variational-bayes
 
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
Rebuilding Factorized Information Criterion: Asymptotically Accurate Marginal...
 
Reading revue of "Inferring Multiple Graphical Structures"
Reading revue of "Inferring Multiple Graphical Structures"Reading revue of "Inferring Multiple Graphical Structures"
Reading revue of "Inferring Multiple Graphical Structures"
 
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
QMC Program: Trends and Advances in Monte Carlo Sampling Algorithms Workshop,...
 
Deep generative model.pdf
Deep generative model.pdfDeep generative model.pdf
Deep generative model.pdf
 
Presentation OCIP 2015
Presentation OCIP 2015Presentation OCIP 2015
Presentation OCIP 2015
 
17_monte_carlo.pdf
17_monte_carlo.pdf17_monte_carlo.pdf
17_monte_carlo.pdf
 
Paper Summary of Disentangling by Factorising (Factor-VAE)
Paper Summary of Disentangling by Factorising (Factor-VAE)Paper Summary of Disentangling by Factorising (Factor-VAE)
Paper Summary of Disentangling by Factorising (Factor-VAE)
 
Signals and Systems Homework Help.pptx
Signals and Systems Homework Help.pptxSignals and Systems Homework Help.pptx
Signals and Systems Homework Help.pptx
 
Automatic bayesian cubature
Automatic bayesian cubatureAutomatic bayesian cubature
Automatic bayesian cubature
 
the ABC of ABC
the ABC of ABCthe ABC of ABC
the ABC of ABC
 
GAN(と強化学習との関係)
GAN(と強化学習との関係)GAN(と強化学習との関係)
GAN(と強化学習との関係)
 
Curve fitting and Optimization
Curve fitting and OptimizationCurve fitting and Optimization
Curve fitting and Optimization
 

More from Olga Zinkevych

Overview of text classification approaches algorithms & software v lyubin...
Overview of text classification approaches algorithms & software v lyubin...Overview of text classification approaches algorithms & software v lyubin...
Overview of text classification approaches algorithms & software v lyubin...
Olga Zinkevych
 
Evolution of words through time a malenko dataconf 21 04_18
Evolution of words through time a malenko dataconf 21 04_18Evolution of words through time a malenko dataconf 21 04_18
Evolution of words through time a malenko dataconf 21 04_18
Olga Zinkevych
 
What it takes to build a model for detecting patients that defaults from medi...
What it takes to build a model for detecting patients that defaults from medi...What it takes to build a model for detecting patients that defaults from medi...
What it takes to build a model for detecting patients that defaults from medi...
Olga Zinkevych
 
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
Olga Zinkevych
 
Azure data catalog your data your way eugene polonichko dataconf 21 04 18
Azure data catalog your data your way eugene polonichko dataconf 21 04 18Azure data catalog your data your way eugene polonichko dataconf 21 04 18
Azure data catalog your data your way eugene polonichko dataconf 21 04 18
Olga Zinkevych
 
Aibdconference chat bot for every product Maksym Volchenko
Aibdconference chat bot for every product Maksym VolchenkoAibdconference chat bot for every product Maksym Volchenko
Aibdconference chat bot for every product Maksym Volchenko
Olga Zinkevych
 
Ai&bigdataconference oleksandr saienko machine learning use cases in telecom
Ai&bigdataconference oleksandr saienko machine learning use cases in telecomAi&bigdataconference oleksandr saienko machine learning use cases in telecom
Ai&bigdataconference oleksandr saienko machine learning use cases in telecom
Olga Zinkevych
 
Ai big dataconference_volodymyr getmanskyi colorization distance measuring
Ai big dataconference_volodymyr getmanskyi colorization distance measuringAi big dataconference_volodymyr getmanskyi colorization distance measuring
Ai big dataconference_volodymyr getmanskyi colorization distance measuring
Olga Zinkevych
 
Ai big dataconference_taras firman how to build advanced prediction with addi...
Ai big dataconference_taras firman how to build advanced prediction with addi...Ai big dataconference_taras firman how to build advanced prediction with addi...
Ai big dataconference_taras firman how to build advanced prediction with addi...
Olga Zinkevych
 
Ai big dataconference_sparkinonehour_vitalii bashun
Ai big dataconference_sparkinonehour_vitalii bashunAi big dataconference_sparkinonehour_vitalii bashun
Ai big dataconference_sparkinonehour_vitalii bashun
Olga Zinkevych
 
Ai big dataconference_semantic image segmentatation using word embeddings_ole...
Ai big dataconference_semantic image segmentatation using word embeddings_ole...Ai big dataconference_semantic image segmentatation using word embeddings_ole...
Ai big dataconference_semantic image segmentatation using word embeddings_ole...
Olga Zinkevych
 
Ai big dataconference_ml_fastdata_vitalii bondarenko
Ai big dataconference_ml_fastdata_vitalii bondarenkoAi big dataconference_ml_fastdata_vitalii bondarenko
Ai big dataconference_ml_fastdata_vitalii bondarenko
Olga Zinkevych
 
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutionsAi big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
Olga Zinkevych
 
Ai big dataconference_jeffrey ricker_kappa_architecture
Ai big dataconference_jeffrey ricker_kappa_architectureAi big dataconference_jeffrey ricker_kappa_architecture
Ai big dataconference_jeffrey ricker_kappa_architecture
Olga Zinkevych
 
Ai big dataconference_eugene_polonichko_azure data lake
Ai big dataconference_eugene_polonichko_azure data lake Ai big dataconference_eugene_polonichko_azure data lake
Ai big dataconference_eugene_polonichko_azure data lake
Olga Zinkevych
 

More from Olga Zinkevych (15)

Overview of text classification approaches algorithms & software v lyubin...
Overview of text classification approaches algorithms & software v lyubin...Overview of text classification approaches algorithms & software v lyubin...
Overview of text classification approaches algorithms & software v lyubin...
 
Evolution of words through time a malenko dataconf 21 04_18
Evolution of words through time a malenko dataconf 21 04_18Evolution of words through time a malenko dataconf 21 04_18
Evolution of words through time a malenko dataconf 21 04_18
 
What it takes to build a model for detecting patients that defaults from medi...
What it takes to build a model for detecting patients that defaults from medi...What it takes to build a model for detecting patients that defaults from medi...
What it takes to build a model for detecting patients that defaults from medi...
 
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
Dataservices based on mesos and kafka kostiantyn bokhan dataconf 21 04 18
 
Azure data catalog your data your way eugene polonichko dataconf 21 04 18
Azure data catalog your data your way eugene polonichko dataconf 21 04 18Azure data catalog your data your way eugene polonichko dataconf 21 04 18
Azure data catalog your data your way eugene polonichko dataconf 21 04 18
 
Aibdconference chat bot for every product Maksym Volchenko
Aibdconference chat bot for every product Maksym VolchenkoAibdconference chat bot for every product Maksym Volchenko
Aibdconference chat bot for every product Maksym Volchenko
 
Ai&bigdataconference oleksandr saienko machine learning use cases in telecom
Ai&bigdataconference oleksandr saienko machine learning use cases in telecomAi&bigdataconference oleksandr saienko machine learning use cases in telecom
Ai&bigdataconference oleksandr saienko machine learning use cases in telecom
 
Ai big dataconference_volodymyr getmanskyi colorization distance measuring
Ai big dataconference_volodymyr getmanskyi colorization distance measuringAi big dataconference_volodymyr getmanskyi colorization distance measuring
Ai big dataconference_volodymyr getmanskyi colorization distance measuring
 
Ai big dataconference_taras firman how to build advanced prediction with addi...
Ai big dataconference_taras firman how to build advanced prediction with addi...Ai big dataconference_taras firman how to build advanced prediction with addi...
Ai big dataconference_taras firman how to build advanced prediction with addi...
 
Ai big dataconference_sparkinonehour_vitalii bashun
Ai big dataconference_sparkinonehour_vitalii bashunAi big dataconference_sparkinonehour_vitalii bashun
Ai big dataconference_sparkinonehour_vitalii bashun
 
Ai big dataconference_semantic image segmentatation using word embeddings_ole...
Ai big dataconference_semantic image segmentatation using word embeddings_ole...Ai big dataconference_semantic image segmentatation using word embeddings_ole...
Ai big dataconference_semantic image segmentatation using word embeddings_ole...
 
Ai big dataconference_ml_fastdata_vitalii bondarenko
Ai big dataconference_ml_fastdata_vitalii bondarenkoAi big dataconference_ml_fastdata_vitalii bondarenko
Ai big dataconference_ml_fastdata_vitalii bondarenko
 
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutionsAi big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
Ai big dataconference_krakovetskyi_microsoft ai a new era of smart solutions
 
Ai big dataconference_jeffrey ricker_kappa_architecture
Ai big dataconference_jeffrey ricker_kappa_architectureAi big dataconference_jeffrey ricker_kappa_architecture
Ai big dataconference_jeffrey ricker_kappa_architecture
 
Ai big dataconference_eugene_polonichko_azure data lake
Ai big dataconference_eugene_polonichko_azure data lake Ai big dataconference_eugene_polonichko_azure data lake
Ai big dataconference_eugene_polonichko_azure data lake
 

Recently uploaded

Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdfIron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
RadiNasr
 
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
insn4465
 
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
University of Maribor
 
Wearable antenna for antenna applications
Wearable antenna for antenna applicationsWearable antenna for antenna applications
Wearable antenna for antenna applications
Madhumitha Jayaram
 
Modelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdfModelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdf
camseq
 
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdfBPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
MIGUELANGEL966976
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
thanhdowork
 
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTCHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
jpsjournal1
 
Properties Railway Sleepers and Test.pptx
Properties Railway Sleepers and Test.pptxProperties Railway Sleepers and Test.pptx
Properties Railway Sleepers and Test.pptx
MDSABBIROJJAMANPAYEL
 
Advanced control scheme of doubly fed induction generator for wind turbine us...
Advanced control scheme of doubly fed induction generator for wind turbine us...Advanced control scheme of doubly fed induction generator for wind turbine us...
Advanced control scheme of doubly fed induction generator for wind turbine us...
IJECEIAES
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
anoopmanoharan2
 
New techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdfNew techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdf
wisnuprabawa3
 
Generative AI leverages algorithms to create various forms of content
Generative AI leverages algorithms to create various forms of contentGenerative AI leverages algorithms to create various forms of content
Generative AI leverages algorithms to create various forms of content
Hitesh Mohapatra
 
ML Based Model for NIDS MSc Updated Presentation.v2.pptx
ML Based Model for NIDS MSc Updated Presentation.v2.pptxML Based Model for NIDS MSc Updated Presentation.v2.pptx
ML Based Model for NIDS MSc Updated Presentation.v2.pptx
JamalHussainArman
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
Dr Ramhari Poudyal
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
SyedAbiiAzazi1
 
22CYT12-Unit-V-E Waste and its Management.ppt
22CYT12-Unit-V-E Waste and its Management.ppt22CYT12-Unit-V-E Waste and its Management.ppt
22CYT12-Unit-V-E Waste and its Management.ppt
KrishnaveniKrishnara1
 
A review on techniques and modelling methodologies used for checking electrom...
A review on techniques and modelling methodologies used for checking electrom...A review on techniques and modelling methodologies used for checking electrom...
A review on techniques and modelling methodologies used for checking electrom...
nooriasukmaningtyas
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
heavyhaig
 
basic-wireline-operations-course-mahmoud-f-radwan.pdf
basic-wireline-operations-course-mahmoud-f-radwan.pdfbasic-wireline-operations-course-mahmoud-f-radwan.pdf
basic-wireline-operations-course-mahmoud-f-radwan.pdf
NidhalKahouli2
 

Recently uploaded (20)

Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdfIron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
Iron and Steel Technology Roadmap - Towards more sustainable steelmaking.pdf
 
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
 
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
Presentation of IEEE Slovenia CIS (Computational Intelligence Society) Chapte...
 
Wearable antenna for antenna applications
Wearable antenna for antenna applicationsWearable antenna for antenna applications
Wearable antenna for antenna applications
 
Modelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdfModelagem de um CSTR com reação endotermica.pdf
Modelagem de um CSTR com reação endotermica.pdf
 
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdfBPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
BPV-GUI-01-Guide-for-ASME-Review-Teams-(General)-10-10-2023.pdf
 
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
 
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTCHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
 
Properties Railway Sleepers and Test.pptx
Properties Railway Sleepers and Test.pptxProperties Railway Sleepers and Test.pptx
Properties Railway Sleepers and Test.pptx
 
Advanced control scheme of doubly fed induction generator for wind turbine us...
Advanced control scheme of doubly fed induction generator for wind turbine us...Advanced control scheme of doubly fed induction generator for wind turbine us...
Advanced control scheme of doubly fed induction generator for wind turbine us...
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
 
New techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdfNew techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdf
 
Generative AI leverages algorithms to create various forms of content
Generative AI leverages algorithms to create various forms of contentGenerative AI leverages algorithms to create various forms of content
Generative AI leverages algorithms to create various forms of content
 
ML Based Model for NIDS MSc Updated Presentation.v2.pptx
ML Based Model for NIDS MSc Updated Presentation.v2.pptxML Based Model for NIDS MSc Updated Presentation.v2.pptx
ML Based Model for NIDS MSc Updated Presentation.v2.pptx
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
 
14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application14 Template Contractual Notice - EOT Application
14 Template Contractual Notice - EOT Application
 
22CYT12-Unit-V-E Waste and its Management.ppt
22CYT12-Unit-V-E Waste and its Management.ppt22CYT12-Unit-V-E Waste and its Management.ppt
22CYT12-Unit-V-E Waste and its Management.ppt
 
A review on techniques and modelling methodologies used for checking electrom...
A review on techniques and modelling methodologies used for checking electrom...A review on techniques and modelling methodologies used for checking electrom...
A review on techniques and modelling methodologies used for checking electrom...
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
 
basic-wireline-operations-course-mahmoud-f-radwan.pdf
basic-wireline-operations-course-mahmoud-f-radwan.pdfbasic-wireline-operations-course-mahmoud-f-radwan.pdf
basic-wireline-operations-course-mahmoud-f-radwan.pdf
 

Variational autoencoders for speech processing d.bielievtsov dataconf 21 04 18

  • 1. Variational Autoencoders for Speech Dmytro Bielievtsov 2018
  • 4. Generative models Generative models model the distribution of high-dimensional objects like images, audio, etc. They model these distributions either by allowing you to sample from the distribution or by computing likelihoods Three families of generative model dominate the deep learning landscape today: autoregressive, GANs, and VAEs
  • 5. Generative model comparison Autoregressive models can be very accurate (WaveNet) but are slow to generate (but cf. Distilled WaveNet), and produce no latent representation of the data GANs produce a latent representation of the data but are finicky to train and don’t give likelihoods (at least in their pure form) VAEs lack the disadvantages of autoregressive models and GANs, but in their basic form are prone to producing blurry samples
  • 6. Disentangled representations Each latent controls one “factor of variation”. You don’t have one latent controlling, e.g., pitch + volume and another pitch - volume The factor of variation each latent controls is independent of the setting of the other latents
  • 9. Latent variable models, easy things to do Drawing samples is easy: z ∼ pθ(z), x ∼ pθ(x | z) As is computing joint likelihoods
  • 10. Latent variable models, hard things to do To train the model, we need to maximize pθ(x) We also need pθ(x) if we want to find likelihoods of samples To find latent codes of samples (e.g., to do morphing), we need to know pθ(z | x) Unfortunately, pθ(x) and pθ(z | x) are intractible
  • 11. Latent variable models, intractable integrals pθ(x) = pθ(z)pθ(x | z) dz pθ(z | x) = pθ(z)pθ(x | z)/pθ(x) Since z is high-dimensional, pθ(x) and pθ(z | x) are intractable By the way, from now on we will drop θ from our notation for less clutter
  • 12. The idea of an encoder For fixed x, for almost all values of z, p(z | x) will be very nearly zero If we could somehow train a network to approximate p(z | x), maybe we could estimate p(x) much more efficiently by sampling z values that are likely to generate something close to x
  • 14. The encoder network Instead of using fixed variances, as with the decoder network, the encoder network also produces variances Intuitively, we want the network to find an efficient representation of the data, so we penalize it for outputting very small variances
  • 15. The fundamental VAE equation Start with the definition of KL-divergence and then use Bayes’ Law: D (q(z | x) p(z | x)) = Ez∼q(z|x)[log q(z | x) − log p(z | x)] = Ez∼q(z|x)[log q(z | x) − log p(x | z) − log p(z)] + log p(x) Rearranging terms and using the definition of KL-divergence, OELBO := log p(x) − D (q(z | x) p(z | x)) = Ez∼q(z|x)[log p(x | z)] tractable to optimize − D (q(z | x) p(z)) tractable to optimize
  • 16. Recap pθ(x | z) := N(x | fθ(z), σ2 I) p(z) := N(z | 0, I) qφ(z | x) := N(x | gφ(x), diag(hφ(x)) OELBO(x; θ, φ) := Ez∼qφ(z|x)[log pθ(x | z)] − D (qφ(z | x) p(z))
  • 17. A picture, and the reparameterization trick
  • 18. Interpretability of latent variables Latent variables produced by vanilla VAE’s are usually meaningless when considered individually. Adapted from Higgins et al. 2017
  • 19. Disentangling variables Simple approaches to encourage disentangled latent variables: β-VAE factor-VAE
  • 20. Disentangling variables: β-VAE Original objective function (ELBO): OELBO = Ez∼q(z|x)[log p(x | z)] − D (q(z | x) p(z))
  • 21. Disentangling variables: β-VAE Original objective function (ELBO): OELBO = Ez∼q(z|x)[log p(x | z)] − D (q(z | x) p(z)) New objective function: Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
  • 22. Disentangling variables: β-VAE Adapted from Higgins et al. 2017
  • 23. Disentangling variables: factor-VAE Upon closer inspection, β-VAE creates some tradeoff between disentanglement and reconstruction: β-VAE objective: Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z))
  • 24. Disentangling variables: factor-VAE Upon closer inspection, β-VAE creates some tradeoff between disentanglement and reconstruction: β-VAE objective: Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z)) Let’s look closer at that regularizer: Ex∼pdata(x)[D (q(z | x) p(z))] = Ienc(x; z) + D (q(z) p(z))
  • 25. Disentangling variables: factor-VAE Upon closer inspection, β-VAE creates some tradeoff between disentanglement and reconstruction: β-VAE objective: Oβ = Ez∼q(z|x)[log p(x | z)] − β D (q(z | x) p(z)) Let’s look closer at that regularizer: Ex∼pdata(x)[D (q(z | x) p(z))] = Ienc(x; z) + D (q(z) p(z)) We don’t want to hurt that mutual information term!
  • 26. Disentangling variables: factor-VAE Yet another variation on the ELBO objective: Oγ = OELBO − γD (q(z) ¯q(z))
  • 27. Disentangling variables: factor-VAE Yet another variation on the ELBO objective: Oγ = OELBO − γD (q(z) ¯q(z)) Adapted from Kim et al. 2018
  • 28. Disentangling variables: factor-VAE Yet another variation on the ELBO objective: Oγ = OELBO − γD (q(z) ¯q(z)) Adapted from Kim et al. 2018
  • 29. VQ-VAE, quick look Adapted from Oord et al. 2017
  • 30. References Doersch, Carl. “Tutorial on Variational Autoencoders.” ArXiv:1606.05908 [Cs, Stat], June 19, 2016. http://arxiv.org/abs/1606.05908. Higgins, Irina, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner. “β-VAE: LEARNING BASIC VISUAL CONCEPTS WITH A CONSTRAINED VARIATIONAL FRAMEWORK,” 2017, 22. Hsu, Wei-Ning, Yu Zhang, and James Glass. “Learning Latent Representations for Speech Generation and Transformation.” ArXiv:1704.04222 [Cs, Stat], April 13, 2017. http://arxiv.org/abs/1704.04222. ———. “Unsupervised Learning of Disentangled and Interpretable Representations from Sequential Data.” ArXiv:1709.07902 [Cs, Eess, Stat], September 22, 2017. http://arxiv.org/abs/1709.07902. Kim, Hyunjik, and Andriy Mnih. “Disentangling by Factorising.” ArXiv:1802.05983 [Cs, Stat], February 16, 2018. http://arxiv.org/abs/1802.05983. Kingma, Diederik P., and Max Welling. “Auto-Encoding Variational Bayes.” ArXiv:1312.6114 [Cs, Stat], December 20, 2013. http://arxiv.org/abs/1312.6114. Oord, Aaron van den, Oriol Vinyals, and Koray Kavukcuoglu. “Neural Discrete Representation Learning.” ArXiv:1711.00937 [Cs], November 2, 2017. http://arxiv.org/abs/1711.00937.