Browsing by Subject "Generative models"

Now showing 1 - 4 of 4

Open Access
Learning efficient visual embedding models under data constraints
(2019-09) Sarıyıldız, Mert Bülent
Deep learning models require large-scale datasets to learn rich sets of low and mid-level patterns and high-level semantics. Therefore, given a high-capacity neural network, one way to improve the performance of a model is increasing the size of the dataset which the model is trained over on. Considering that it is easy to get the amount of computational power required to train a network, data becomes a serious bottleneck in scaling up the existing machine learning pipelines. In this thesis, we look into two main data bottlenecks that rise in computer vision applications: I. the difficulty of finding training data for diverse sets of object categories, II. the complication of utilizing data containing sensitive user information for the purpose of training neural network models. To address these issues, we study zero-shot learning and decentralized learning schemes, respectively. Zero-shot learning (ZSL) is one of the most promising problems where substantial progress can potentially be achieved through unsupervised learning, due to distributional differences between supervised and zero-shot classes. For this reason, several works investigate the incorporation of discriminative domain adaptation techniques into ZSL, which, however, lead to modest improvements in ZSL accuracy. In contrast, we propose a generative model that can naturally learn from unsupervised examples, and synthesize training examples for unseen classes purely based on their class embeddings, and therefore, reduce the zero-shot learning problem into a supervised classification task. The proposed approach consists of two important components: I. a conditional Generative Adversarial Network that learns to produce samples that mimic the characteristics of unsupervised data examples, and II. the Gradient Matching (GM) loss that measures the quality of the gradient signal obtained from the synthesized examples. Using our GM loss formulation, we enforce the generator to produce examples from which accurate classifiers can be trained. Experimental results on several ZSL benchmark datasets show that our approach leads to significant improvements over the state of the art in generalized zero-shot classification. Collaborative learning techniques provide a privacy-preserving solution, by enabling training over a number of private datasets that are not shared by their owners. However, recently, it has been shown that the existing collaborative learning frameworks are vulnerable to an active adversary that runs a generative adversarial network (GAN) attack. In this work, we propose a novel classification model that is resilient against such attacks by design. More specifically, we introduce a key-based classification model and a principled training scheme that protects class scores by using class-specific private keys, which effectively hides the information necessary for a GAN attack. We additionally show how to utilize high dimensional keys to improve the robustness against attacks without increasing the model complexity. Our detailed experiments demonstrate the effectiveness of the proposed technique.
Open Access
Spatio-temporal forecasting over graphs with deep learning
(2020-12) Ceyani, Emir
We study spatiotemporal forecasting of high-dimensional rectangular grid graph structured data, which exhibits both complex spatial and temporal dependencies. In most high-dimensional spatiotemporal forecasting scenarios, deep learningbased methods are widely used. However, deep learning algorithms are overconfident in their predictions, and this overconfidence causes problems in the human-in-the-loop domains such as medical diagnosis and many applications of 5 th generation wireless networks. We propose spatiotemporal extensions to variational autoencoders for regularization, robustness against out-of data distribution, and incorporating uncertainty in predictions to resolve overconfident predictions. However, variational inference methods are prone to biased posterior approximations due to using explicit exponential family densities and mean-field assumption in their posterior factorizations. To mitigate these problems, we utilize variational inference & learning with semi-implicit distributions and apply this inference scheme into convolutional long-short term memory networks(ConvLSTM) for the first time in the literature. In chapter 3, we propose variational autoencoders with convolutional long-short term memory networks, called VarConvLSTM. In chapter 4, we improve our algorithm via semi-implicit & doubly semi-implicit variational inference to model multi-modalities in the data distribution . In chapter 5, we demonstrate that proposed algorithms are applicable for spatiotemporal forecasting tasks, including space-time mobile traffic forecasting over Turkcell base station networks.
Open Access
Style synthesizing conditional generative adversarial networks
(2020-01) Çetin, Yarkın Deniz
Neural style transfer (NST) models aim to transfer a particular visual style to a image while preserving its content using neural networks. Style transfer models that can apply arbitrary styles without requiring style-speciﬁc models or architectures are called universal style transfer (UST) models. Typically a UST model takes a content image and a style image as inputs and outputs the corresponding stylized image. It is, therefore, required to have a style image with the required characteristics to facilitate the transfer. However, in practical applications, where the user wants to apply variations of a style class or a mixture of multiple style classes, such style images may be diﬃcult to ﬁnd or simply non-existent. In this work we propose a conditional style transfer network which can model multiple style classes. While our model requires training examples (style images) for each class at training time, it does not require any style images at test time. The model implicitly learns the manifold of each style and is able to generate diverse stylization outputs corresponding to a single style class or a mixture of the available style classes. This requires the model to be able to learn one-to-many mappings, from an input single class label to multiple styles. For this reason, we build our model based on generative adversarial networks (GAN), which have been shown to generate realistic data in highly complex and multi-modal distributions in numerous domains. More speciﬁcally, we design a conditional GAN model that takes a semantic conditioning vector specifying the desired style class(es) and a noise vector as input and outputs the statistics required for applying style transfer. In order to achieve style transfer, we adapt a preexisting encoder-decoder based universal style transfer model. The encoder component extracts convolutional feature maps from the content image. These features are ﬁrst whitened and then colorized using the statistics of the input style image. The decoder component then reconstructs the stylized image from the colorized features. In our adaptation, instead of using full covariance matrices, we approximate the whitening and coloring transforms using diagonal elements of the covariance matrices. We then remove the dependence to the input style image by learning to generate the statistics via our GAN model. In our experiments, we use a subset of the WikiArt dataset to train and validate our approach. We demonstrate that our approximation method achieves stylization results similar to the preexisting model but with higher speeds and using a fraction of target style statistics. We also show that our conditional GAN model leads to successful style transfer results by learning the manifold of styles corresponding to each style class. We additionally show that the GAN model can be used to generate novel style class combinations, which are highly correlated with the corresponding actual stylization results that are not seen during training.
Open Access
Utrgan: Learning to generate 5’ UTR sequences for optimized translation efficiency and gene expression
(2024-07) Barazandeh, Sina
The 5’ untranslated region (5’ UTR) of the messenger RNA plays a crucial role in the translatability and stability of the molecule. Thus, it is an important component in the design of synthetic biological circuits for high and stable expression of intermediate proteins. Several UTR sequences are patented and used frequently in laboratories. We present a novel model, UTRGAN, which is a Generative Adversarial Network (GAN)-based model designed to generate 5’ UTR sequences coupled with an optimization procedure to ensure a target feature such as high expression for a target gene sequence or high ribosome load and translation efficiency. We rigorously analyze and show that the model can generate sequences that mimic various properties of natural UTR sequences. Then, we show that the optimization procedure yields sequences that are expected to yield (i) up to 5-fold higher average expression on a set of target genes, (ii) up to 2-fold higher mean ribosome load on average, and (iii) a 34-fold higher average translation efficiency, compared to the initially generated UTR sequences. We compare our method with other approaches and find that UTRGANgenerated sequences contain motives that show higher sequence similarity to known regularity motives in various regions such as (i) internal ribosome entry sites, (ii) upstream open reading frames, (iii) G-quadruplexes, (iv) Kozak and initiation start codon regions. Finally, we show in-vitro that the UTR sequences we designed yield a higher translation rate for the human TNF-α protein compared to the human Beta Globin 5’ UTR, which is a 5’ UTR with high production capacity.