Literatura académica sobre el tema "Deep Generatve Models"
Crea una cita precisa en los estilos APA, MLA, Chicago, Harvard y otros
Consulte las listas temáticas de artículos, libros, tesis, actas de conferencias y otras fuentes académicas sobre el tema "Deep Generatve Models".
Junto a cada fuente en la lista de referencias hay un botón "Agregar a la bibliografía". Pulsa este botón, y generaremos automáticamente la referencia bibliográfica para la obra elegida en el estilo de cita que necesites: APA, MLA, Harvard, Vancouver, Chicago, etc.
También puede descargar el texto completo de la publicación académica en formato pdf y leer en línea su resumen siempre que esté disponible en los metadatos.
Artículos de revistas sobre el tema "Deep Generatve Models"
1
Mehmood, Rayeesa, Rumaan Bashir y Kaiser J. Giri. "Deep Generative Models: A Review". Indian Journal Of Science And Technology 16, n.º 7 (21 de febrero de 2023): 460–67. http://dx.doi.org/10.17485/ijst/v16i7.2296.
Texto completo
2
Ragoza, Matthew, Tomohide Masuda y David Ryan Koes. "Generating 3D molecules conditional on receptor binding sites with deep generative models". Chemical Science 13, n.º 9 (2022): 2701–13. http://dx.doi.org/10.1039/d1sc05976a.
Texto completoResumen
We generate 3D molecules conditioned on receptor binding sites by training a deep generative model on protein–ligand complexes. Our model uses the conditional receptor information to make chemically relevant changes to the generated molecules.
3
Salakhutdinov, Ruslan. "Learning Deep Generative Models". Annual Review of Statistics and Its Application 2, n.º 1 (10 de abril de 2015): 361–85. http://dx.doi.org/10.1146/annurev-statistics-010814-020120.
Texto completo
4
Partaourides, Harris y Sotirios P. Chatzis. "Asymmetric deep generative models". Neurocomputing 241 (junio de 2017): 90–96. http://dx.doi.org/10.1016/j.neucom.2017.02.028.
Texto completo
5
Changsheng Du, Changsheng Du, Yong Li Changsheng Du y Ming Wen Yong Li. "G-DCS: GCN-Based Deep Code Summary Generation Model". 網際網路技術學刊 24, n.º 4 (julio de 2023): 965–73. http://dx.doi.org/10.53106/160792642023072404014.
Texto completoResumen
<p>In software engineering, software personnel faced many large-scale software and complex systems, these need programmers to quickly and accurately read and understand the code, and efficiently complete the tasks of software change or maintenance tasks. Code-NN is the first model to use deep learning to accomplish the task of code summary generation, but it is not used the structural information in the code itself. In the past five years, researchers have designed different code summarization systems based on neural networks. They generally use the end-to-end neural machine translation framework, but many current research methods do not make full use of the structural information of the code. This paper raises a new model called G-DCS to automatically generate a summary of java code; the generated summary is designed to help programmers quickly comprehend the effect of java methods. G-DCS uses natural language processing technology, and training the model uses a code corpus. This model could generate code summaries directly from the code files in the coded corpus. Compared with the traditional method, it uses the information of structural on the code. Through Graph Convolutional Neural Network (GCN) extracts the structural information on the code to generate the code sequence, which makes the generated code summary more accurate. The corpus used for training was obtained from GitHub. Evaluation criteria using BLEU-n. Experimental results show that our approach outperforms models that do not utilize code structure information.</p> <p> </p>
6
Wu, Han. "Face image generation and feature visualization using deep convolutional generative adversarial networks". Journal of Physics: Conference Series 2634, n.º 1 (1 de noviembre de 2023): 012041. http://dx.doi.org/10.1088/1742-6596/2634/1/012041.
Texto completoResumen
Abstract Generative Neural Networks (GAN) aims to generate realistic and recognizable images, including portraits, cartoons and other modalities. Image generation has broad application prospects and important research value in the fields of public security and digital entertainment, and has become one of the current research hotspots. This article will introduce and apply an important image generation model called GAN, which stands for Generative Adversarial Network. Unlike recent image processing models such as Variational Autoencoders (VAE), The discriminative network evaluates potential candidates while the GAN generates candidates. As a result, the discriminative network distinguishes created and real candidates, while the generative network learns to map from a latent space to an interest data distribution. In this article, the GAN model and some of its extensions will be thoroughly applied and implemented based on the dataset of CelebA, and details will be discussed through the images and graphs generated by the model. Specific training methods for various models and optimization algorithms can be produced by the GAN framework. The experiment’s findings in this article will show how the framework’s potential may be quantified and qualitatively assessed using the samples that were produced.
7
Berrahal, Mohammed, Mohammed Boukabous, Mimoun Yandouzi, Mounir Grari y Idriss Idrissi. "Investigating the effectiveness of deep learning approaches for deep fake detection". Bulletin of Electrical Engineering and Informatics 12, n.º 6 (1 de diciembre de 2023): 3853–60. http://dx.doi.org/10.11591/eei.v12i6.6221.
Texto completoResumen
As a result of notable progress in image processing and machine learning algorithms, generating, modifying, and manufacturing superior quality images has become less complicated. Nonetheless, malevolent individuals can exploit these tools to generate counterfeit images that seem genuine. Such fake images can be used to harm others, evade image detection algorithms, or deceive recognition classifiers. In this paper, we propose the implementation of the best-performing convolutional neural network (CNN) based classifier to distinguish between generated fake face images and real images. This paper aims to provide an in-depth discussion about the challenge of generated fake face image detection. We explain the different datasets and the various proposed deep learning models for fake face image detection. The models used were trained on a large dataset of real data from CelebA-HQ and fake data from a trained generative adversarial network (GAN) based generator. All testing models achieved high accuracy in detecting the fake images, especially residual neural network (ResNet50) which performed the best among with an accuracy of 99.43%.
8
Che, Tong, Xiaofeng Liu, Site Li, Yubin Ge, Ruixiang Zhang, Caiming Xiong y Yoshua Bengio. "Deep Verifier Networks: Verification of Deep Discriminative Models with Deep Generative Models". Proceedings of the AAAI Conference on Artificial Intelligence 35, n.º 8 (18 de mayo de 2021): 7002–10. http://dx.doi.org/10.1609/aaai.v35i8.16862.
Texto completoResumen
AI Safety is a major concern in many deep learning applications such as autonomous driving. Given a trained deep learning model, an important natural problem is how to reliably verify the model's prediction. In this paper, we propose a novel framework --- deep verifier networks (DVN) to detect unreliable inputs or predictions of deep discriminative models, using separately trained deep generative models. Our proposed model is based on conditional variational auto-encoders with disentanglement constraints to separate the label information from the latent representation. We give both intuitive and theoretical justifications for the model. Our verifier network is trained independently with the prediction model, which eliminates the need of retraining the verifier network for a new model. We test the verifier network on both out-of-distribution detection and adversarial example detection problems, as well as anomaly detection problems in structured prediction tasks such as image caption generation. We achieve state-of-the-art results in all of these problems.
9
Scurto, Hugo, Thomas Similowski, Samuel Bianchini y Baptiste Caramiaux. "Probing Respiratory Care With Generative Deep Learning". Proceedings of the ACM on Human-Computer Interaction 7, CSCW2 (28 de septiembre de 2023): 1–34. http://dx.doi.org/10.1145/3610099.
Texto completoResumen
This paper combines design, machine learning and social computing to explore generative deep learning as both tool and probe for respiratory care. We first present GANspire, a deep learning tool that generates fine-grained breathing waveforms, which we crafted in collaboration with one respiratory physician, attending to joint materialities of human breathing data and deep generative models. We then relate a probe, produced with breathing waveforms generated with GANspire, and led with a group of ten respiratory care experts, responding to its material attributes. Qualitative annotations showed that respiratory care experts interpreted both realistic and ambiguous attributes of breathing waveforms generated with GANspire, according to subjective aspects of physiology, activity and emotion. Semi-structured interviews also revealed experts' broader perceptions, expectations and ethical concerns on AI technology, based on their clinical practice of respiratory care, and reflexive analysis of GANspire. These findings suggest design implications for technological aids in respiratory care, and show how ambiguity of deep generative models can be leveraged as a resource for qualitative inquiry, enabling socio-material research with generative deep learning. Our paper contributes to the CSCW community by broadening how generative deep learning may be approached not only as a tool to design human-computer interactions, but also as a probe to provoke open conversations with communities of practice about their current and speculative uses of AI technology.
10
Prakash Patil, Et al. "GAN-Enhanced Medical Image Synthesis: Augmenting CXR Data for Disease Diagnosis and Improving Deep Learning Performance". Journal of Electrical Systems 19, n.º 3 (25 de enero de 2024): 53–61. http://dx.doi.org/10.52783/jes.651.
Texto completoResumen
Deep learning is increasing the need for accurate and reliable medical image analysis tools, especially for CXR disease diagnosis. This study proposes the Attention Mechanisms based Cycle-Consistent GAN (AM-CGAN) to address the lack of annotated medical data. To produce realistic and clinically relevant CXR images, our model uses Generative Adversarial Networks (GAN) and attention mechanisms. Downstream deep learning models for disease classification improve with this enhancement. The Attention Mechanisms based Cycle-Consistent GAN (AM-CGAN) improves the accuracy and reliability of deep learning models used for medical image analysis, specifically for Chest X-ray (CXR) data. CXR data is enhanced to improve disease diagnosis. Generative Adversarial Networks (GAN) create realistic medical images in the proposed model. It also uses attention mechanisms to highlight key areas in generated images. This research aims to address the lack of annotated medical data, particularly for CXR images. Training deep learning models is difficult due to the lack of diverse and well-annotated datasets. Our proposed AM-CGAN uses attention mechanisms to generate synthetic CXR images that closely resemble medical images and highlight disease-specific characteristics. AM-CGAN uses Cycle-Consistent GAN to ensure that generated images match the input distribution and prevent mode collapse. While synthesizing images, the model can selectively focus on important anatomical structures and pathological indicators using attention mechanisms. This attention-driven approach improves the clinical significance of generated images, making them better for training accurate and reliable disease classification models. Many experiments were done to test the AM-CGAN on CXR images of COVID-19, pneumonia, and normal cases. The quantitative results show high precision (98.15% accuracy). This shows the model's ability to create medical-data-like synthetic images. Downstream deep learning models trained on the augmented dataset perform better at capturing disease-specific characteristics. This study advances GAN-enhanced medical image synthesis research and addresses the data shortage in medical imaging research. The AM-CGAN attention-driven focus on disease-related regions in CXR data suggests a promising way to improve diagnostic models, especially in situations with few labeled datasets. The AM-CGAN bridges the gap between diverse data and sophisticated deep learning models for disease diagnosis, making it a major advancement in medical image analysis.
Tesis sobre el tema "Deep Generatve Models"
1
Miao, Yishu. "Deep generative models for natural language processing". Thesis, University of Oxford, 2017. http://ora.ox.ac.uk/objects/uuid:e4e1f1f9-e507-4754-a0ab-0246f1e1e258.
Texto completoResumen
Deep generative models are essential to Natural Language Processing (NLP) due to their outstanding ability to use unlabelled data, to incorporate abundant linguistic features, and to learn interpretable dependencies among data. As the structure becomes deeper and more complex, having an effective and efficient inference method becomes increasingly important. In this thesis, neural variational inference is applied to carry out inference for deep generative models. While traditional variational methods derive an analytic approximation for the intractable distributions over latent variables, here we construct an inference network conditioned on the discrete text input to provide the variational distribution. The powerful neural networks are able to approximate complicated non-linear distributions and grant the possibilities for more interesting and complicated generative models. Therefore, we develop the potential of neural variational inference and apply it to a variety of models for NLP with continuous or discrete latent variables. This thesis is divided into three parts. Part I introduces a generic variational inference framework for generative and conditional models of text. For continuous or discrete latent variables, we apply a continuous reparameterisation trick or the REINFORCE algorithm to build low-variance gradient estimators. To further explore Bayesian non-parametrics in deep neural networks, we propose a family of neural networks that parameterise categorical distributions with continuous latent variables. Using the stick-breaking construction, an unbounded categorical distribution is incorporated into our deep generative models which can be optimised by stochastic gradient back-propagation with a continuous reparameterisation. Part II explores continuous latent variable models for NLP. Chapter 3 discusses the Neural Variational Document Model (NVDM): an unsupervised generative model of text which aims to extract a continuous semantic latent variable for each document. In Chapter 4, the neural topic models modify the neural document models by parameterising categorical distributions with continuous latent variables, where the topics are explicitly modelled by discrete latent variables. The models are further extended to neural unbounded topic models with the help of stick-breaking construction, and a truncation-free variational inference method is proposed based on a Recurrent Stick-breaking construction (RSB). Chapter 5 describes the Neural Answer Selection Model (NASM) for learning a latent stochastic attention mechanism to model the semantics of question-answer pairs and predict their relatedness. Part III discusses discrete latent variable models. Chapter 6 introduces latent sentence compression models. The Auto-encoding Sentence Compression Model (ASC), as a discrete variational auto-encoder, generates a sentence by a sequence of discrete latent variables representing explicit words. The Forced Attention Sentence Compression Model (FSC) incorporates a combined pointer network biased towards the usage of words from source sentence, which significantly improves the performance when jointly trained with the ASC model in a semi-supervised learning fashion. Chapter 7 describes the Latent Intention Dialogue Models (LIDM) that employ a discrete latent variable to learn underlying dialogue intentions. Additionally, the latent intentions can be interpreted as actions guiding the generation of machine responses, which could be further refined autonomously by reinforcement learning. Finally, Chapter 8 summarizes our findings and directions for future work.
2
Misino, Eleonora. "Deep Generative Models with Probabilistic Logic Priors". Master's thesis, Alma Mater Studiorum - Università di Bologna, 2021. http://amslaurea.unibo.it/24058/.
Texto completoResumen
Many different extensions of the VAE framework have been introduced in the past. How ever, the vast majority of them focused on pure sub-symbolic approaches that are not sufficient for solving generative tasks that require a form of reasoning. In this thesis, we propose the probabilistic logic VAE (PLVAE), a neuro-symbolic deep generative model that combines the representational power of VAEs with the reasoning ability of probabilistic logic programming. The strength of PLVAE resides in its probabilistic logic prior, which provides an interpretable structure to the latent space that can be easily changed in order to apply the model to different scenarios. We provide empirical results of our approach by training PLVAE on a base task and then using the same model to generalize to novel tasks that involve reasoning with the same set of symbols.
3
Nilsson, Mårten. "Augmenting High-Dimensional Data with Deep Generative Models". Thesis, KTH, Robotik, perception och lärande, RPL, 2018. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-233969.
Texto completoResumen
Data augmentation is a technique that can be performed in various ways to improve the training of discriminative models. The recent developments in deep generative models offer new ways of augmenting existing data sets. In this thesis, a framework for augmenting annotated data sets with deep generative models is proposed together with a method for quantitatively evaluating the quality of the generated data sets. Using this framework, two data sets for pupil localization was generated with different generative models, including both well-established models and a novel model proposed for this purpose. The unique model was shown both qualitatively and quantitatively to generate the best data sets. A set of smaller experiments on standard data sets also revealed cases where this generative model could improve the performance of an existing discriminative model. The results indicate that generative models can be used to augment or replace existing data sets when training discriminative models.Dataaugmentering är en teknik som kan utföras på flera sätt för att förbättra träningen av diskriminativa modeller. De senaste framgångarna inom djupa generativa modeller har öppnat upp nya sätt att augmentera existerande dataset. I detta arbete har ett ramverk för augmentering av annoterade dataset med hjälp av djupa generativa modeller föreslagits. Utöver detta så har en metod för kvantitativ evaulering av kvaliteten hos genererade data set tagits fram. Med hjälp av detta ramverk har två dataset för pupillokalisering genererats med olika generativa modeller. Både väletablerade modeller och en ny modell utvecklad för detta syfte har testats. Den unika modellen visades både kvalitativt och kvantitativt att den genererade de bästa dataseten. Ett antal mindre experiment på standardiserade dataset visade exempel på fall där denna generativa modell kunde förbättra prestandan hos en existerande diskriminativ modell. Resultaten indikerar att generativa modeller kan användas för att augmentera eller ersätta existerande dataset vid träning av diskriminativa modeller.
4
Lindqvist, Niklas. "Automatic Question Paraphrasing in Swedish with Deep Generative Models". Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-294320.
Texto completoResumen
Paraphrase generation refers to the task of automatically generating a paraphrase given an input sentence or text. Paraphrase generation is a fundamental yet challenging natural language processing (NLP) task and is utilized in a variety of applications such as question answering, information retrieval, conversational systems etc. In this study, we address the problem of paraphrase generation of questions in Swedish by evaluating two different deep generative models that have shown promising results on paraphrase generation of questions in English. The first model is a Conditional Variational Autoencoder (C-VAE) and the other model is an extension of the first one where a discriminator network is introduced into the model to form a Generative Adversarial Network (GAN) architecture. In addition to these models, a method not based on machine-learning was implemented to act as a baseline. The models were evaluated using both quantitative and qualitative measures including grammatical correctness and equivalence to source question. The results show that the deep generative models outperformed the baseline across all quantitative metrics. Furthermore, from the qualitative evaluation it was shown that the deep generative models outperformed the baseline at generating grammatically correct sentences, but there was no noticeable difference in terms of equivalence to the source question between the models.Parafrasgenerering syftar på uppgiften att, utifrån en given mening eller text, automatiskt generera en parafras, det vill säga en annan text med samma betydelse. Parafrasgenerering är en grundläggande men ändå utmanande uppgift inom naturlig språkbehandling och används i en rad olika applikationer som informationssökning, konversionssystem, att besvara frågor givet en text etc. I den här studien undersöker vi problemet med parafrasgenerering av frågor på svenska genom att utvärdera två olika djupa generativa modeller som visat lovande resultat på parafrasgenerering av frågor på engelska. Den första modellen är en villkorsbaserad variationsautokodare (C-VAE). Den andra modellen är också en C-VAE men introducerar även en diskriminator vilket gör modellen till ett generativt motståndarnätverk (GAN). Förutom modellerna presenterade ovan, implementerades även en icke maskininlärningsbaserad metod som en baslinje. Modellerna utvärderades med både kvantitativa och kvalitativa mått inklusive grammatisk korrekthet och likvärdighet mellan parafras och originalfråga. Resultaten visar att de djupa generativa modellerna presterar bättre än baslinjemodellen på alla kvantitativa mätvärden. Vidare, visade the kvalitativa utvärderingen att de djupa generativa modellerna kunde generera grammatiskt korrekta frågor i större utsträckning än baslinjemodellen. Det var däremot ingen större skillnad i semantisk ekvivalens mellan parafras och originalfråga för de olika modellerna.
5
Gane, Georgiana Andreea. "Building generative models over discrete structures : from graphical models to deep learning". Thesis, Massachusetts Institute of Technology, 2019. https://hdl.handle.net/1721.1/121611.
Texto completoResumen
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019Cataloged from PDF version of thesis. Page 173 blank.
Includes bibliographical references (pages 159-172).
The goal of this thesis is to investigate generative models over discrete structures, such as binary grids, alignments or arbitrary graphs. We focused on developing models easy to sample from, and we approached the task from two broad perspectives: defining models via structured potential functions, and via neural network based decoders. In the first case, we investigated Perturbation Models, a family of implicit distributions where samples emerge through optimization of randomized potential functions. Designed explicitly for efficient sampling, Perturbation Models are strong candidates for building generative models over structures, and the leading open questions pertain to understanding the properties of the induced models and developing practical learning algorithms.
In this thesis, we present theoretical results showing that, in contrast to the more established Gibbs models, low-order potential functions, after undergoing randomization and maximization, lead to high-order dependencies in the induced distributions. Furthermore, while conditioning in Gibbs' distributions is straightforward, conditioning in Perturbation Models is typically not, but we theoretically characterize cases where the straightforward approach produces the correct results. Finally, we introduce a new Perturbation Models learning algorithm based on Inverse Combinatorial Optimization. We illustrate empirically both the induced dependencies and the inverse optimization approach, in learning tasks inspired by computer vision problems. In the second case, we sequentialize the structures, converting structure generation into a sequence of discrete decisions, to enable the use of sequential models.
We explore maximum likelihood training with step-wise supervision and continuous relaxations of the intermediate decisions. With respect to intermediate discrete representations, the main directions consist of using gradient estimators or designing continuous relaxations. We discuss these solutions in the context of unsupervised scene understanding with generative models. In particular, we asked whether a continuous relaxation of the counting problem also discovers the objects in an unsupervised fashion (given the increased training stability that continuous relaxations provide) and we proposed an approach based on Adaptive Computation Time (ACT) which achieves the desired result. Finally, we investigated the task of iterative graph generation. We proposed a variational lower-bound to the maximum likelihood objective, where the approximate posterior distribution renormalizes the prior distribution over local predictions which are plausible for the target graph.
For instance, the local predictions may be binary values indicating the presence or absence of an edge indexed by the given time step, for a canonical edge indexing chosen a-priori. The plausibility of each local prediction is assessed by solving a combinatorial optimization problem, and we discuss relevant approaches, including an induced sub-graph isomorphism-based algorithm for the generic graph generation case, and a polynomial algorithm for the special case of graph generation resulting from solving graph clustering tasks. In this thesis, we focused on the generic case, and we investigated the approximate posterior's relevance on synthetic graph datasets.
by Georgiana Andreea Gane.
Ph. D.
Ph.D. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science
6
Mescheder, Lars Morten [Verfasser]. "Stability and Expressiveness of Deep Generative Models / Lars Morten Mescheder". Tübingen : Universitätsbibliothek Tübingen, 2020. http://d-nb.info/1217249257/34.
Texto completo
7
Rastgoufard, Rastin. "Multi-Label Latent Spaces with Semi-Supervised Deep Generative Models". ScholarWorks@UNO, 2018. https://scholarworks.uno.edu/td/2486.
Texto completoResumen
Expert labeling, tagging, and assessment are far more costly than the processes of collecting raw data. Generative modeling is a very powerful tool to tackle this real-world problem. It is shown here how these models can be used to allow for semi-supervised learning that performs very well in label-deficient conditions.
The foundation for the work in this dissertation is built upon visualizing generative models' latent spaces to gain deeper understanding of data, analyze faults, and propose solutions. A number of novel ideas and approaches are presented to improve single-label classification. This dissertation's main focus is on extending semi-supervised Deep Generative Models for solving the multi-label problem by proposing unique mathematical and programming concepts and organization.
In all naive mixtures, using multiple labels is detrimental and causes each label's predictions to be worse than models that utilize only a single label. Examining latent spaces reveals that in many cases, large regions in the models generate meaningless results. Enforcing a priori independence is essential, and only when applied can multi-label models outperform the best single-label models. Finally, a novel learning technique called open-book learning is described that is capable of surpassing the state-of-the-art classification performance of generative models for multi-labeled, semi-supervised data sets.
8
Douwes, Constance. "On the Environmental Impact of Deep Generative Models for Audio". Electronic Thesis or Diss., Sorbonne université, 2023. http://www.theses.fr/2023SORUS074.
Texto completoResumen
Cette thèse étudie l'impact environnemental des modèles d'apprentissage profond pour la génération audio et vise à mettre le coût de calcul au cœur du processus d'évaluation. En particulier, nous nous concentrons sur différents types de modèles d'apprentissage profond spécialisés dans la synthèse audio de formes d'onde brutes. Ces modèles sont désormais un élément clé des systèmes audio modernes, et leur utilisation a considérablement augmenté ces dernières années. Leur flexibilité et leurs capacités de généralisation en font des outils puissants dans de nombreux contextes, de la synthèse de texte à la parole à la génération audio inconditionnelle. Cependant, ces avantages se font au prix de sessions d'entraînement coûteuses sur de grandes quantités de données, exploitées sur du matériel dédié à forte consommation d'énergie, ce qui entraîne d'importantes émissions de gaz à effet de serre. Les mesures que nous utilisons en tant que communauté scientifique pour évaluer nos travaux sont au cœur de ce problème. Actuellement, les chercheurs en apprentissage profond évaluent leurs travaux principalement sur la base des améliorations de la précision, de la log-vraisemblance, de la reconstruction ou des scores d'opinion, qui occultent tous le coût de calcul des modèles génératifs. Par conséquent, nous proposons d'utiliser une nouvelle méthodologie basée sur l'optimalité de Pareto pour aider la communauté à mieux évaluer leurs travaux tout en ramenant l'empreinte énergétique -- et in fine les émissions de carbone -- au même niveau d'intérêt que la qualité du son. Dans la première partie de cette thèse, nous présentons un rapport complet sur l'utilisation de diverses mesures d'évaluation des modèles génératifs profonds pour les tâches de synthèse audio. Bien que l'efficacité de calcul soit de plus en plus abordée, les mesures de qualité sont les plus couramment utilisées pour évaluer les modèles génératifs profonds, alors que la consommation d'énergie n'est presque jamais mentionnée. Nous abordons donc cette question en estimant le coût en carbone de la formation des modèles génératifs et en le comparant à d'autres coûts en carbone notables pour démontrer qu'il est loin d'être insignifiant. Dans la deuxième partie de cette thèse, nous proposons une évaluation à grande échelle des vocodeurs neuronaux pervasifs, qui sont une classe de modèles génératifs utilisés pour la génération de la parole, conditionnée par le mel-spectrogramme. Nous introduisons une analyse multi-objectifs basée sur l'optimalité de Pareto à la fois de la qualité de l'évaluation humaine et de la consommation d'énergie. Dans ce cadre, nous montrons que des modèles plus légers peuvent être plus performants que des modèles plus coûteux. En proposant de s'appuyer sur une nouvelle définition de l'efficacité, nous entendons fournir aux praticiens une base de décision pour choisir le meilleur modèle en fonction de leurs exigences. Dans la dernière partie de la thèse, nous proposons une méthode pour réduire les coûts associés à l'inférence des modèle génératif profonds, basée sur la quantification des réseaux de neurones. Nous montrons un gain notable sur la taille des modèles et donnons des pistes pour l'utilisation future de ces modèles dans des systèmes embarqués. En somme, nous fournissons des clés pour mieux comprendre l'impact des modèles génératifs profonds pour la synthèse audio ainsi qu'un nouveau cadre pour développer des modèles tout en tenant compte de leur impact environnemental. Nous espérons que ce travail permettra de sensibiliser les chercheurs à la nécessité d'étudier des modèles efficaces sur le plan énergétique tout en garantissant une qualité audio élevéeIn this thesis, we investigate the environmental impact of deep learning models for audio generation and we aim to put computational cost at the core of the evaluation process. In particular, we focus on different types of deep learning models specialized in raw waveform audio synthesis. These models are now a key component of modern audio systems, and their use has increased significantly in recent years. Their flexibility and generalization capabilities make them powerful tools in many contexts, from text-to-speech synthesis to unconditional audio generation. However, these benefits come at the cost of expensive training sessions on large amounts of data, operated on energy-intensive dedicated hardware, which incurs large greenhouse gas emissions. The measures we use as a scientific community to evaluate our work are at the heart of this problem. Currently, deep learning researchers evaluate their works primarily based on improvements in accuracy, log-likelihood, reconstruction, or opinion scores, all of which overshadow the computational cost of generative models. Therefore, we propose using a new methodology based on Pareto optimality to help the community better evaluate their work's significance while bringing energy footprint -- and in fine carbon emissions -- at the same level of interest as the sound quality. In the first part of this thesis, we present a comprehensive report on the use of various evaluation measures of deep generative models for audio synthesis tasks. Even though computational efficiency is increasingly discussed, quality measurements are the most commonly used metrics to evaluate deep generative models, while energy consumption is almost never mentioned. Therefore, we address this issue by estimating the carbon cost of training generative models and comparing it to other noteworthy carbon costs to demonstrate that it is far from insignificant. In the second part of this thesis, we propose a large-scale evaluation of pervasive neural vocoders, which are a class of generative models used for speech generation, conditioned on mel-spectrogram. We introduce a multi-objective analysis based on Pareto optimality of both quality from human-based evaluation and energy consumption. Within this framework, we show that lighter models can perform better than more costly models. By proposing to rely on a novel definition of efficiency, we intend to provide practitioners with a decision basis for choosing the best model based on their requirements. In the last part of the thesis, we propose a method to reduce the inference costs of neural vocoders, based on quantizated neural networks. We show a significant gain on the memory size and give some hints for the future use of these models on embedded hardware. Overall, we provide keys to better understand the impact of deep generative models for audio synthesis as well as a new framework for developing models while accounting for their environmental impact. We hope that this work raises awareness on the need to investigate energy-efficient models simultaneously with high perceived quality
9
Patsanis, Alexandros. "Network Anomaly Detection and Root Cause Analysis with Deep Generative Models". Thesis, Uppsala universitet, Institutionen för informationsteknologi, 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-397367.
Texto completoResumen
The project's objective is to detect network anomalies happening in a telecommunication network due to hardware malfunction or software defects after a vast upgrade on the network's system over a specific area, such as a city. The network's system generates statistical data at a 15-minute interval for different locations in the area of interest. For every interval, all statistical data generated over an area are aggregated and converted to images. In this way, an image represents a snapshot of the network for a specific interval, where statistical data are represented as points having different density values. To that problem, this project makes use of Generative Adversarial Networks (GANs), which learn a manifold of the normal network pattern. Additionally, mapping from new unseen images to the learned manifold results in an anomaly score used to detect anomalies. The anomaly score is a combination of the reconstruction error and the learned feature representation. Two models for detecting anomalies are used in this project, AnoGAN and f-AnoGAN. Furthermore, f-AnoGAN uses a state-of-the-art approach called Wasstestein GAN with gradient penalty, which improves the initial implementation of GANs. Both quantitative and qualitative evaluation measurements are used to assess GANs models, where F1 Score and Wasserstein loss are used for the quantitative evaluation and linear interpolation in the hidden space for qualitative evaluation. Moreover, to set a threshold, a prediction model used to predict the expected behaviour of the network for a specific interval. Then, the predicted behaviour is used over the anomaly detection model to define a threshold automatically. Our experiments were implemented successfully for both prediction and anomaly detection models. We additionally tested known abnormal behaviours which were detected and visualised. However, more research has to be done over the evaluation of GANs, as there is no universal approach to evaluate them.
10
Alabdallah, Abdallah. "Human Understandable Interpretation of Deep Neural Networks Decisions Using Generative Models". Thesis, Högskolan i Halmstad, Halmstad Embedded and Intelligent Systems Research (EIS), 2019. http://urn.kb.se/resolve?urn=urn:nbn:se:hh:diva-41035.
Texto completoResumen
Deep Neural Networks have long been considered black box systems, where their interpretability is a concern when applied in safety critical systems. In this work, a novel approach of interpreting the decisions of DNNs is proposed. The approach depends on exploiting generative models and the interpretability of their latent space. Three methods for ranking features are explored, two of which depend on sensitivity analysis, and the third one depends on Random Forest model. The Random Forest model was the most successful to rank the features, given its accuracy and inherent interpretability.
Libros sobre el tema "Deep Generatve Models"
1
Mukhopadhyay, Anirban, Ilkay Oksuz, Sandy Engelhardt, Dajiang Zhu y Yixuan Yuan, eds. Deep Generative Models. Cham: Springer Nature Switzerland, 2022. http://dx.doi.org/10.1007/978-3-031-18576-2.
Texto completo
2
Mukhopadhyay, Anirban, Ilkay Oksuz, Sandy Engelhardt, Dajiang Zhu y Yixuan Yuan, eds. Deep Generative Models. Cham: Springer Nature Switzerland, 2024. http://dx.doi.org/10.1007/978-3-031-53767-7.
Texto completo
3
Engelhardt, Sandy, Ilkay Oksuz, Dajiang Zhu, Yixuan Yuan, Anirban Mukhopadhyay, Nicholas Heller, Sharon Xiaolei Huang, Hien Nguyen, Raphael Sznitman y Yuan Xue, eds. Deep Generative Models, and Data Augmentation, Labelling, and Imperfections. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-88210-5.
Texto completo
4
Ali, Hazrat, Mubashir Husain Rehmani y Zubair Shah, eds. Advances in Deep Generative Models for Medical Artificial Intelligence. Cham: Springer Nature Switzerland, 2023. http://dx.doi.org/10.1007/978-3-031-46341-9.
Texto completo
5
Bongard, Josh. Modeling self and others. Oxford University Press, 2018. http://dx.doi.org/10.1093/oso/9780199674923.003.0011.
Texto completoResumen
Embodied cognition is the view that intelligence arises out of the interaction between an agent’s body and its environment. Taking such a view generates novel scientific hypotheses about biological intelligence and opportunities for advancing artificial intelligence. In this chapter we review one such set of hypotheses regarding how a robot may generate models of self, and others, and then exploit those models to recover from damage or exhibit the rudiments of social cognition. This modeling of self and others draws mainly on three concepts from neuroscience and AI: forward and inverse models in the brain, the neuronal replicator hypothesis, and the brain as a hierarchical prediction machine. The chapter concludes with future directions, including the integration of deep learning methods with embodied cognition.
6
Mukhopadhyay, Anirban, Dajiang Zhu, Sandy Engelhardt, Ilkay Oksuz y Yixuan Yuan. Deep Generative Models: Second MICCAI Workshop, DGM4MICCAI 2022, Held in Conjunction with MICCAI 2022, Singapore, September 22, 2022, Proceedings. Springer, 2022.
Buscar texto completo
7
Mukhopadhyay, Anirban, Dajiang Zhu, Sandy Engelhardt, Ilkay Oksuz y Yixuan Yuan. Deep Generative Models, and Data Augmentation, Labelling, and Imperfections: First Workshop, DGM4MICCAI 2021, and First Workshop, DALI 2021, Held in Conjunction with MICCAI 2021, Strasbourg, France, October 1, 2021, Proceedings. Springer International Publishing AG, 2021.
Buscar texto completo
8
Constantinesco, Thomas. Writing Pain in the Nineteenth-Century United States. Oxford University Press, 2022. http://dx.doi.org/10.1093/oso/9780192855596.001.0001.
Texto completoResumen
This book examines how pain is represented in a range of literary texts and genres from the nineteenth-century United States. It considers the aesthetic, philosophical, and ethical implications of pain across the works of Ralph Waldo Emerson, Harriet Jacobs, Emily Dickinson, Henry James, Elizabeth Stuart Phelps, and Alice James, as the national culture of pain progressively transformed in the wake of the invention of anesthesia. Through these writers, it argues that pain, while undeniably destructive, also generates language and identities, and demonstrates how literature participates in theorizing the problems of mind and body that undergird the deep chasms of selfhood, sociality, gender, and race of a formative period in American history. Writing Pain considers first Emerson’s philosophy of compensation, which promises to convert pain into gain. It then explores the limitations of this model, showing how Jacobs contests the division of body and mind that underwrites it and how Dickinson challenges its alleged universalism by foregrounding the unshareability of pain as a paradoxical measure of togetherness. The book investigates next the concurrent economies of affects in which pain was implicated during and after the Civil War and argues, through the example of James and Phelps, for queer sociality as a response to the heteronormative violence of sentimentalism. The last chapter on Alice James extends the critique of sentimental sympathy while returning to the book’s premise that pain is generative and the site of thought. By linking literary formalism with individual and social formation, Writing Pain eventually claims close reading as a method to recover the theoretical work of literature.
9
Joho, Tobias. Thucydides, Epic, and Tragedy. Editado por Sara Forsdyke, Edith Foster y Ryan Balot. Oxford University Press, 2017. http://dx.doi.org/10.1093/oxfordhb/9780199340385.013.40.
Texto completoResumen
Despite initially acknowledging the deep gulf between his concern with the truth and the fanciful compositions of the poets, Thucydides is strongly influenced by the examples of Homer and Attic tragedy. Three areas of influence can be distinguished. First, Homer and the tragedians provide fundamental structural principles: Thucydides adopts Homer’s solution to narrating simultaneous strands of events, follows Homer’s example in heightening suspense, and learns from both Homer and tragedy to generate unity through narrative patterning. By adopting Homeric and tragic features, Thucydides infuses his narrative with a specific tone: tragic irony and reversals create an atmosphere of eerie inevitability, Homeric “Almost episodes” underscore the fragility of human endeavors, the Homeric emphasis on factual precision heightens Thucydides’ tone of objective pathos. Third, Thucydides’ allusions to specific Homeric and tragic episodes, besides demonstrating Thucydides’ engagement with Homer and tragedy, reveal his persisting distance from his poetic models.
10
Aguayo, Angela J. Documentary Resistance. Oxford University Press, 2019. http://dx.doi.org/10.1093/oso/9780190676216.001.0001.
Texto completoResumen
The potential of documentary moving images to foster democratic exchange has been percolating within media production culture for the last century, and now, with mobile cameras at our fingertips and broadcasts circulating through unpredictable social networks, the documentary impulse is coming into its own as a political force of social change. The exploding reach and power of audio and video are multiplying documentary modes of communication. Once considered an outsider media practice, documentary is finding mass appeal in the allure of moving images, collecting participatory audiences that create meaningful challenges to the social order. Documentary is adept at collecting frames of human experience, challenging those insights, and turning these stories into public knowledge that is palpable for audiences. Generating pathways of exchange between unlikely interlocutors, collective identification forged with documentary discourse constitutes a mode of political agency that is directing energy toward acting in the world. Reflecting experiences of life unfolding before the camera, documentary representations help order social relationships that deepen our public connections and generate collective roots. As digital culture creates new pathways through which information can flow, the connections generated from social change documentary constitute an emerging public commons. Considering the deep ideological divisions that are fracturing U.S. democracy, it is of critical significance to understand how communities negotiate power and difference by way of an expanding documentary commons. Investment in the force of documentary resistance helps cultivate an understanding of political life from the margins, where documentary production practices are a form of survival.
Capítulos de libros sobre el tema "Deep Generatve Models"
1
Vasudevan, Shriram K., Sini Raj Pulari y Subashri Vasudevan. "Generative Models". En Deep Learning, 209–25. New York: Chapman and Hall/CRC, 2021. http://dx.doi.org/10.1201/9781003185635-9.
Texto completo
2
Calin, Ovidiu. "Generative Models". En Deep Learning Architectures, 591–609. Cham: Springer International Publishing, 2020. http://dx.doi.org/10.1007/978-3-030-36721-3_19.
Texto completo
3
Tomczak, Jakub M. "Autoregressive Models". En Deep Generative Modeling, 13–25. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-93158-2_2.
Texto completo
4
Tomczak, Jakub M. "Energy-Based Models". En Deep Generative Modeling, 143–58. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-93158-2_6.
Texto completo
5
Tomczak, Jakub M. "Flow-Based Models". En Deep Generative Modeling, 27–56. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-93158-2_3.
Texto completo
6
Tomczak, Jakub M. "Latent Variable Models". En Deep Generative Modeling, 57–127. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-93158-2_4.
Texto completo
7
Schön, Julian, Raghavendra Selvan y Jens Petersen. "Interpreting Latent Spaces of Generative Models for Medical Images Using Unsupervised Methods". En Deep Generative Models, 24–33. Cham: Springer Nature Switzerland, 2022. http://dx.doi.org/10.1007/978-3-031-18576-2_3.
Texto completo
8
Mensing, Daniel, Jochen Hirsch, Markus Wenzel y Matthias Günther. "3D (c)GAN for Whole Body MR Synthesis". En Deep Generative Models, 97–105. Cham: Springer Nature Switzerland, 2022. http://dx.doi.org/10.1007/978-3-031-18576-2_10.
Texto completo
9
Dietrichstein, Marc, David Major, Martin Trapp, Maria Wimmer, Dimitrios Lenis, Philip Winter, Astrid Berg, Theresa Neubauer y Katja Bühler. "Anomaly Detection Using Generative Models and Sum-Product Networks in Mammography Scans". En Deep Generative Models, 77–86. Cham: Springer Nature Switzerland, 2022. http://dx.doi.org/10.1007/978-3-031-18576-2_8.
Texto completo
10
Trullo, Roger, Quoc-Anh Bui, Qi Tang y Reza Olfati-Saber. "Image Translation Based Nuclei Segmentation for Immunohistochemistry Images". En Deep Generative Models, 87–96. Cham: Springer Nature Switzerland, 2022. http://dx.doi.org/10.1007/978-3-031-18576-2_9.
Texto completoActas de conferencias sobre el tema "Deep Generatve Models"
1
Han, Tian, Jiawen Wu y Ying Nian Wu. "Replicating Active Appearance Model by Generator Network". En Twenty-Seventh International Joint Conference on Artificial Intelligence {IJCAI-18}. California: International Joint Conferences on Artificial Intelligence Organization, 2018. http://dx.doi.org/10.24963/ijcai.2018/305.
Texto completoResumen
A recent Cell paper [Chang and Tsao, 2017] reports an interesting discovery. For the face stimuli generated by a pre-trained active appearance model (AAM), the responses of neurons in the areas of the primate brain that are responsible for face recognition exhibit strong linear relationship with the shape variables and appearance variables of the AAM that generates the face stimuli. In this paper, we show that this behavior can be replicated by a deep generative model called the generator network, which assumes that the observed signals are generated by latent random variables via a top-down convolutional neural network. Specifically, we learn the generator network from the face images generated by a pre-trained AAM model using variational auto-encoder, and we show that the inferred latent variables of the learned generator network have strong linear relationship with the shape and appearance variables of the AAM model that generates the face images. Unlike the AAM model that has an explicit shape model where the shape variables generate the control points or landmarks, the generator network has no such shape model and shape variables. Yet the generator network can learn the shape knowledge in the sense that some of the latent variables of the learned generator network capture the shape variations in the face images generated by AAM.
2
Misra, Siddharth, Jungang Chen, Polina Churilova y Yusuf Falola. "Generative Artificial Intelligence for Geomodeling". En International Petroleum Technology Conference. IPTC, 2024. http://dx.doi.org/10.2523/iptc-23477-ms.
Texto completoResumen
Abstract Subsurface earth models, also known as geomodels, are essential for characterizing and developing complex subsurface systems. Traditional geomodel generation methods, such as multiple-point statistics, can be time-consuming and computationally expensive. Generative Artificial Intelligence (AI) offers a promising alternative, with the potential to generate high-quality geomodels more quickly and efficiently. This paper proposes a deep-learning-based generative AI for geomodeling that comprises two deep learning models: a hierarchical vector-quantized variational autoencoder (VQ-VAE-2) and a PixelSNAIL autoregressive model. The VQ-VAE-2 learns to massively compress geomodels into a low-dimensional, discrete latent representation. The PixelSNAIL then learns the prior distribution of the latent codes. To generate a geomodel, the PixelSNAIL samples from the prior distribution of latent codes and the decoder of the VQ-VAE-2 converts the sampled latent code to a newly constructed geomodel. The PixelSNAIL can be used for unconditional or conditional geomodel generation. In unconditional generation, the generative workflow generates an ensemble of geomodels without any constraint. In conditional geomodel generation, the generative workflow generates an ensemble of geomodels similar to a user-defined source geomodel. This facilitates the control and manipulation of the generated geomodels. To improve the generation of fluvial channels in the geomodels, we use perceptual loss instead of the traditional mean absolute error loss in the VQ-VAE-2 model. At a specific compression ratio, the proposed Generative AI method generates multi-attribute geomodels of higher quality than single-attribute geomodels.
3
Li, Chen, Chikashige Yamanaka, Kazuma Kaitoh y Yoshihiro Yamanishi. "Transformer-based Objective-reinforced Generative Adversarial Network to Generate Desired Molecules". En Thirty-First International Joint Conference on Artificial Intelligence {IJCAI-22}. California: International Joint Conferences on Artificial Intelligence Organization, 2022. http://dx.doi.org/10.24963/ijcai.2022/539.
Texto completoResumen
Deep generative models of sequence-structure data have attracted widespread attention in drug discovery. However, such models cannot fully extract the semantic features of molecules from sequential representations. Moreover, mode collapse reduces the diversity of the generated molecules. This paper proposes a transformer-based objective-reinforced generative adversarial network (TransORGAN) to generate molecules. TransORGAN leverages a transformer architecture as a generator and uses a stochastic policy gradient for reinforcement learning to generate plausible molecules with rich semantic features. The discriminator grants rewards that guide the policy update of the generator, while an objective-reinforced penalty encourages the generation of diverse molecules. Experiments were performed using the ZINC chemical dataset, and the results demonstrated the usefulness of TransORGAN in terms of uniqueness, novelty, and diversity of the generated molecules.
4
Chen, Wei y Faez Ahmed. "PaDGAN: A Generative Adversarial Network for Performance Augmented Diverse Designs". En ASME 2020 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. American Society of Mechanical Engineers, 2020. http://dx.doi.org/10.1115/detc2020-22729.
Texto completoResumen
Abstract Deep generative models are proven to be a useful tool for automatic design synthesis and design space exploration. When applied in engineering design, existing generative models face three challenges: 1) generated designs lack diversity and do not cover all areas of the design space, 2) it is difficult to explicitly improve the overall performance or quality of generated designs, and 3) existing models generate do not generate novel designs, outside the domain of the training data. In this paper, we simultaneously address these challenges by proposing a new Determinantal Point Processes based loss function for probabilistic modeling of diversity and quality. With this new loss function, we develop a variant of the Generative Adversarial Network, named “Performance Augmented Diverse Generative Adversarial Network” or PaDGAN, which can generate novel high-quality designs with good coverage of the design space. Using three synthetic examples and one real-world airfoil design example, we demonstrate that PaDGAN can generate diverse and high-quality designs. In comparison to a vanilla Generative Adversarial Network, on average, it generates samples with 28% higher mean quality score with larger diversity and without the mode collapse issue. Unlike typical generative models that usually generate new designs by interpolating within the boundary of training data, we show that PaDGAN expands the design space boundary outside the training data towards high-quality regions. The proposed method is broadly applicable to many tasks including design space exploration, design optimization, and creative solution recommendation.
5
Xiao, Chaowei, Bo Li, Jun-yan Zhu, Warren He, Mingyan Liu y Dawn Song. "Generating Adversarial Examples with Adversarial Networks". En Twenty-Seventh International Joint Conference on Artificial Intelligence {IJCAI-18}. California: International Joint Conferences on Artificial Intelligence Organization, 2018. http://dx.doi.org/10.24963/ijcai.2018/543.
Texto completoResumen
Deep neural networks (DNNs) have been found to be vulnerable to adversarial examples resulting from adding small-magnitude perturbations to inputs. Such adversarial examples can mislead DNNs to produce adversary-selected results. Different attack strategies have been proposed to generate adversarial examples, but how to produce them with high perceptual quality and more efficiently requires more research efforts. In this paper, we propose AdvGAN to generate adversarial exam- ples with generative adversarial networks (GANs), which can learn and approximate the distribution of original instances. For AdvGAN, once the generator is trained, it can generate perturbations efficiently for any instance, so as to potentially accelerate adversarial training as defenses. We apply Adv- GAN in both semi-whitebox and black-box attack settings. In semi-whitebox attacks, there is no need to access the original target model after the generator is trained, in contrast to traditional white-box attacks. In black-box attacks, we dynamically train a distilled model for the black-box model and optimize the generator accordingly. Adversarial examples generated by AdvGAN on different target models have high attack success rate under state-of-the-art defenses compared to other attacks. Our attack has placed the first with 92.76% accuracy on a public MNIST black-box attack challenge.
6
Oussidi, Achraf y Azeddine Elhassouny. "Deep generative models: Survey". En 2018 International Conference on Intelligent Systems and Computer Vision (ISCV). IEEE, 2018. http://dx.doi.org/10.1109/isacv.2018.8354080.
Texto completo
7
Liu, Bochao, Pengju Wang, Shikun Li, Dan Zeng y Shiming Ge. "Model Conversion via Differentially Private Data-Free Distillation". En Thirty-Second International Joint Conference on Artificial Intelligence {IJCAI-23}. California: International Joint Conferences on Artificial Intelligence Organization, 2023. http://dx.doi.org/10.24963/ijcai.2023/243.
Texto completoResumen
While massive valuable deep models trained on large-scale data have been released to facilitate the artificial intelligence community, they may encounter attacks in deployment which leads to privacy leakage of training data. In this work, we propose a learning approach termed differentially private data-free distillation (DPDFD) for model conversion that can convert a pretrained model (teacher) into its privacy-preserving counterpart (student) via an intermediate generator without access to training data. The learning collaborates three parties in a unified way. First, massive synthetic data are generated with the generator. Then, they are fed into the teacher and student to compute differentially private gradients by normalizing the gradients and adding noise before performing descent. Finally, the student is updated with these differentially private gradients and the generator is updated by taking the student as a fixed discriminator in an alternate manner. In addition to a privacy-preserving student, the generator can generate synthetic data in a differentially private way for other down-stream tasks. We theoretically prove that our approach can guarantee differential privacy and well convergence. Extensive experiments that significantly outperform other differentially private generative approaches demonstrate the effectiveness of our approach.
8
Cintas, Celia, Payel Das, Brian Quanz, Girmaw Abebe Tadesse, Skyler Speakman y Pin-Yu Chen. "Towards Creativity Characterization of Generative Models via Group-Based Subset Scanning". En Thirty-First International Joint Conference on Artificial Intelligence {IJCAI-22}. California: International Joint Conferences on Artificial Intelligence Organization, 2022. http://dx.doi.org/10.24963/ijcai.2022/683.
Texto completoResumen
Deep generative models, such as Variational Autoencoders (VAEs) and Generative Adversarial Networks (GANs), have been employed widely in computational creativity research. However, such models discourage out-of-distribution generation to avoid spurious sample generation, thereby limiting their creativity. Thus, incorporating research on human creativity into generative deep learning techniques presents an opportunity to make their outputs more compelling and human-like. As we see the emergence of generative models directed toward creativity research, a need for machine learning-based surrogate metrics to characterize creative output from these models is imperative. We propose group-based subset scanning to identify, quantify, and characterize creative processes by detecting a subset of anomalous node-activations in the hidden layers of the generative models. Our experiments on the standard image benchmarks and their ``creatively generated'' variants reveal that the proposed subset scores distribution is more useful for detecting novelty in creative processes in the activation space rather than the pixel space. Further, we found that creative samples generate larger subsets of anomalies than normal or non-creative samples across datasets. The node activations highlighted during the creative decoding process are different from those responsible for the normal sample generation. Lastly, we assess if the images from the subsets selected by our method were also found creative by human evaluators, presenting a link between creativity perception in humans and node activations within deep neural nets.
9
Çelik, Mustafa y Ahmet HaydarÖrnek. "GAN-Based Data Augmentation and Anonymization for Mask Classification". En 10th International Conference on Natural Language Processing (NLP 2021). Academy and Industry Research Collaboration Center (AIRCC), 2021. http://dx.doi.org/10.5121/csit.2021.112315.
Texto completoResumen
Deep learning methods, especially convolutional neural networks (CNNs), have made a major contribution to computer vision. However, deep learning classifiers need large-scale annotated datasets to be trained without over-fitting. Also, in high-data diversity, trained models generalize better. However, collecting such a large-scale dataset remains challenging. Furthermore, it is invaluable for researchers to protect the subjects' confidentiality when using their personal data such as face images. In this paper, we propose a deep learning Generative Adversarial Networks (GANs) which generates synthetic samples for our mask classification model. Our contributions in this work are two-fold that the synthetics images provide. First, GANs' models can be used as an anonymization tool when the subjects' confidentiality is matters. Second, the generated masked/unmasked face images boost the performance of the mask classification model by using the synthetic images as a form of data augmentation. In our work, the classification accuracy using only traditional data augmentations is 93.71 %. By using both synthetic data and original data with traditional data augmentations the result is 95.50 %. It is shown that the GAN-generated synthetic data boosts the performance of deep learning classifiers.
10
Lopez, Christian, Scarlett R. Miller y Conrad S. Tucker. "Human Validation of Computer vs Human Generated Design Sketches". En ASME 2018 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. American Society of Mechanical Engineers, 2018. http://dx.doi.org/10.1115/detc2018-85698.
Texto completoResumen
The objective of this work is to explore the perceived visual and functional characteristics of computer generated sketches, compared to human created sketches. In addition, this work explores the possible biases that humans may have towards the perceived functionality of computer generated sketches. Recent advancements in deep generative design methods have allowed designers to implement computational tools to automatically generate large pools of new design ideas. However, if computational tools are to co-create ideas and solutions alongside designers, their ability to generate not only novel but also functional ideas, needs to be explored. Moreover, since decision-makers need to select those creative ideas for further development to ensure innovation, their possible biases towards computer generated ideas need to be explored. In this study, 619 human participants were recruited to analyze the perceived visual and functional characteristics of 50 human created 2D sketches, and 50 2D sketches generated by a deep learning generative model (i.e., computer generated). The results indicate that participants perceived the computer generated sketches as more functional than the human generated sketches. This perceived functionality was not biased by the presence of labels that explicitly presented the sketches as either human or computer generated. Moreover, the results reveal that participants were not able to classify the 2D sketches as human or computer generated with accuracies greater than random chance. The results provide evidence that supports the capabilities of deep learning generative design tools and their potential to assist designers in creative tasks such as ideation.
Informes sobre el tema "Deep Generatve Models"
1
Ogunbire, Abimbola, Panick Kalambay, Hardik Gajera y Srinivas Pulugurtha. Deep Learning, Machine Learning, or Statistical Models for Weather-related Crash Severity Prediction. Mineta Transportation Institute, diciembre de 2023. http://dx.doi.org/10.31979/mti.2023.2320.
Texto completoResumen
Nearly 5,000 people are killed and more than 418,000 are injured in weather-related traffic incidents each year. Assessments of the effectiveness of statistical models applied to crash severity prediction compared to machine learning (ML) and deep learning techniques (DL) help researchers and practitioners know what models are most effective under specific conditions. Given the class imbalance in crash data, the synthetic minority over-sampling technique for nominal (SMOTE-N) data was employed to generate synthetic samples for the minority class. The ordered logit model (OLM) and the ordered probit model (OPM) were evaluated as statistical models, while random forest (RF) and XGBoost were evaluated as ML models. For DL, multi-layer perceptron (MLP) and TabNet were evaluated. The performance of these models varied across severity levels, with property damage only (PDO) predictions performing the best and severe injury predictions performing the worst. The TabNet model performed best in predicting severe injury and PDO crashes, while RF was the most effective in predicting moderate injury crashes. However, all models struggled with severe injury classification, indicating the potential need for model refinement and exploration of other techniques. Hence, the choice of model depends on the specific application and the relative costs of false negatives and false positives. This conclusion underscores the need for further research in this area to improve the prediction accuracy of severe and moderate injury incidents, ultimately improving available data that can be used to increase road safety.
2
Huang, Lei, Meng Song, Hui Shen, Huixiao Hong, Ping Gong, Deng Hong-Wen y Zhang Chaoyang. Deep learning methods for omics data imputation. Engineer Research and Development Center (U.S.), febrero de 2024. http://dx.doi.org/10.21079/11681/48221.
Texto completoResumen
One common problem in omics data analysis is missing values, which can arise due to various reasons, such as poor tissue quality and insufficient sample volumes. Instead of discarding missing values and related data, imputation approaches offer an alternative means of handling missing data. However, the imputation of missing omics data is a non-trivial task. Difficulties mainly come from high dimensionality, non-linear or nonmonotonic relationships within features, technical variations introduced by sampling methods, sample heterogeneity, and the non-random missingness mechanism. Several advanced imputation methods, including deep learning-based methods, have been proposed to address these challenges. Due to its capability of modeling complex patterns and relationships in large and high-dimensional datasets, many researchers have adopted deep learning models to impute missing omics data. This review provides a comprehensive overview of the currently available deep learning-based methods for omics imputation from the perspective of deep generative model architectures such as autoencoder, variational autoencoder, generative adversarial networks, and Transformer, with an emphasis on multi-omics data imputation. In addition, this review also discusses the opportunities that deep learning brings and the challenges that it might face in this field.
3
Pascal Notin, Pascal Notin. Designing ultrastable carbonic anhydrase with deep generative models and high-throughput assays. Experiment, octubre de 2023. http://dx.doi.org/10.18258/57574.
Texto completo
4
Sadoune, Igor, Marcelin Joanis y Andrea Lodi. Implementing a Hierarchical Deep Learning Approach for Simulating multilevel Auction Data. CIRANO, septiembre de 2023. http://dx.doi.org/10.54932/lqog8430.
Texto completoResumen
We present a deep learning solution to address the challenges of simulating realistic synthetic first-price sealed-bid auction data. The complexities encountered in this type of auction data include high-cardinality discrete feature spaces and a multilevel structure arising from multiple bids associated with a single auction instance. Our methodology combines deep generative modeling (DGM) with an artificial learner that predicts the conditional bid distribution based on auction characteristics, contributing to advancements in simulation-based research. This approach lays the groundwork for creating realistic auction environments suitable for agent-based learning and modeling applications. Our contribution is twofold: we introduce a comprehensive methodology for simulating multilevel discrete auction data, and we underscore the potential ofDGMas a powerful instrument for refining simulation techniques and fostering the development of economic models grounded in generative AI.
5
Bidier, S., U. Khristenko, A. Kodakkal, C. Soriano y R. Rossi. D7.4 Final report on Stochastic Optimization results. Scipedia, 2022. http://dx.doi.org/10.23967/exaqute.2022.3.02.
Texto completoResumen
This deliverable report focuses on the final stochastic optimization results obtained within the EXAscale Quantification of Uncertainties for Technology and Science Simulation (ExaQUte) project. Details on a novel wind inlet generator that is able to incorporate local wind-field data through a deep-learned rapid distortion model and generates the turbulent wind data during run-time is presented in section 2. Section 3 presents the results of the overall stochastic optimization procedure applied to a twisted tapered tower with multiple design parameters within an uncertain synthetic wind field. Thereby, the significance of the developed methods and the obtained results are discussed and their integration in industrial wind-engineering workflows is outlined in section 4.
6
Malej, Matt y Fengyan Shi. Suppressing the pressure-source instability in modeling deep-draft vessels with low under-keel clearance in FUNWAVE-TVD. Engineer Research and Development Center (U.S.), mayo de 2021. http://dx.doi.org/10.21079/11681/40639.
Texto completoResumen
This Coastal and Hydraulics Engineering Technical Note (CHETN) documents the development through verification and validation of three instability-suppressing mechanisms in FUNWAVE-TVD, a Boussinesq-type numerical wave model, when modeling deep-draft vessels with a low under-keel clearance (UKC). Many large commercial ports and channels (e.g., Houston Ship Channel, Galveston, US Army Corps of Engineers [USACE]) are traveled and affected by tens of thousands of commercial vessel passages per year. In a series of recent projects undertaken for the Galveston District (USACE), it was discovered that when deep-draft vessels are modeled using pressure-source mechanisms, they can suffer from model instabilities when low UKC is employed (e.g., vessel draft of 12 m¹ in a channel of 15 m or less of depth), rendering a simulation unstable and obsolete. As an increasingly large number of deep-draft vessels are put into service, this problem is becoming more severe. This presents an operational challenge when modeling large container-type vessels in busy shipping channels, as these often will come as close as 1 m to the bottom of the channel, or even touch the bottom. This behavior would subsequently exhibit a numerical discontinuity in a given model and could severely limit the sample size of modeled vessels. This CHETN outlines a robust approach to suppressing such instability without compromising the integrity of the far-field vessel wave/wake solution. The three methods developed in this study aim to suppress high-frequency spikes generated nearfield of a vessel. They are a shock-capturing method, a friction method, and a viscosity method, respectively. The tests show that the combined shock-capturing and friction method is the most effective method to suppress the local high-frequency noises, while not affecting the far-field solution. A strong test, in which the target draft is larger than the channel depth, shows that there are no high-frequency noises generated in the case of ship squat as long as the shock-capturing method is used.
7
Mohammadi, N., D. Corrigan, A. A. Sappin y N. Rayner. Evidence for a Neoarchean to earliest-Paleoproterozoic mantle metasomatic event prior to formation of the Mesoproterozoic-age Strange Lake REE deposit, Newfoundland and Labrador, and Quebec, Canada. Natural Resources Canada/CMSS/Information Management, 2022. http://dx.doi.org/10.4095/330866.
Texto completoResumen
A complete suite of bulk major- and trace-elements measurements combined with macroscopic/microscopic observations and mineralogy guided by scanning electron microscope-energy dispersive spectrometry (SEM-EDS) analyses were applied on Nekuashu (2.55 Ga) and Pelland (2.32 Ga) intrusions in northern Canada, near the Strange Lake rare earth elements (REE) deposit, to evaluate their magmatic evolution and possible relations to the Mesoproterozoic Strange Lake Peralkaline Complex (SLPC). These Neoarchean to earliest-Paleoproterozoic intrusions, part of the Core Zone in southeastern Churchill Province, comprise mainly hypersolvus suites, including hornblendite, gabbro, monzogabbro/monzodiorite, monzonite, syenite/augite-syenite, granodiorite, and mafic diabase/dyke. However, the linkage of the suites and their petrogenesis are poorly understood. Geochemical evidence suggests a combination of 'intra-crustal multi-stage differentiation', mainly controlled by fractional crystallization (to generate mafic to felsic suites), and 'accumulation' (to form hornblendite suite) was involved in the evolution history of this system. Our model proposes that hornblendite and mafic to felsic intrusive rocks of both intrusions share a similar basaltic parent magma, generated from melting of a hydrous metasomatized mantle source that triggered an initial REE and incompatible element enrichment that prepared the ground for the subsequent enrichment in the SLPC. Geochemical signature of the hornblendite suite is consistent with a cumulate origin and its formation during the early stages of the magma evolution, however, the remaining suites were mainly controlled by 'continued fractional crystallization' processes, producing more evolved suites: gabbronorite/hornblende-gabbro ? monzogabbro/monzodiorite ? monzonite ? syenite/augite-syenite. In this proposed model, the hydrous mantle-derived basaltic magma was partly solidified to form the mafic suites (gabbronorite/hornblende-gabbro) by early-stage plagioclase-pyroxene-amphibole fractionation in the deep crust while settling of the early crystallized hornblende (+pyroxene) led to the formation of the hornblendite cumulates. The subsequent fractionation of plagioclase, pyroxene, and amphibole from the residual melt produced the more intermediate suites of monzogabbro/monzodiorite. The evolved magma ascended upward into the shallow crust to form monzonite by K-feldspar fractionation. The residual melt then intruded at shallower depth to form syenite/augite-syenite with abundant microcline crystals. The granodiorite suite was probably generated from lower crustal melts associated with the mafic end members. Later mafic diabase/dykes were likely generated by further partial melting of the same source at depth that were injected into the other suites.
8
Huijser, MP, J. W. Duffield, C. Neher, A. P. Clevenger y T. Mcguire. Final Report 2022: Update and expansion of the WVC mitigation measures and their cost-benefit model. Nevada Department of Transportation, octubre de 2022. http://dx.doi.org/10.15788/ndot2022.10.
Texto completoResumen
This report contains an update and an expansion of a cost-benefit model for wildlife-vehicle collisions and associated mitigation measures along highways, that was originally calculated in 2007 US$ and published in 2009. The direct cost values (vehicle repair, human injuries, human fatalities) were updated for deer, elk, and moose, and expanded by including additional species: gray wolf (Canis lupus), grizzly bear (Ursus arctos), and free ranging or feral domesticated species including cattle, horse, and burro. The costs associated with collisions were also expanded by including passive use, or nonuse values associated with the conservation value of selected wild animal species. The total costs (in 2020 US$) associated with a collision with deer, elk and moose were about 2-3 times (direct costs only) or about 3-4 times higher (direct costs and passive use values combined) compared to the values in 2007 US$. The passive use costs associated with threatened species (wolf, grizzly bear) were higher or much higher than the direct costs. The costs associated with mitigation measures (especially fences and wildlife crossing structures) were also updated and supplemented with new data. New cost-benefit analyses generated updated or entirely new threshold values for deer, elk, moose, and grizzly bear. If collisions with these large wild mammal species reach or surpass the threshold values, it is economically defensible to install the associated type and combination of mitigation measures, both based on direct use and passive use parameters and their associated values. The trend in increasing costs associated with vehicle repair costs, costs associated with human injuries and fatalities, and through including passive use values for wildlife is that we learn that the implementation of effective mitigation measures can be considered earlier and more readily than based on the cost-benefit model published in 2009.
9
Mbani, Benson, Timm Schoening y Jens Greinert. Automated and Integrated Seafloor Classification Workflow (AI-SCW). GEOMAR, mayo de 2023. http://dx.doi.org/10.3289/sw_2_2023.
Texto completoResumen
The Automated and Integrated Seafloor Classification Workflow (AI-SCW) is a semi-automated underwater image processing pipeline that has been customized for use in classifying the seafloor into semantic habitat categories. The current implementation has been tested against a sequence of underwater images collected by the Ocean Floor Observation System (OFOS), in the Clarion-Clipperton Zone of the Pacific Ocean. Despite this, the workflow could also be applied to images acquired by other platforms such as an Autonomous Underwater Vehicle (AUV), or Remotely Operated Vehicle (ROV). The modules in AI-SCW have been implemented using the python programming language, specifically using libraries such as scikit-image for image processing, scikit-learn for machine learning and dimensionality reduction, keras for computer vision with deep learning, and matplotlib for generating visualizations. Therefore, AI-SCW modularized implementation allows users to accomplish a variety of underwater computer vision tasks, which include: detecting laser points from the underwater images for use in scale determination; performing contrast enhancement and color normalization to improve the visual quality of the images; semi-automated generation of annotations to be used downstream during supervised classification; training a convolutional neural network (Inception v3) using the generated annotations to semantically classify each image into one of pre-defined seafloor habitat categories; evaluating sampling strategies for generation of balanced training images to be used for fitting an unsupervised k-means classifier; and visualization of classification results in both feature space view and in map view geospatial co-ordinates. Thus, the workflow is useful for a quick but objective generation of image-based seafloor habitat maps to support monitoring of remote benthic ecosystems.
10
Buesseler, Buessele, Daniele Bianchi, Fei Chai, Jay T. Cullen, Margaret Estapa, Nicholas Hawco, Seth John et al. Paths forward for exploring ocean iron fertilization. Woods Hole Oceanographic Institution, octubre de 2023. http://dx.doi.org/10.1575/1912/67120.
Texto completoResumen
We need a new way of talking about global warming. UN Secretary General António Guterres underscored this when he said the “era of global boiling” has arrived. Although we have made remarkable progress on a very complex problem over the past thirty years, we have a long way to go before we can keep the global temperature increase to below 2°C relative to the pre-industrial times. Climate models suggest that this next decade is critical if we are to avert the worst consequences of climate change. The world must continue to reduce greenhouse gas emissions, and find ways to adapt and build resilience among vulnerable communities. At the same time, we need to find new ways to remove carbon dioxide from the atmosphere in order to chart a “net negative” emissions pathway. Given their large capacity for carbon storage, the oceans must be included in consideration of our multiple carbon dioxide removal (CDR) options. This report focused on ocean iron fertilization (OIF) for marine CDR. This is by no means a new scientific endeavor. Several members of ExOIS (Exploring Ocean Iron Solutions) have been studying this issue for decades, but the emergence of runaway climate impacts has motivated this group to consider a responsible path forward for marine CDR. That path needs to ensure that future choices are based upon the best science and social considerations required to reduce human suffering and counter economic and ecological losses, while limiting and even reversing the negative impacts that climate change is already having on the ocean and the rest of the planet. Prior studies have confirmed that the addition of small amounts of iron in some parts of the ocean is effective at stimulating phytoplankton growth. Through enhanced photosynthesis, carbon dioxide can not only be removed from the atmosphere but a fraction can also be transferred to durable storage in the deep sea. However, prior studies were not designed to quantify how effective this storage can be, or how wise OIF might be as a marine CDR approach. ExOIS is a consortium that was created in 2022 to consider what OIF studies are needed to answer critical questions about the potential efficiency and ecological impacts of marine CDR (http://oceaniron.org). Owing to concerns surrounding the ethics of marine CDR, ExOIS is organized around a responsible code of conduct that prioritizes activities for the collective benefit of our planet with an emphasis on open and transparent studies that include public engagement. Our goal is to establish open-source conventions for implementing OIF for marine CDR that can be assessed with appropriate monitoring, reporting, and verification (MRV) protocols, going beyond just carbon accounting, to assess ecological and other non-carbon environmental effects (eMRV). As urgent as this is, it will still take 5 to 10 years of intensive work and considerable resources to accomplish this goal. We present here a “Paths Forward’’ report that stems from a week-long workshop held at the Moss Landing Marine Laboratories in May 2023 that was attended by international experts spanning atmospheric, oceanographic, and social sciences as well as legal specialists (see inside back cover). At the workshop, we reviewed prior OIF studies, distilled the lessons learned, and proposed several paths forward over the next decade to lay the foundation for evaluating OIF for marine CDR. Our discussion very quickly resulted in a recommendation for the need to establish multiple “Ocean Iron Observatories’’ where, through observations and modeling, we would be able to assess with a high degree of certainty both the durable removal of atmospheric carbon dioxide—which we term the “centennial tonne”—and the ecological response of the ocean. In a five-year phase I period, we prioritize five major research activities: 1. Next generation field studies: Studies of long-term (durable) carbon storage will need to be longer (year or more) and larger (>10,000 km2) than past experiments, organized around existing tools and models, but with greater reliance on autonomous platforms. While prior studies suggested that ocean systems return to ambient conditions once iron infusion is stopped, this needs to be verified. We suggest that these next field experiments take place in the NE Pacific to assess the processes controlling carbon removal efficiencies, as well as the intended and unintended ecological and geochemical consequences. 2. Regional, global and field study modeling Incorporation of new observations and model intercomparisons are essential to accurately represent how iron cycling processes regulate OIF effects on marine ecosystems and carbon sequestration, to support experimental planning for large-scale MRV, and to guide decision making on marine CDR choices. 3. New forms of iron and delivery mechanisms Rigorous testing and comparison of new forms of iron and their potential delivery mechanisms is needed to optimize phytoplankton growth while minimizing the financial and carbon costs of OIF. Efficiency gains are expected to generate responses closer to those of natural OIF events. 4. Monitoring, reporting, and verification: Advances in observational technologies and platforms are needed to support the development, validation, and maintenance of models required for MRV of large-scale OIF deployment. In addition to tracking carbon storage and efficiency, prioritizing eMRV will be key to developing regulated carbon markets. 5. Governance and stakeholder engagement: Attention to social dimensions, governance, and stakeholder perceptions will be essential from the start, with particular emphasis on expanding the diversity of groups engaged in marine CDR across the globe. This feedback will be a critical component underlying future decisions about whether to proceed, or not, with OIF for marine CDR. Paramount in the plan is the need to move carefully. Our goal is to conduct these five activities in parallel to inform decisions steering the establishment of ocean iron observatories at multiple locations in phase II. When completed, this decadal plan will provide a rich knowledge base to guide decisions about if, when, where, and under what conditions OIF might be responsibly implemented for marine CDR. The consensus of our workshop and this report is that now is the time for actionable studies to begin. Quite simply, we suggest that some form of marine CDR will be essential to slow down and reverse the most severe consequences of our disrupted climate. OIF has the potential to be one of these climate mitigation strategies. We have the opportunity and obligation to invest in the knowledge necessary to ensure that we can make scientifically and ethically sound decisions for the future of our planet.