• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 58
  • 4
  • 2
  • 2
  • Tagged with
  • 87
  • 87
  • 68
  • 41
  • 37
  • 34
  • 29
  • 29
  • 28
  • 27
  • 27
  • 26
  • 23
  • 23
  • 22
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
71

Navigating the Metric Zoo: Towards a More Coherent Model For Quantitative Evaluation of Generative ML Models

Dozier, Robbie 26 August 2022 (has links)
No description available.
72

Deep Scenario Generation of Financial Markets / Djup scenario generering av finansiella marknader

Carlsson, Filip, Lindgren, Philip January 2020 (has links)
The goal of this thesis is to explore a new clustering algorithm, VAE-Clustering, and examine if it can be applied to find differences in the distribution of stock returns and augment the distribution of a current portfolio of stocks and see how it performs in different market conditions. The VAE-clustering method is as mentioned a newly introduced method and not widely tested, especially not on time series. The first step is therefore to see if and how well the clustering works. We first apply the algorithm to a dataset containing monthly time series of the power demand in Italy. The purpose in this part is to focus on how well the method works technically. When the model works well and generates proper results with the Italian Power Demand data, we move forward and apply the model on stock return data. In the latter application we are unable to find meaningful clusters and therefore unable to move forward towards the goal of the thesis. The results shows that the VAE-clustering method is applicable for time series. The power demand have clear differences from season to season and the model can successfully identify those differences. When it comes to the financial data we hoped that the model would be able to find different market regimes based on time periods. The model is though not able distinguish different time periods from each other. We therefore conclude that the VAE-clustering method is applicable on time series data, but that the structure and setting of the financial data in this thesis makes it to hard to find meaningful clusters. The major finding is that the VAE-clustering method can be applied to time series. We highly encourage further research to find if the method can be successfully used on financial data in different settings than tested in this thesis. / Syftet med den här avhandlingen är att utforska en ny klustringsalgoritm, VAE-Clustering, och undersöka om den kan tillämpas för att hitta skillnader i fördelningen av aktieavkastningar och förändra distributionen av en nuvarande aktieportfölj och se hur den presterar under olika marknadsvillkor. VAE-klusteringsmetoden är som nämnts en nyinförd metod och inte testad i stort, särskilt inte på tidsserier. Det första steget är därför att se om och hur klusteringen fungerar. Vi tillämpar först algoritmen på ett datasätt som innehåller månatliga tidsserier för strömbehovet i Italien. Syftet med denna del är att fokusera på hur väl metoden fungerar tekniskt. När modellen fungerar bra och ger tillfredställande resultat, går vi vidare och tillämpar modellen på aktieavkastningsdata. I den senare applikationen kan vi inte hitta meningsfulla kluster och kan därför inte gå framåt mot målet som var att simulera olika marknader och se hur en nuvarande portfölj presterar under olika marknadsregimer. Resultaten visar att VAE-klustermetoden är väl tillämpbar på tidsserier. Behovet av el har tydliga skillnader från säsong till säsong och modellen kan framgångsrikt identifiera dessa skillnader. När det gäller finansiell data hoppades vi att modellen skulle kunna hitta olika marknadsregimer baserade på tidsperioder. Modellen kan dock inte skilja olika tidsperioder från varandra. Vi drar därför slutsatsen att VAE-klustermetoden är tillämplig på tidsseriedata, men att strukturen på den finansiella data som undersöktes i denna avhandling gör det svårt att hitta meningsfulla kluster. Den viktigaste upptäckten är att VAE-klustermetoden kan tillämpas på tidsserier. Vi uppmuntrar ytterligare forskning för att hitta om metoden framgångsrikt kan användas på finansiell data i andra former än de testade i denna avhandling
73

Attribute Embedding for Variational Auto-Encoders : Regularization derived from triplet loss / Inbäddning av attribut för Variationsautokodare : Strukturering av det Latenta Rummet

E. L. Dahlin, Anton January 2022 (has links)
Techniques for imposing a structure on the latent space of neural networks have seen much development in recent years. Clustering techniques used for classification have been used to great success, and with this work we hope to bridge the gap between contrastive losses and Generative models. We introduce an embedding loss derived from Triplet loss to show that attributes and information can be clustered in specific dimensions in the latent space of Variational Auto-Encoders. This allows control over the embedded attributes via manipulation of these latent space dimensions. This work also serves to take steps towards the usage of any data augmentation when applying Triplet loss to Variational Auto-Encoders. In this work three different Variational Auto-Encoders are trained on three different datasets to embed information in three different ways using this novel method. Our results show the method working to varying degrees depending on the implementation and the information embedded. Two experiments using image data and one using waveform audio shows that the method is modality invariant. / Tekniker för att införa en struktur i det latenta utrymmet i neurala nätverk har sett mycket utveckling under de senaste åren. Kluster metoder som används för klassificering har använts till stor framgång, och med detta arbete hoppas vi kunna brygga gapet mellan kontrastiva förlustfunktioner och generativa modeller. Vi introducerar en förlustfunktion för inbäddning härledd från triplet loss för att visa att attribut och information kan klustras i specifika dimensioner i det latenta utrymmet hos variationsautokodare. Detta tillåter kontroll över de inbäddade attributen via manipulering av dessa dimensioner i latenta utrymmet. Detta arbete tjänar också till att ta steg mot användningen av olika data augmentationer när triplet loss tillämpas på generativa modeller. Tre olika Variationsautokodare tränas på tre olika dataset för att bädda in information på tre olika sätt med denna nya metod. Våra resultat visar att metoden fungerar i varierande grad beroende på hur den tillämpas och vilken information som inbäddas. Två experiment använder bild-data och ett använder sig av ljud, vilket visar på att metoden är modalitetsinvariant.
74

Generative Image-to-Image Translation with Applications in Computational Pathology

Fangda Li (17272816) 24 October 2023 (has links)
<p dir="ltr">Generative Image-to-Image Translation (I2IT) involves transforming an input image from one domain to another. Typically, this transformation retains the content in the input image while adjusting the domain-dependent style elements. Generative I2IT finds utility in a wide range of applications, yet its effectiveness hinges on adaptations to the unique characteristics of the data at hand. This dissertation pushes the boundaries of I2IT by applying it to stain-related problems in computational pathology. Particularly, the main contributions span two major applications of stain translation: H&E-to-H&E and H&E-to-IHC, each with its unique requirements and challenges. More specifically, the first contribution addresses the generalization challenge posed by the high variability in H&E stain appearances to any task-specific machine learning models. To this end, the Generative Stain Augmentation Network (G-SAN) is introduced to augment the training images in any downstream task with random and diverse H&E stain appearances. Experimental results demonstrate G-SAN’s ability to enhance model generalization across stain variations in downstream tasks. The second key contribution in this dissertation focuses on H&E-to-IHC stain translation. The major challenge in learning accurate H&E-to-IHC stain translation is the frequent and sometimes severe inconsistencies in the groundtruth H&E-IHC image pairs. To make training more robust to these inconsistencies, a novel contrastive learning based loss, named the Adaptive Supervised PatchNCE (ASP) loss is presented. Experimental results suggest that the proposed ASP-based framework outperforms the state-of-the-art in H&E-to-IHC stain translation by significant margins. Additionally, a new dataset for H&E-to-IHC translation – the Multi-IHC Stain Translation (MIST) dataset, is released to the public, featuring paired images from H&E to four different IHC stains. For future directions of generative I2IT in stain translation problems, a proof-of-concept study of applying the latest diffusion model based I2IT methods to the problem of virtual H&E staining is presented.</p>
75

Understanding, improving, and generalizing generative models

Jolicoeur-Martineau, Alexia 08 1900 (has links)
Les modèles génératifs servent à générer des échantillons d'une loi de probabilité (ex. : du texte, des images, de la musique, des vidéos, des molécules, et beaucoup plus) à partir d'un jeu de données (ex. : une banque d'images, de texte, ou autre). Entrainer des modèles génératifs est une tâche très difficile, mais ces outils ont un très grand potentiel en termes d'applications. Par exemple, dans le futur lointain, on pourrait envisager qu'un modèle puisse générer les épisodes d'une émission de télévision à partir d'un script et de voix générés par d'autres modèles génératifs. Il existe plusieurs types de modèles génératifs. Pour la génération d'images, l'approche la plus fructueuse est sans aucun doute la méthode de réseaux adverses génératifs (GANs). Les GANs apprennent à générer des images par un jeu compétitif entre deux joueurs, le Discriminateur et le Générateur. Le Discriminateur tente de prédire si une image est vraie ou fausse, tandis que le Générateur tente de générer des images plus réalistes en apprenant à faire croire au discriminateur que ces fausses images générées sont vraies. En complétant ce jeu, les GANs arrivent à générer des images presque photo-réalistes. Il est souvent possible pour des êtres humains de distinguer les fausses images (générés par les GANs) des vraies images (ceux venant du jeu de données), mais la tâche devient plus difficile au fur et à mesure que cette technologie s'améliore. Le plus gros défaut des GANs est que les données générées par les GANs manquent souvent de diversité (ex. : les chats au visage aplati sont rares dans la banque d'images, donc les GANs génèrent juste des races de chats plus fréquentes). Ces méthodes souvent aussi souvent très instables. Il y a donc encore beaucoup de chemin à faire avant l'obtention d'images parfaitement photo-réalistes et diverses. De nouvelles méthodes telles que les modèles de diffusion à la base de score semblent produire de meilleurs résultats que les GANs, donc tout n'est pas gagné pour les GANs. C'est pourquoi cette thèse n'est pas concentrée seulement sur les GANs, mais aussi sur les modèles de diffusion. Notez que cette thèse est exclusivement concentrée sur la génération de données continues (ex. : images, musique, vidéos) plutôt que discrètes (ex. : texte), car cette dernière fait usage de méthodes complètement différentes. Le premier objectif de cette thèse est d'étudier les modèles génératifs de façon théorique pour mieux les comprendre. Le deuxième objectif de cette thèse est d'inventer de nouvelles astuces (nouvelles fonctions objectives, régularisations, architectures, etc.) permettant d'améliorer les modèles génératifs. Le troisième objectif est de généraliser ces approches au-delà de leur formulation initiale, pour permettre la découverte de nouveaux liens entre différentes approches. Ma première contribution est de proposer un discriminateur relativiste qui estime la probabilité qu'une donnée réelle, soit plus réaliste qu'une donnée fausse (inventée par un modèle générateur). Les GANs relativistes forment une nouvelle classe de fonctions de perte qui apportent beaucoup de stabilité durant l'entrainement. Ma seconde contribution est de prouver que les GANs relativistes forment une mesure de dissimilarité. Ma troisième contribution est de concevoir une variante adverse au appariement de score pour produire des données de meilleure qualité avec les modèles de diffusion. Ma quatrième contribution est d'améliorer la vitesse de génération des modèles de diffusion par la création d'une méthode numérique de résolution pour équations différentielles stochastiques (SDEs). / Generative models are powerful tools to generate samples (e.g., images, music, text) from an unknown distribution given a finite set of examples. Generative models are hard to train successfully, but they have the potential to revolutionize arts, science, and business. These models can generate samples from various data types (e.g., text, images, audio, videos, 3d). In the future, we can envision generative models being used to create movies or episodes from a TV show given a script (possibly also generated by a generative model). One of the most successful methods for generating images is Generative Adversarial Networks (GANs). This approach consists of a game between two players, the Discriminator and the Generator. The goal of the Discriminator is to classify an image as real or fake, while the Generator attempts to fool the Discriminator into thinking that the fake images it generates are real. Through this game, GANs are able to generate very high-quality samples, such as photo-realistic images. Humans are still generally able to distinguish real images (from the training dataset) from fake images (generated by GANs), but the gap is lessening as GANs become better over time. The biggest weakness of GANs is that they have trouble generating diverse data representative of the full range of the data distribution. Thus, there is still much progress to be made before GANs reach their full potential. New methods performing better than GANs are also appearing. One prime example is score-based diffusion models. This thesis focuses on generative models that seemed promising at the time for continuous data generation: GANs and score-based diffusion models. I seek to improve generative models so that they reach their full potential (Objective 1: Improving) and to understand these approaches better on a theoretical level (Objective 2: Theoretical understanding). I also want to generalize these approaches beyond their original setting (Objective 3: Generalizing), allowing the discovery of new connections between different concepts/fields. My first contribution is to propose using a relativistic discriminator, which estimates the probability that a given real data is more realistic than a randomly sampled fake data. Relativistic GANs form a new class of GAN loss functions that are much more stable with respect to optimization hyperparameters. My second contribution is to take a more rigorous look at relativistic GANs and prove that they are proper statistical divergences. My third contribution is to devise an adversarial variant to denoising score matching, which leads to higher quality data with score-based diffusion models. My fourth contribution is to significantly improve the speed of score-based diffusion models through a carefully devised Stochastic Differential Equation (SDE) solver.
76

Expressivité et contrôle de modèles d’apprentissage automatique dans un corpus d’installations audiovisuelles

Lavoie Viau, Gabriel 12 1900 (has links)
L’appropriation d’algorithmes existants, la création d’outils numériques et des recherches conceptuelles ont mené à la création de deux installations audiovisuelles interactives. La première, Deep Duo, met en scène des réseaux de neurones artificiels contrôlant des synthétiseurs modulaires. La deuxième, Morphogenèse, l’œuvre d’envergure de ce mémoire, met en relation le spectateur avec des modèles profonds génératifs et le place face à des représentations artificielles de sa voix et de son visage. Les installations et leurs fonctionnements seront décrits et, à travers des exemples de stratégies créatives et des concepts théoriques en lien avec l’interactivité et l’esthétique des comportements, des pistes pour favoriser l’utilisation d’algorithmes d’apprentissage automatique à des fins créatives seront proposées. / The appropriation of existing algorithms, the creation of digital tools and conceptual research have led to the creation of two interactive audiovisual installations. The first, Deep Duo, features artificial neural networks controlling modular synthesizers. The second, Morphogenesis, the major work of this dissertation, connects the viewer with generative deep models and places them in front of artificial representations of their voice and face. We will describe these installations and their functioning and, through examples of creative strategies and theoretical concepts related to interactivity and the aesthetics of behaviour, we will propose ways to promote the use of machine learning algorithms for creative purposes.
77

Basil-GAN / Basilika-GAN

Risberg, Jonatan January 2022 (has links)
Developments in computer vision has sought to design deep neural networks which trained on a large set of images are able to generate high quality artificial images which share semantic qualities with the original image set. A pivotal shift was made with the introduction of the generative adversarial network (GAN) by Goodfellow et al.. Building on the work by Goodfellow more advanced models using the same idea have shown great improvements in terms of both image quality and data diversity. GAN models generate images by feeding samples from a vector space into a generative neural network. The structure of these so called latent vector samples show to correspond to semantic similarities of their corresponding generated images. In this thesis the DCGAN model is trained on a novel data set consisting of image sequences of the growth process of basil plants from germination to harvest. We evaluate the trained model by comparing the DCGAN performance on benchmark data sets such as MNIST and CIFAR10 and conclude that the model trained on the basil plant data set achieved similar results compared to the MNIST data set and better results in comparison to the CIFAR10 data set. To argue for the potential of using more advanced GAN models we compare the results from the DCGAN model with the contemporary StyleGAN2 model. We also investigate the latent vector space produced by the DCGAN model and confirm that in accordance with previous research, namely that the DCGAN model is able to generate a latent space with data specific semantic structures. For the DCGAN model trained on the data set of basil plants, the latent space is able to distinguish between images of early stage basil plants from late stage plants in the growth phase. Furthermore, utilizing the sequential semantics of the basil plant data set, an attempt at generating an artificial growth sequence is made using linear interpolation. Finally we present an unsuccessful attempt at visualising the latent space produced by the DCGAN model using a rudimentary approach at inverting the generator network function. / Utvecklingen inom datorseende har syftat till att utforma djupa neurala nätverk som tränas på en stor mängd bilder och kan generera konstgjorda bilder av hög kvalitet med samma semantiska egenskaper som de ursprungliga bilderna. Ett avgörande skifte skedde när Goodfellow et al. introducerade det generativa adversariella nätverket (GAN). Med utgångspunkt i Goodfellows arbete har flera mer avancerade modeller som använder samma idé uppvisat stora förbättringar när det gäller både bildkvalitet och datamångfald. GAN-modeller genererar bilder genom att mata in vektorer från ett vektorrum till ett generativt neuralt nätverk. Strukturen hos dessa så kallade latenta vektorer visar sig motsvara semantiska likheter mellan motsvarande genererade bilder. I detta examensarbete har DCGAN-modellen tränats på en ny datamängd som består av bildsekvenser av basilikaplantors tillväxtprocess från groning till skörd. Vi utvärderar den tränade modellen genom att jämföra DCGAN-modellen mot referensdataset som MNIST och CIFAR10 och drar slutsatsen att DCGAN tränad på datasetet för basilikaväxter uppnår liknande resultat jämfört med MNIST-dataset och bättre resultat jämfört med CIFAR10-datasetet. För att påvisa potentialen av att använda mer avancerade GAN-modeller jämförs resultaten från DCGAN-modellen med den mer avancerade StyleGAN2-modellen. Vi undersöker också det latenta vektorrum som produceras av DCGAN-modellen och bekräftar att DCGAN-modellen i enlighet med tidigare forskning kan generera ett latent rum med dataspecifika semantiska strukturer. För DCGAN-modellen som tränats på datamängden med basilikaplantor lyckas det latenta rummet skilja mellan bilder av basilikaplantor i tidiga stadier och sena stadier av plantor i tillväxtprocessen. Med hjälp av den sekventiella semantiken i datamängden för basilikaväxter gjörs dessutom ett försök att generera en artificiell tillväxtsekvens med hjälp av linjär interpolation. Slutligen presenterar vi ett misslyckat försök att visualisera det latenta rummet som produceras av DCGAN-modellen med hjälp av ett rudimentärt tillvägagångssätt för att invertera den generativa nätverksfunktionen.
78

Some phenomenological investigations in deep learning

Baratin, Aristide 12 1900 (has links)
Les remarquables performances des réseaux de neurones profonds dans de nombreux domaines de l'apprentissage automatique au cours de la dernière décennie soulèvent un certain nombre de questions théoriques. Par exemple, quels mecanismes permettent à ces reseaux, qui ont largement la capacité de mémoriser entièrement les exemples d'entrainement, de généraliser correctement à de nouvelles données, même en l'absence de régularisation explicite ? De telles questions ont fait l'objet d'intenses efforts de recherche ces dernières années, combinant analyses de systèmes simplifiés et études empiriques de propriétés qui semblent être corrélées à la performance de généralisation. Les deux premiers articles présentés dans cette thèse contribuent à cette ligne de recherche. Leur but est de mettre en évidence et d'etudier des mécanismes de biais implicites permettant à de larges modèles de prioriser l'apprentissage de fonctions "simples" et d'adapter leur capacité à la complexité du problème. Le troisième article aborde le problème de l'estimation de information mutuelle en haute, en mettant à profit l'expressivité et la scalabilité des reseaux de neurones profonds. Il introduit et étudie une nouvelle classe d'estimateurs, dont il présente plusieurs applications en apprentissage non supervisé, notamment à l'amélioration des modèles neuronaux génératifs. / The striking empirical success of deep neural networks in machine learning raises a number of theoretical puzzles. For example, why can they generalize to unseen data despite their capacity to fully memorize the training examples? Such puzzles have been the subject of intense research efforts in the past few years, which combine rigorous analysis of simplified systems with empirical studies of phenomenological properties shown to correlate with generalization. The first two articles presented in these thesis contribute to this line of work. They highlight and discuss mechanisms that allow large models to prioritize learning `simple' functions during training and to adapt their capacity to the complexity of the problem. The third article of this thesis addresses the long standing problem of estimating mutual information in high dimension, by leveraging the scalability of neural networks. It introduces and studies a new class of estimators and present several applications in unsupervised learning, especially on enhancing generative models.
79

Exploring Normalizing Flow Modifications for Improved Model Expressivity / Undersökning av normalizing flow-modifikationer för förbättrad modelluttrycksfullhet

Juschak, Marcel January 2023 (has links)
Normalizing flows represent a class of generative models that exhibit a number of attractive properties, but do not always achieve state-of-the-art performance when it comes to perceived naturalness of generated samples. To improve the quality of generated samples, this thesis examines methods to enhance the expressivity of discrete-time normalizing flow models and thus their ability to capture different aspects of the data. In the first part of the thesis, we propose an invertible neural network architecture as an alternative to popular architectures like Glow that require an individual neural network per flow step. Although our proposal greatly reduces the number of parameters, it has not been done before, as such architectures are believed to not be powerful enough. For this reason, we define two optional extensions that could greatly increase the expressivity of the architecture. We use augmentation to add Gaussian noise variables to the input to achieve arbitrary hidden-layer widths that are no longer dictated by the dimensionality of the data. Moreover, we implement Piecewise Affine Activation Functions that represent a generalization of Leaky ReLU activations and allow for more powerful transformations in every individual step. The resulting three models are evaluated on two simple synthetic datasets – the two moons dataset and one generated from a mixture of eight Gaussians. Our findings indicate that the proposed architectures cannot adequately model these simple datasets and thus do not represent alternatives to current stateof-the-art models. The Piecewise Affine Activation Function significantly improved the expressivity of the invertible neural network, but could not make use of its full potential due to inappropriate assumptions about the function’s input distribution. Further research is needed to ensure that the input to this function is always standard normal distributed. We conducted further experiments with augmentation using the Glow model and could show minor improvements on the synthetic datasets when only few flow steps (two, three or four) were used. However, in a more realistic scenario, the model would encompass many more flow steps. Lastly, we generalized the transformation in the coupling layers of modern flow architectures from an elementwise affine transformation to a matrixbased affine transformation and studied the effect this had on MoGlow, a flow-based model of motion. We could show that McMoGlow, our modified version of MoGlow, consistently achieved a better training likelihood than the original MoGlow on human locomotion data. However, a subjective user study found no statistically significant difference in the perceived naturalness of the samples generated. As a possible reason for this, we hypothesize that the improvements are subtle and more visible in samples that exhibit slower movements or edge cases which may have been underrepresented in the user study. / Normalizing flows representerar en klass av generativa modeller som besitter ett antal eftertraktade egenskaper, men som inte alltid uppnår toppmodern prestanda när det gäller upplevd naturlighet hos genererade data. För att förbättra kvaliteten på dessa modellers utdata, undersöker detta examensarbete metoder för att förbättra uttrycksfullheten hos Normalizing flows-modeller i diskret tid, och därmed deras förmåga att fånga olika aspekter av datamaterialet. I den första delen av uppsatsen föreslår vi en arkitektur uppbyggt av ett inverterbart neuralt nätverk. Vårt förslag är ett alternativ till populära arkitekturer som Glow, vilka kräver individuella neuronnät för varje flödessteg. Även om vårt förslag kraftigt minskar antalet parametrar har detta inte gjorts tidigare, då sådana arkitekturer inte ansetts vara tillräckligt kraftfulla. Av den anledningen definierar vi två oberoende utökningar till arkitekturen som skulle kunna öka dess uttrycksfullhet avsevärt. Vi använder så kallad augmentation, som konkatenerar Gaussiska brusvariabler till observationsvektorerna för att uppnå godtyckliga bredder i de dolda lagren, så att deras bredd inte längre begränsas av datadimensionaliteten. Dessutom implementerar vi Piecewise Affine Activation-funktioner (PAAF), vilka generaliserar Leaky ReLU-aktiveringar genom att möjliggöra mer kraftfulla transformationer i varje enskilt steg. De resulterande tre modellerna utvärderas med hjälp av två enkla syntetiska datamängder - ”the two moons dataset” och ett som genererats genom att blanda av åtta Gaussfördelningar. Våra resultat visar att de föreslagna arkitekturerna inte kan modellera de enkla datamängderna på ett tillfredsställande sätt, och därmed inte utgör kompetitiva alternativ till nuvarande moderna modeller. Den styckvisa aktiveringsfunktionen förbättrade det inverterbara neurala nätverkets uttrycksfullhet avsevärt, men kunde inte utnyttja sin fulla potential på grund av felaktiga antaganden om funktionens indatafördelning. Ytterligare forskning behövs för att hantera detta problem. Vi genomförde ytterligare experiment med augmentation av Glow-modellen och kunde påvisa vissa förbättringar på de syntetiska dataseten när endast ett fåtal flödessteg (två, tre eller fyra) användes. Däremot omfattar modeller i mer realistiska scenarion många fler flödessteg. Slutligen generaliserade vi transformationen i kopplingslagren hos moderna flödesarkitekturer från en elementvis affin transformation till en matrisbaserad affin transformation, samt studerade vilken effekt detta hade på MoGlow, en flödesbaserad modell av 3D-rörelser. Vi kunde visa att McMoGlow, vår modifierade version av MoGlow, konsekvent uppnådde bättre likelihood i träningen än den ursprungliga MoGlow gjorde på mänskliga rörelsedata. En subjektiv användarstudie på exempelrörelser genererade från MoGlow och McMoGlow visade dock ingen statistiskt signifikant skillnad i användarnas uppfattning av hur naturliga rörelserna upplevdes. Som en möjlig orsak till detta antar vi att förbättringarna är subtila och mer synliga i situationer som uppvisar långsammare rörelser eller i olika gränsfall som kan ha varit underrepresenterade i användarstudien.
80

Fast high-dimensional posterior inference with deep generative models : application to CMB delensing

Sotoudeh, Mohammad-Hadi 08 1900 (has links)
Nous vivons à une époque marquée par une abondance de données cosmologiques de haute résolution. Cet afflux de données engendré par les missions d'observation de nouvelle génération au sol et dans l'espace porte le potentiel de remodeler fondamentalement notre compréhension de l'univers et de ses principes physiques sous-jacents. Cependant, la complexité grande des données observées pose des défis aux approches conventionnelles d'analyse de données, soit en raison de coûts de calcul irréalisables, soit en raison des hypothèses simplificatrices utilisées dans ces algorithmes qui deviennent inadéquates dans des contextes haute résolution à faible bruit, conduisant à des résultats sous-optimaux. En réponse, la communauté scientifique s'est tournée vers des méthodes innovantes d'analyse de données, notamment les techniques d'apprentissage automatique (ML). Les modèles de ML, lorsqu'ils sont bien entraînés, peuvent identifier de manière autonome des correlations significatives dans les données de manière plus efficace et sans hypothèses restrictives inutiles. Bien que les méthodes de ML aient montré des promesses en astrophysique, elles présentent également des problèmes tels que le manque d'interprétabilité, les biais cachés et les estimations d'incertitude non calibrées, ce qui, jusqu'a maintenant, a entrave leur application dans d'importantes découvertes scientifiques. Ce projet s'inscrit dans le cadre de la collaboration "Learning the Universe" (LtU), axée sur la reconstruction des conditions initiales de l'univers, en utilisant une approche de modélisation bayésienne et en exploitant la puissance du ML. L'objectif de ce projet est de développer un cadre pour mener une inférence bayésienne au niveau des pixels dans des problèmes multidimensionnels. Dans cette thèse, je présente le développement d'un cadre d'apprentissage profond pour un échantillonnage rapide des postérieurs en dimensions élevées. Ce cadre utilise l'architecture "Hierarchical Probabilistic U-Net", qui combine la puissance de l'architecture U-Net dans l'apprentissage de cartes multidimensionnelles avec le rigoureux cadre d'inférence des autoencodeurs variationnels conditionnels. Notre modèle peut quantifier les incertitudes dans ses données d'entraînement et générer des échantillons à partir de la distribution a posteriori des paramètres, pouvant être utilisés pour dériver des estimations d'incertitude pour les paramètres inférés. L'efficacité de notre cadre est démontrée en l'appliquant au problème de la reconstruction de cartes du fond diffus cosmologique (CMB) pour en retirer de l'effet de lentille gravitationnelle faible. Notre travail constitue un atout essentiel pour effectuer une inférence de vraisemblance implicite en dimensions élevées dans les domaines astrophysiques. Il permet d'exploiter pleinement le potentiel des missions d'observation de nouvelle génération pour améliorer notre compréhension de l'univers et de ses lois physiques fondamentales. / We live in an era marked by an abundance of high-resolution cosmological data. This influx of data brought about by next-generation observational missions on the ground and in space, bears the potential of fundamentally reshaping our understanding of the universe and its underlying physical principles. However, the elevated complexity of the observed data poses challenges to conventional data analysis approaches, either due to infeasible computational costs or the simplifying assumptions used in these algorithms that become inadequate in high-resolution, low-noise contexts, leading to suboptimal results. In response, the scientific community has turned to innovative data analysis methods, including machine learning (ML) techniques. ML models, when well-trained, can autonomously identify meaningful patterns in data more efficiently and without unnecessary restrictive assumptions. Although ML methods have shown promise in astrophysics, they also exhibit issues like lack of interpretability, hidden biases, and uncalibrated uncertainty estimates, which have hindered their application in significant scientific discoveries. This project is defined within the context of the Learning the Universe (LtU) collaboration, focused on reconstructing the initial conditions of the universe, utilizing a Bayesian forward modeling approach and harnessing the power of ML. The goal of this project is to develop a framework for conducting Bayesian inference at the pixel level in high-dimensional problems. In this thesis, I present the development of a deep learning framework for fast high-dimensional posterior sampling. This framework utilizes the Hierarchical Probabilistic U-Net architecture, which combines the power of the U-Net architecture in learning high-dimensional mappings with the rigorous inference framework of Conditional Variational Autoencoders. Our model can quantify uncertainties in its training data and generate samples from the posterior distribution of parameters, which can be used to derive uncertainty estimates for the inferred parameters. The effectiveness of our framework is demonstrated by applying it to the problem of removing the weak gravitational lensing effect from the CMB. Our work stands as an essential asset to performing high-dimensional implicit likelihood inference in astrophysical domains. It enables utilizing the full potential of next-generation observational missions to improve our understanding of the universe and its fundamental physical laws.

Page generated in 0.1222 seconds