The work deals with the issue of handrwritten text recognition problem with deep neural networks. It focuses on the use of sequence to sequence method using encoder-decoder model. It also includes design of encoder-decoder model for handwritten text recognition using a transformer instead of recurrent neurons and a set of experiments that were performed on it.

The aim of this Master's thesis is to describe basic technics of evolutionary computing, convolutional neural networks (CNN), and automated design of neural networks using neuroevolution ( NAS - Neural Architecture Search ). NAS techniques are currently being researched more and more, as they speed up and simplify the lengthy and complicated process of designing artificial neural networks. These techniques are also able to search for unconventional architectures that would not be found by classic methods. The work also contains the design and implementation of software capable of automated development of convolutional neural networks using the open-source library TensorFlow. The program uses a multiobjective NSGA-II algorithm for designing accurate and compact CNNs.

The aim of this diploma thesis is to study and design experimental improvement of the convolutional neural network for disease detection. Another goal is to extend the classifier with a new type of detection. he new type of detection is damage fingerprint by pressure. The experimentally improved convolutional network is implemented by PyTorch. The network detects which part of the fingerprint is damaged and draws this part into the fingerprint. Synthetic fingerprints are used when training the net. Real fingerprints are added to the synthetic fingerprints.

No description available.

We address the problem of estimating a high quality dense depth map from a single RGB input image. We start out with a baseline encoder-decoder convolutional neural network architecture and pose the question of how the global processing of information can help improve overall depth estimation. To this end, we propose a transformer-based architecture block that divides the depth range into bins whose center value is estimated adaptively per image. The final depth values are estimated as linear combinations of the bin centers. We call our new building block AdaBins. Our results show a decisive improvement over the state-of-the-art on several popular depth datasets across all metrics. We also validate the effectiveness of the proposed block with an ablation study.

This thesis is concerned with the automatic altitude estimation from a single landscape photograph. I solved this task using convolutional neural networks. There was no suitable training dataset available having information about image altitude, thus I  had to create a new one. To estimate human performance in altitude estimation task, an experiment was conducted counting 100 subjects. The goal of this experiment was to measure the accuracy of the human estimate of camera altitude from an image. The measured average estimation error of subjects was 879 m. An automatic system based on convolutional neural networks outperforms humans with an average elevation error 712 m. The proposed system can be used in more complex scenario like the visual camera geo-localization.

To be able to operate in the world around us, we need to store visual information for further processing. Since we are able to memorize a vast array of visual scenes (photographs of the outside world), it is still an open question of how we represent these scenes in memory. Research shows that perception and memory for visual scenes is a complex problem that requires contribution from many subfields of vision science. In this work we focused on the visual scene memory on the creation of perceptual prototypes. Using convolutional neural networks, we defined the similarity of scenes in the scene space, which we used in two experiments. In the first experiment, we validated this space using a paradigm for detecting an odd scene. In the second experiment, using the Deese-Roediger-McDermott paradigm, we verified the creation of false memories and thus visual prototypes. The results show that people intuitively understand the scene space (Experiment 1) and that a visual prototype is created even in the case of the complex stimuli such as scenes. The results have wide application either for machine evaluation of image similarities or for visual memory research.

Machine learning and its wide range of applications is becoming increasingly prevalent in both academia and industry. This thesis will focus on the two machine learning methods convolutional neural networks and reinforcement learning. Convolutional neural networks has seen great success in various applications for both classification and regression problems in a diverse range of fields, e.g. vision for self-driving cars or facial recognition. These networks are built on a set of trainable weights optimized on data, and a set of hyperparameters set by the designer of the network which will remain constant. For the network to perform well, the hyperparameters have to be optimized separately. The goal of this thesis is to investigate the use of reinforcement learning as a method for optimizing hyperparameters in convolutional neural networks built for classification problems. The reinforcement learning methods used are a tabular Q-learning and a new Q-learning inspired algorithm denominated max-table. These algorithms have been tested with different exploration policies based on each hyperparameter value’s covariance, precision or relevance to the performance metric. The reinforcement learning algorithms were mostly tested on the datasets CIFAR10 and MNIST fashion against a baseline set by random search. While the Q-learning algorithm was not able to perform better than random search, max-table was able to perform better than random search in 50% of the time on both datasets. Hyperparameterbased exploration policy using covariance and relevance were shown to decrease the optimizers’ performance. No significant difference was found between a hyperparameter based exploration policy using performance and an equally distributed exploration policy. / Maskininlärning och dess många tillämpningsområden blir vanligare i både akademin och industrin. Den här uppsatsen fokuserar på två maskininlärningsmetoder, faltande neurala nätverk och förstärkningsinlärning. Faltande neurala nätverk har sett stora framgångar inom olika applikationsområden både för klassifieringsproblem och regressionsproblem inom diverse fält, t.ex. syn för självkörande bilar eller ansiktsigenkänning. Dessa nätverk är uppbyggda på en uppsättning av tränbara parameterar men optimeras på data, samt en uppsättning hyperparameterar bestämda av designern och som hålls konstanta vilka behöver optimeras separat för att nätverket ska prestera bra. Målet med denna uppsats är att utforska användandet av förstärkningsinlärning som en metod för att optimera hyperparameterar i faltande neurala nätverk gjorda för klassifieringsproblem. De förstärkningsinlärningsmetoder som använts är en tabellarisk "Q-learning" samt en ny "Q-learning" inspirerad metod benämnd "max-table". Dessa algoritmer har testats med olika handlingsmetoder för utforskning baserade på hyperparameterarnas värdens kovarians, precision eller relevans gentemot utvärderingsmetriken. Förstärkningsinlärningsalgoritmerna var i största del testade på dataseten CIFAR10 och MNIST fashion och jämförda mot en baslinje satt av en slumpmässig sökning. Medan "Q-learning"-algoritmen inte kunde visas prestera bättre än den slumpmässiga sökningen, kunde "max-table" prestera bättre på 50\% av tiden på både dataseten. De handlingsmetoder för utforskning som var baserade på kovarians eller relevans visades minska algoritmens prestanda. Ingen signifikant skillnad kunde påvisas mellan en handlingsmetod baserad på hyperparametrarnas precision och en jämnt fördelad handlingsmetod för utforsking.

This work researches named entity recognition (NER) with respect to images of documents with a domain-specific layout, by means of Convolutional Neural Networks (CNNs). Examples of such documents are receipts, invoices, forms and scientific papers, the latter of which are used in this work. An NER task is first performed statically, where a static number of entity classes is extracted per document. Networks based on the deep VGG-16 network are used for this task. Here, experimental evaluation shows that framing the task as a classification task, where the network classifies each bounding box coordinate separately, leads to the best network performance. Also, a multi-headed architecture is introduced, where the network has an independent fully-connected classification head per entity. VGG-16 achieves better performance with the multi-headed architecture than with its default, single-headed architecture. Additionally, it is shown that transfer learning does not improve performance of these networks. Analysis suggests that the networks trained for the static NER task learn to recognise document templates, rather than the entities themselves, and therefore do not generalize well to new, unseen templates. For a dynamic NER task, where the type and number of entity classes vary per document, experimental evaluation shows that, on large entities in the document, the Faster R-CNN object detection framework achieves comparable performance to the networks trained on the static task. Analysis suggests that Faster R-CNN generalizes better to new templates than the networks trained for the static task, as Faster R-CNN is trained on local features rather than the full document template. Finally, analysis shows that Faster R-CNN performs poorly on small entities in the image and suggestions are made to improve its performance.

Machine learning methods provide an opportunity to improve the classification of skin lesions and the early diagnosis of melanoma by providing decision support for general practitioners. So far most studies have been looking at the creation of features that best indicate melanoma. Representation learning methods such as neural networks have outperformed hand-crafted features in many areas. This work aims to evaluate the performance of convolutional neural networks in relation to earlier machine learning algorithms and expert diagnosis. In this work, convolutional neural networks were trained on datasets of dermoscopy images using weights initialized from a random distribution, a network trained on the ImageNet dataset and a network trained on Dermnet, a skin disease atlas.  The ensemble sum prediction of the networks achieved an accuracy of 89.3% with a sensitivity of 77.1% and a specificity of 93.0% when based on the weights learned from the ImageNet dataset and the Dermnet skin disease atlas and trained on non-polarized light dermoscopy images.  The results from the different networks trained on little or no prior data confirms the idea that certain features are transferable between different data. Similar classification accuracies to that of the highest scoring network are achieved by expert dermatologists and slightly higher results are achieved by referenced hand-crafted classifiers.  The trained networks are found to be comparable to practicing dermatologists and state-of-the-art machine learning methods in binary classification accuracy, benign – melanoma, with only little pre-processing and tuning.