'
\nThe following is a list of academic research papers that will make you laugh, cry, and maybe even question your entire existence.
\nEnjoy! <3
Deep learning for whole-body medical image generation
\n\n\nTL;DR: This study investigated deep learning transformation in medical imaging. Whole-body MR patient data were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach.
\n
Artificial intelligence (AI) algorithms based on deep convolutional networks have demonstrated remarkable success for image transformation tasks. State-of-the-art results have been achieved by generative adversarial networks (GANs) and training approaches which do not require paired data. Recently, these techniques have been applied in the medical field for cross-domain image translation. This study investigated deep learning transformation in medical imaging. It was motivated to identify generalizable methods which would satisfy the simultaneous requirements of quality and anatomical accuracy across the entire human body. Specifically, whole-body MR patient data acquired on a PET/MR system were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods, which typically use multiphase Dixon sequences to segment various tissue types. This work aimed to investigate the technical performance of a GAN system for general MR-to-CT volumetric transformation and to evaluate the performance of the generated images for PET AC. A dataset comprising matched, same-day PET/MR and PET/CT patient scans was used for validation. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach. Over the entire body, the total amounts of reconstructed PET activities were similar between the two MR-AC methods, but the synthetic CT method yielded higher accuracy for quantifying the tracer uptake in specific regions. The findings reported here demonstrate the feasibility of this technique and its potential to improve certain aspects of attenuation correction for PET/MR systems. Moreover, this work may have larger implications for establishing generalized methods for inter-modality, whole-body transformation in medical imaging. Unsupervised deep learning techniques can produce high-quality synthetic images, but additional constraints may be needed to maintain medical integrity in the generated data.
\n\n\nTL;DR: We compared the performance of two deep learning models, a GAN and a ResNet, for the task of MR-based synthetic CT (sCT) generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR tumor pretreatment planning.
\n
Background: Currently, MRI-only radiotherapy (RT) eliminates some of the concerns about using CT images in RT chains such as the registration of MR images to a separate CT, extra dose delivery, and the additional cost of repeated imaging. However, one remaining challenge is that the signal intensities of MRI are not related to the attenuation coefficient of the biological tissue. This work compares the performance of two state-of-the-art deep learning models; a generative adversarial network (GAN) and a residual network (ResNet) for synthetic CTs (sCT) generation from MR images. Materials and Methods: The brain MR and CT images of 86 participants were analyzed. GAN and ResNet models were implemented for the generation of synthetic CTs from the 3D T1-weighted MR images using a six-fold cross-validation scheme. The resulting sCTs were compared, considering the CT images as a reference using standard metrics such as the mean absolute error (MAE), peak signal-to-noise-ratio (PSNR) and the structural similarity index (SSIM). Results: Overall, the ResNet model exhibited higher accuracy in relation to the delineation of brain tissues. The ResNet model estimated the CT values for the entire head region with an MAE of 114.1±27.5 HU compared to MAE=-10.9±147.0 HU obtained from the GAN model. Moreover, both models offered comparable SSIM and PSNR values, although the ResNet method exhibited a slightly superior performance over the GAN method. Conclusion: We compared two state-of-the-art deep learning models for the task of MR-based sCT generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR AC and MR-only RT planning. ► Original article
\nDeep learning for molecular generation and optimization - a review of the state of the art
\nWe review a recent groundswell of work which uses deep learning techniques to generate and optimize molecules.
\nDeep Learning for Source Code Modeling and Generation
\n\n\nTL;DR: Deep Learning methods are showing great promise for natural language processing tasks such as modeling and generation of source code. We provide a comprehensive review of these methods, grouped by the type of task they address. We then formulate common program learning tasks under an encoder-decoder framework, and introduce recent Deep Learning mechanisms suitable to solve them. Finally, we discuss the current state-of-the-art and challenges for practitioners and researchers.
\n
Deep Learning (DL) techniques for Natural Language Processing have been evolving remarkably fast. Recently, the DL advances in language modeling, machine translation, and paragraph understanding are so prominent that the potential of DL in Software Engineering cannot be overlooked, especially in the field of program learning. To facilitate further research and applications of DL in this field, we provide a comprehensive review to categorize and investigate existing DL methods for source code modeling and generation. To address the limitations of the traditional source code models, we formulate common program learning tasks under an encoder-decoder framework. After that, we introduce recent DL mechanisms suitable to solve such problems. Then, we present the state-of-the-art practices and discuss their challenges with some recommendations for practitioners and researchers as well.
\nTowards a new generation of parton densities with deep learning models
\n\n\nTL;DR: We use a deep learning inspired approach to regression for the determination of parton distribution functions. Our approach uses efficient computing frameworks based on graph generated models for PDF parametrization and gradient descent optimization. Results from our approach outperform the current state-of-the-art PDF fitting methodology in terms of model selection and computational resources usage.
\n
We present a new regression model for the determination of parton distribution functions (PDF) using techniques inspired from deep learning projects. In the context of the NNPDF methodology, we implement a new efficient computing framework based on graph generated models for PDF parametrization and gradient descent optimization. The best model configuration is derived from a robust cross-validation mechanism through a hyperparametrization tune procedure. We show that results provided by this new framework outperforms the current state-of-the-art PDF fitting methodology in terms of best model selection and computational resources usage.
\nReAssert: Deep Learning for Assert Generation
\n\n\nTL;DR: Re-ASSERT is a method for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first.
\n
The automated generation of test code can reduce the time and effort required to build software while increasing its correctness and robustness. In this paper, we present RE-ASSERT, an approach for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first. We also utilise Reformer, a state-of-the-art deep learning model, along with two models from previous work to evaluate ReAssert and an existing approach, known as ATLAS, using lexical accuracy,uniqueness, and dynamic analysis. Our evaluation of ReAssert shows up to 44% of generated asserts for a single project match exactly with the ground truth, increasing to 51% for generated asserts that compile. We also improve on the ATLAS results through our use of Reformer with 28% of generated asserts matching exactly with the ground truth. Reformer also produces the greatest proportion of unique asserts (71%), giving further evidence that Reformer produces the most useful asserts.
\nGenerative Deep Learning Techniques for Password Generation
\nPassword guessing approaches via deep learning have recently been investigated with significant breakthroughs in their ability to generate novel, realistic password candidates. In the present work we study a broad collection of deep learning and probabilistic based models in the light of password guessing: attention-based deep neural networks, autoencoding mechanisms and generative adversarial networks. We provide novel generative deep-learning models in terms of variational autoencoders exhibiting state-of-art sampling performance, yielding additional latent-space features such as interpolations and targeted sampling. Lastly, we perform a thorough empirical analysis in a unified controlled framework over well-known datasets (RockYou, LinkedIn, Youku, Zomato, Pwnd). Our results not only identify the most promising schemes driven by deep neural networks, but also illustrate the strengths of each approach in terms of generation variability and sample uniqueness.
\nResearch on Image Generation and Style Transfer Algorithm Based on Deep Learning
\nat the current process of artistic creation and animation creation, there are a
\nlot of repeated manual operations in the process of conversion from sketch to the stylized image. This paper presented a solution based on a deep
\nlearning framework to realize image generation and style transfer. The method
\nfirst used the conditional generation to resist the network, optimizes the loss
\nfunction of the training mapping relationship, and generated the actual image
\nfrom the input sketch. Then, by defining and optimizing the perceptual loss
\nfunction of the style transfer model, the style features are extracted from the
\nimage, thereby forming the actual The conversion between images and stylized
\nart images. Experiments show that this method can greatly reduce the work of
\ncoloring and converting with different artistic effects, and achieve the purpose of transforming simple stick
\nfigures into actual object images.
Applications of Deep Learning to Audio Generation
\n\n\nTL;DR: Deep learning algorithms are characterized by multiple hidden layers of processing units, which can be trained to automatically learn features of data. These deep learning architectures have demonstrated remarkable success in a number of domains, including speech recognition and acoustic signal processing. In this article, we provide an overview of the deep learning techniques and their effectiveness for speech/acoustic signal processing applications.
\n
In the recent past years, deep learning based machine learning systems have demonstrated remarkable success for a wide range of learning tasks in multiple domains such as computer vision, speech recognition and other pattern recognition based applications. The purpose of this article is to contribute a timely review and introduction of state-of-the-art deep learning techniques and their effectiveness in speech/acoustic signal processing. Thorough investigations of various deep learning architectures are provided under the categories of discriminative and generative algorithms, including the up-to-date Generative Adversarial Networks (GANs) as an integrated model. A comprehensive overview of applications in audio generation is highlighted. Based on understandings from these approaches, we discuss how deep learning methods can benefit the field of speech/acoustic signal synthesis and the potential issues that need to be addressed for prospective real-world scenarios. We hope this survey provides a valuable reference for practitioners seeking to innovate in the usage of deep learning approaches for speech/acoustic signal generation.
\nMusic Generation Using Deep Learning
\nDeep learning has recently been used for many art-related activities such as automatic generation of music and pictures. This paper deals with music generation by using raw audio files in the frequency domain using Restricted Boltzmann Machine and Long Short- Term Memory architectures. The work does not use any information about musical structure to aid the learning, instead, it learns from a previous permutation of notes and generates an optimal and pleasant permutation. It also serves as a comparative study for music generation using Long Short-Term Memory and Restricted Boltzmann Machine.
\nDeep Learning for Expressive Music Generation
\n\n\nTL;DR: We study the expressive potential of musical elements such as note velocity and duration in artificially generated music compositions. We find that these elements are important for musical expression and that including them in the generated pieces improves their quality.
\n
In the last decade, Deep Learning (DL) algorithms have been increasing its popularity in several fields such as computer vision, speech recognition, natural language processing and many others. DL models, however, are not limited to scientific domains as they have recently been applied to content generation in diverse art forms - both in the generation of novel contents and as co-creative tools. Artificial music generation is one of the fields where DL architectures have been applied. They have been mostly used to create new compositions exhibiting promising results when compared to human compositions. Despite this, the majority of these artificial pieces lack some expression when compared to music compositions performed by humans. In this document, we propose a system capable of artificially generating expressive music compositions. Our main goal is to improve the quality of the musical compositions generated by the artificial system by exploring perceptually relevant musical elements such as note velocity and duration. To assess this hypothesis, we perform user tests. Results suggest that expressive elements such as duration and velocity are key aspects in a music composition expression, making the ones who include these preferable to non-expressive ones.
\nDeep Learning Based Program Generation From Requirements Text: Are We There Yet?
\n\n\nTL;DR: We present a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements
\n
To release developers from time-consuming software development, many approaches have been proposed to generate source code automatically according to software requirements. With significant advances in deep learning and natural language processing, deep learning-based approaches are proposed to generate source code from natural language descriptions. The key insight is that given a large corpus of software requirements and their corresponding implementations, advanced deep learning techniques may learn how to translate software requirements into source code that fulfill such requirements. Although such approaches are reported to be highly accurate, they are evaluated on datasets that are rather small, lack of diversity, and significantly different from real-world software requirements. To this end, we build a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements, and most of the successfully generated statements are popular statements in the training programs. Based on this finding, we propose a popularity-based approach that always generates the most popular statements in training programs regardless of the input (software requirements). Evaluation results suggest that none of the state-of-the-art approaches can outperform this simple statistics-based approach. As a conclusion, deep learning-based program generation requires significant improvement in the future, and our dataset may serve as a basis for future research in this direction.
\nA Review of Text Style Transfer using Deep Learning
\nStyle is an integral component of a sentence indicated by the choice of words a person makes. Different people have different ways of expressing themselves, however, they adjust their speaking and writing style to a social context, an audience, an interlocutor or the formality of an occasion. Text style transfer is defined as a task of adapting and/or changing the stylistic manner in which a sentence is written, while preserving the meaning of the original sentence. A systematic review of text style transfer methodologies using deep learning is presented in this paper. We point out the technological advances in deep neural networks that have been the driving force behind current successes in the fields of natural language understanding and generation. The review is structured around two key stages in the text style transfer process, namely, representation learning and sentence generation in a new style. The discussion highlights the commonalities and differences between proposed solutions as well as challenges and opportunities that are expected to direct and foster further research in the field. Impact Statement—Motivated by recent advancements in the field, we have carried out a systematic review of state-of-the-art research to highlight the trends, commonalities and differences across style transfer methodologies using deep learning. The discussion is organized around key stages of the process, namely, representation learning of style and content of a given sentence, and generation of the sentence in a new style. A comprehensive view of methodologies, available datasets and evaluation metrics is compiled to foster further research in the field.
\n'
\nThe following is a list of academic research papers that will make you laugh, cry, and maybe even question your entire existence.
\nEnjoy! <3
Deep learning for whole-body medical image generation
\n\n\nTL;DR: This study investigated deep learning transformation in medical imaging. Whole-body MR patient data were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach.
\n
Artificial intelligence (AI) algorithms based on deep convolutional networks have demonstrated remarkable success for image transformation tasks. State-of-the-art results have been achieved by generative adversarial networks (GANs) and training approaches which do not require paired data. Recently, these techniques have been applied in the medical field for cross-domain image translation. This study investigated deep learning transformation in medical imaging. It was motivated to identify generalizable methods which would satisfy the simultaneous requirements of quality and anatomical accuracy across the entire human body. Specifically, whole-body MR patient data acquired on a PET/MR system were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods, which typically use multiphase Dixon sequences to segment various tissue types. This work aimed to investigate the technical performance of a GAN system for general MR-to-CT volumetric transformation and to evaluate the performance of the generated images for PET AC. A dataset comprising matched, same-day PET/MR and PET/CT patient scans was used for validation. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach. Over the entire body, the total amounts of reconstructed PET activities were similar between the two MR-AC methods, but the synthetic CT method yielded higher accuracy for quantifying the tracer uptake in specific regions. The findings reported here demonstrate the feasibility of this technique and its potential to improve certain aspects of attenuation correction for PET/MR systems. Moreover, this work may have larger implications for establishing generalized methods for inter-modality, whole-body transformation in medical imaging. Unsupervised deep learning techniques can produce high-quality synthetic images, but additional constraints may be needed to maintain medical integrity in the generated data.
\n\n\nTL;DR: We compared the performance of two deep learning models, a GAN and a ResNet, for the task of MR-based synthetic CT (sCT) generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR tumor pretreatment planning.
\n
Background: Currently, MRI-only radiotherapy (RT) eliminates some of the concerns about using CT images in RT chains such as the registration of MR images to a separate CT, extra dose delivery, and the additional cost of repeated imaging. However, one remaining challenge is that the signal intensities of MRI are not related to the attenuation coefficient of the biological tissue. This work compares the performance of two state-of-the-art deep learning models; a generative adversarial network (GAN) and a residual network (ResNet) for synthetic CTs (sCT) generation from MR images. Materials and Methods: The brain MR and CT images of 86 participants were analyzed. GAN and ResNet models were implemented for the generation of synthetic CTs from the 3D T1-weighted MR images using a six-fold cross-validation scheme. The resulting sCTs were compared, considering the CT images as a reference using standard metrics such as the mean absolute error (MAE), peak signal-to-noise-ratio (PSNR) and the structural similarity index (SSIM). Results: Overall, the ResNet model exhibited higher accuracy in relation to the delineation of brain tissues. The ResNet model estimated the CT values for the entire head region with an MAE of 114.1±27.5 HU compared to MAE=-10.9±147.0 HU obtained from the GAN model. Moreover, both models offered comparable SSIM and PSNR values, although the ResNet method exhibited a slightly superior performance over the GAN method. Conclusion: We compared two state-of-the-art deep learning models for the task of MR-based sCT generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR AC and MR-only RT planning. ► Original article
\nDeep learning for molecular generation and optimization - a review of the state of the art
\nWe review a recent groundswell of work which uses deep learning techniques to generate and optimize molecules.
\nDeep Learning for Source Code Modeling and Generation
\n\n\nTL;DR: Deep Learning methods are showing great promise for natural language processing tasks such as modeling and generation of source code. We provide a comprehensive review of these methods, grouped by the type of task they address. We then formulate common program learning tasks under an encoder-decoder framework, and introduce recent Deep Learning mechanisms suitable to solve them. Finally, we discuss the current state-of-the-art and challenges for practitioners and researchers.
\n
Deep Learning (DL) techniques for Natural Language Processing have been evolving remarkably fast. Recently, the DL advances in language modeling, machine translation, and paragraph understanding are so prominent that the potential of DL in Software Engineering cannot be overlooked, especially in the field of program learning. To facilitate further research and applications of DL in this field, we provide a comprehensive review to categorize and investigate existing DL methods for source code modeling and generation. To address the limitations of the traditional source code models, we formulate common program learning tasks under an encoder-decoder framework. After that, we introduce recent DL mechanisms suitable to solve such problems. Then, we present the state-of-the-art practices and discuss their challenges with some recommendations for practitioners and researchers as well.
\nTowards a new generation of parton densities with deep learning models
\n\n\nTL;DR: We use a deep learning inspired approach to regression for the determination of parton distribution functions. Our approach uses efficient computing frameworks based on graph generated models for PDF parametrization and gradient descent optimization. Results from our approach outperform the current state-of-the-art PDF fitting methodology in terms of model selection and computational resources usage.
\n
We present a new regression model for the determination of parton distribution functions (PDF) using techniques inspired from deep learning projects. In the context of the NNPDF methodology, we implement a new efficient computing framework based on graph generated models for PDF parametrization and gradient descent optimization. The best model configuration is derived from a robust cross-validation mechanism through a hyperparametrization tune procedure. We show that results provided by this new framework outperforms the current state-of-the-art PDF fitting methodology in terms of best model selection and computational resources usage.
\nReAssert: Deep Learning for Assert Generation
\n\n\nTL;DR: Re-ASSERT is a method for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first.
\n
The automated generation of test code can reduce the time and effort required to build software while increasing its correctness and robustness. In this paper, we present RE-ASSERT, an approach for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first. We also utilise Reformer, a state-of-the-art deep learning model, along with two models from previous work to evaluate ReAssert and an existing approach, known as ATLAS, using lexical accuracy,uniqueness, and dynamic analysis. Our evaluation of ReAssert shows up to 44% of generated asserts for a single project match exactly with the ground truth, increasing to 51% for generated asserts that compile. We also improve on the ATLAS results through our use of Reformer with 28% of generated asserts matching exactly with the ground truth. Reformer also produces the greatest proportion of unique asserts (71%), giving further evidence that Reformer produces the most useful asserts.
\nGenerative Deep Learning Techniques for Password Generation
\nPassword guessing approaches via deep learning have recently been investigated with significant breakthroughs in their ability to generate novel, realistic password candidates. In the present work we study a broad collection of deep learning and probabilistic based models in the light of password guessing: attention-based deep neural networks, autoencoding mechanisms and generative adversarial networks. We provide novel generative deep-learning models in terms of variational autoencoders exhibiting state-of-art sampling performance, yielding additional latent-space features such as interpolations and targeted sampling. Lastly, we perform a thorough empirical analysis in a unified controlled framework over well-known datasets (RockYou, LinkedIn, Youku, Zomato, Pwnd). Our results not only identify the most promising schemes driven by deep neural networks, but also illustrate the strengths of each approach in terms of generation variability and sample uniqueness.
\nResearch on Image Generation and Style Transfer Algorithm Based on Deep Learning
\nat the current process of artistic creation and animation creation, there are a
\nlot of repeated manual operations in the process of conversion from sketch to the stylized image. This paper presented a solution based on a deep
\nlearning framework to realize image generation and style transfer. The method
\nfirst used the conditional generation to resist the network, optimizes the loss
\nfunction of the training mapping relationship, and generated the actual image
\nfrom the input sketch. Then, by defining and optimizing the perceptual loss
\nfunction of the style transfer model, the style features are extracted from the
\nimage, thereby forming the actual The conversion between images and stylized
\nart images. Experiments show that this method can greatly reduce the work of
\ncoloring and converting with different artistic effects, and achieve the purpose of transforming simple stick
\nfigures into actual object images.
Applications of Deep Learning to Audio Generation
\n\n\nTL;DR: Deep learning algorithms are characterized by multiple hidden layers of processing units, which can be trained to automatically learn features of data. These deep learning architectures have demonstrated remarkable success in a number of domains, including speech recognition and acoustic signal processing. In this article, we provide an overview of the deep learning techniques and their effectiveness for speech/acoustic signal processing applications.
\n
In the recent past years, deep learning based machine learning systems have demonstrated remarkable success for a wide range of learning tasks in multiple domains such as computer vision, speech recognition and other pattern recognition based applications. The purpose of this article is to contribute a timely review and introduction of state-of-the-art deep learning techniques and their effectiveness in speech/acoustic signal processing. Thorough investigations of various deep learning architectures are provided under the categories of discriminative and generative algorithms, including the up-to-date Generative Adversarial Networks (GANs) as an integrated model. A comprehensive overview of applications in audio generation is highlighted. Based on understandings from these approaches, we discuss how deep learning methods can benefit the field of speech/acoustic signal synthesis and the potential issues that need to be addressed for prospective real-world scenarios. We hope this survey provides a valuable reference for practitioners seeking to innovate in the usage of deep learning approaches for speech/acoustic signal generation.
\nMusic Generation Using Deep Learning
\nDeep learning has recently been used for many art-related activities such as automatic generation of music and pictures. This paper deals with music generation by using raw audio files in the frequency domain using Restricted Boltzmann Machine and Long Short- Term Memory architectures. The work does not use any information about musical structure to aid the learning, instead, it learns from a previous permutation of notes and generates an optimal and pleasant permutation. It also serves as a comparative study for music generation using Long Short-Term Memory and Restricted Boltzmann Machine.
\nDeep Learning for Expressive Music Generation
\n\n\nTL;DR: We study the expressive potential of musical elements such as note velocity and duration in artificially generated music compositions. We find that these elements are important for musical expression and that including them in the generated pieces improves their quality.
\n
In the last decade, Deep Learning (DL) algorithms have been increasing its popularity in several fields such as computer vision, speech recognition, natural language processing and many others. DL models, however, are not limited to scientific domains as they have recently been applied to content generation in diverse art forms - both in the generation of novel contents and as co-creative tools. Artificial music generation is one of the fields where DL architectures have been applied. They have been mostly used to create new compositions exhibiting promising results when compared to human compositions. Despite this, the majority of these artificial pieces lack some expression when compared to music compositions performed by humans. In this document, we propose a system capable of artificially generating expressive music compositions. Our main goal is to improve the quality of the musical compositions generated by the artificial system by exploring perceptually relevant musical elements such as note velocity and duration. To assess this hypothesis, we perform user tests. Results suggest that expressive elements such as duration and velocity are key aspects in a music composition expression, making the ones who include these preferable to non-expressive ones.
\nDeep Learning Based Program Generation From Requirements Text: Are We There Yet?
\n\n\nTL;DR: We present a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements
\n
To release developers from time-consuming software development, many approaches have been proposed to generate source code automatically according to software requirements. With significant advances in deep learning and natural language processing, deep learning-based approaches are proposed to generate source code from natural language descriptions. The key insight is that given a large corpus of software requirements and their corresponding implementations, advanced deep learning techniques may learn how to translate software requirements into source code that fulfill such requirements. Although such approaches are reported to be highly accurate, they are evaluated on datasets that are rather small, lack of diversity, and significantly different from real-world software requirements. To this end, we build a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements, and most of the successfully generated statements are popular statements in the training programs. Based on this finding, we propose a popularity-based approach that always generates the most popular statements in training programs regardless of the input (software requirements). Evaluation results suggest that none of the state-of-the-art approaches can outperform this simple statistics-based approach. As a conclusion, deep learning-based program generation requires significant improvement in the future, and our dataset may serve as a basis for future research in this direction.
\nA Review of Text Style Transfer using Deep Learning
\nStyle is an integral component of a sentence indicated by the choice of words a person makes. Different people have different ways of expressing themselves, however, they adjust their speaking and writing style to a social context, an audience, an interlocutor or the formality of an occasion. Text style transfer is defined as a task of adapting and/or changing the stylistic manner in which a sentence is written, while preserving the meaning of the original sentence. A systematic review of text style transfer methodologies using deep learning is presented in this paper. We point out the technological advances in deep neural networks that have been the driving force behind current successes in the fields of natural language understanding and generation. The review is structured around two key stages in the text style transfer process, namely, representation learning and sentence generation in a new style. The discussion highlights the commonalities and differences between proposed solutions as well as challenges and opportunities that are expected to direct and foster further research in the field. Impact Statement—Motivated by recent advancements in the field, we have carried out a systematic review of state-of-the-art research to highlight the trends, commonalities and differences across style transfer methodologies using deep learning. The discussion is organized around key stages of the process, namely, representation learning of style and content of a given sentence, and generation of the sentence in a new style. A comprehensive view of methodologies, available datasets and evaluation metrics is compiled to foster further research in the field.
\n@satoshidatamoto these look interesting, thanks for sharing :)
", "rawMarkdown": "satoshidatamoto these look interesting, thanks for sharing :)", "votes": null }, { "id": "1779010", "postDate": "05/06/2022 02:07:03", "content": "@satoshidatamoto Thank you for sharing
", "rawMarkdown": "satoshidatamoto Thank you for sharing", "votes": null } ], "comments": [ { "id": 1779003, "author_name": "lachlangillian", "author_url": "", "post_date": "05/06/2022 01:56:04", "content": "@satoshidatamoto these look interesting, thanks for sharing :)
", "votes": null, "replies": [] }, { "id": 1779010, "author_name": "alanjo", "author_url": "", "post_date": "05/06/2022 02:07:03", "content": "@satoshidatamoto Thank you for sharing
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1778981": "'\n\nThe following is a list of academic research papers that will make you laugh, cry, and maybe even question your entire existence.\nEnjoy! <3\n\n_____\n**[Deep learning for whole-body medical image generation](https://doi.org/10.1007/s00259-021-05413-0)**\n\n\n> **TL;DR:** This study investigated deep learning transformation in medical imaging. Whole-body MR patient data were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach.\n\nArtificial intelligence (AI) algorithms based on deep convolutional networks have demonstrated remarkable success for image transformation tasks. State-of-the-art results have been achieved by generative adversarial networks (GANs) and training approaches which do not require paired data. Recently, these techniques have been applied in the medical field for cross-domain image translation. This study investigated deep learning transformation in medical imaging. It was motivated to identify generalizable methods which would satisfy the simultaneous requirements of quality and anatomical accuracy across the entire human body. Specifically, whole-body MR patient data acquired on a PET/MR system were used to generate synthetic CT image volumes. The capacity of these synthetic CT data for use in PET attenuation correction (AC) was evaluated and compared to current MR-based attenuation correction (MR-AC) methods, which typically use multiphase Dixon sequences to segment various tissue types. This work aimed to investigate the technical performance of a GAN system for general MR-to-CT volumetric transformation and to evaluate the performance of the generated images for PET AC. A dataset comprising matched, same-day PET/MR and PET/CT patient scans was used for validation. A combination of training techniques was used to produce synthetic images which were of high-quality and anatomically accurate. Higher correlation was found between the values of mu maps calculated directly from CT data and those derived from the synthetic CT images than those from the default segmented Dixon approach. Over the entire body, the total amounts of reconstructed PET activities were similar between the two MR-AC methods, but the synthetic CT method yielded higher accuracy for quantifying the tracer uptake in specific regions. The findings reported here demonstrate the feasibility of this technique and its potential to improve certain aspects of attenuation correction for PET/MR systems. Moreover, this work may have larger implications for establishing generalized methods for inter-modality, whole-body transformation in medical imaging. Unsupervised deep learning techniques can produce high-quality synthetic images, but additional constraints may be needed to maintain medical integrity in the generated data.\n_____\n\n_____\n**[Deep learning-based synthetic CT generation from MR images: comparison of generative adversarial and residual neural networks](https://www.semanticscholar.org/paper/7422d5ed2101758fec38cd19014622a30280c524)**\n\n\n> **TL;DR:** We compared the performance of two deep learning models, a GAN and a ResNet, for the task of MR-based synthetic CT (sCT) generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR tumor pretreatment planning.\n\nBackground: Currently, MRI-only radiotherapy (RT) eliminates some of the concerns about using CT images in RT chains such as the registration of MR images to a separate CT, extra dose delivery, and the additional cost of repeated imaging. However, one remaining challenge is that the signal intensities of MRI are not related to the attenuation coefficient of the biological tissue. This work compares the performance of two state-of-the-art deep learning models; a generative adversarial network (GAN) and a residual network (ResNet) for synthetic CTs (sCT) generation from MR images. Materials and Methods: The brain MR and CT images of 86 participants were analyzed. GAN and ResNet models were implemented for the generation of synthetic CTs from the 3D T1-weighted MR images using a six-fold cross-validation scheme. The resulting sCTs were compared, considering the CT images as a reference using standard metrics such as the mean absolute error (MAE), peak signal-to-noise-ratio (PSNR) and the structural similarity index (SSIM). Results: Overall, the ResNet model exhibited higher accuracy in relation to the delineation of brain tissues. The ResNet model estimated the CT values for the entire head region with an MAE of 114.1±27.5 HU compared to MAE=-10.9±147.0 HU obtained from the GAN model. Moreover, both models offered comparable SSIM and PSNR values, although the ResNet method exhibited a slightly superior performance over the GAN method. Conclusion: We compared two state-of-the-art deep learning models for the task of MR-based sCT generation. The ResNet model exhibited superior results, thus demonstrating its potential to be used for the challenge of synthetic CT generation in PET/MR AC and MR-only RT planning. ► Original article\n_____\n\n_____\n**[Deep learning for molecular generation and optimization - a review of the state of the art](https://www.semanticscholar.org/paper/b2029e552d620600fd6f542ba0e11e39d42c3431)**\n\nWe review a recent groundswell of work which uses deep learning techniques to generate and optimize molecules.\n_____\n\n_____\n**[Deep Learning for Source Code Modeling and Generation](https://paperswithcode.com/paper/deep-learning-for-source-code-modeling-and)**\n\n\n> **TL;DR:** Deep Learning methods are showing great promise for natural language processing tasks such as modeling and generation of source code. We provide a comprehensive review of these methods, grouped by the type of task they address. We then formulate common program learning tasks under an encoder-decoder framework, and introduce recent Deep Learning mechanisms suitable to solve them. Finally, we discuss the current state-of-the-art and challenges for practitioners and researchers.\n\nDeep Learning (DL) techniques for Natural Language Processing have been evolving remarkably fast. Recently, the DL advances in language modeling, machine translation, and paragraph understanding are so prominent that the potential of DL in Software Engineering cannot be overlooked, especially in the field of program learning. To facilitate further research and applications of DL in this field, we provide a comprehensive review to categorize and investigate existing DL methods for source code modeling and generation. To address the limitations of the traditional source code models, we formulate common program learning tasks under an encoder-decoder framework. After that, we introduce recent DL mechanisms suitable to solve such problems. Then, we present the state-of-the-art practices and discuss their challenges with some recommendations for practitioners and researchers as well.\n_____\n\n\n_____\n**[Towards a new generation of parton densities with deep learning models](https://www.semanticscholar.org/paper/8c409498d4888d117f52fea76034774ae2f88ede)**\n\n\n> **TL;DR:** We use a deep learning inspired approach to regression for the determination of parton distribution functions. Our approach uses efficient computing frameworks based on graph generated models for PDF parametrization and gradient descent optimization. Results from our approach outperform the current state-of-the-art PDF fitting methodology in terms of model selection and computational resources usage.\n\nWe present a new regression model for the determination of parton distribution functions (PDF) using techniques inspired from deep learning projects. In the context of the NNPDF methodology, we implement a new efficient computing framework based on graph generated models for PDF parametrization and gradient descent optimization. The best model configuration is derived from a robust cross-validation mechanism through a hyperparametrization tune procedure. We show that results provided by this new framework outperforms the current state-of-the-art PDF fitting methodology in terms of best model selection and computational resources usage.\n_____\n\n_____\n**[ReAssert: Deep Learning for Assert Generation](https://www.semanticscholar.org/paper/24995e7933ed3e3d71f94894fb3e1df0cce681e3)**\n\n\n> **TL;DR:** Re-ASSERT is a method for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first.\n\nThe automated generation of test code can reduce the time and effort required to build software while increasing its correctness and robustness. In this paper, we present RE-ASSERT, an approach for the automated generation of JUnit test asserts which produces more accurate asserts than previous work with fewer constraints. This is achieved by targeting projects individually, using precise code-to-test traceability for learning and by generating assert statements from the method-under-test directly without the need to write an assert-less test first. We also utilise Reformer, a state-of-the-art deep learning model, along with two models from previous work to evaluate ReAssert and an existing approach, known as ATLAS, using lexical accuracy,uniqueness, and dynamic analysis. Our evaluation of ReAssert shows up to 44% of generated asserts for a single project match exactly with the ground truth, increasing to 51% for generated asserts that compile. We also improve on the ATLAS results through our use of Reformer with 28% of generated asserts matching exactly with the ground truth. Reformer also produces the greatest proportion of unique asserts (71%), giving further evidence that Reformer produces the most useful asserts.\n_____\n\n_____\n**[Generative Deep Learning Techniques for Password Generation](https://www.semanticscholar.org/paper/1699d5e8003d83c6f0bca203f5a42978e9723df7)**\n\nPassword guessing approaches via deep learning have recently been investigated with significant breakthroughs in their ability to generate novel, realistic password candidates. In the present work we study a broad collection of deep learning and probabilistic based models in the light of password guessing: attention-based deep neural networks, autoencoding mechanisms and generative adversarial networks. We provide novel generative deep-learning models in terms of variational autoencoders exhibiting state-of-art sampling performance, yielding additional latent-space features such as interpolations and targeted sampling. Lastly, we perform a thorough empirical analysis in a unified controlled framework over well-known datasets (RockYou, LinkedIn, Youku, Zomato, Pwnd). Our results not only identify the most promising schemes driven by deep neural networks, but also illustrate the strengths of each approach in terms of generation variability and sample uniqueness.\n_____\n\n\n_____\n**[Research on Image Generation and Style Transfer Algorithm Based on Deep Learning](https://doi.org/10.4236/ojapps.2019.98053)**\n\nat the current process of artistic creation and animation creation, there are a \nlot of repeated manual operations in the process of conversion from sketch to the stylized image. This paper presented a solution based on a deep \nlearning framework to realize image generation and style transfer. The method \nfirst used the conditional generation to resist the network, optimizes the loss \nfunction of the training mapping relationship, and generated the actual image \nfrom the input sketch. Then, by defining and optimizing the perceptual loss \nfunction of the style transfer model, the style features are extracted from the \nimage, thereby forming the actual The conversion between images and stylized \nart images. Experiments show that this method can greatly reduce the work of \ncoloring and converting with different artistic effects, and achieve the purpose of transforming simple stick \nfigures into actual object images.\n_____\n\n\n_____\n**[Applications of Deep Learning to Audio Generation](https://doi.org/10.1109/MCAS.2019.2945210)**\n\n\n> **TL;DR:** Deep learning algorithms are characterized by multiple hidden layers of processing units, which can be trained to automatically learn features of data. These deep learning architectures have demonstrated remarkable success in a number of domains, including speech recognition and acoustic signal processing. In this article, we provide an overview of the deep learning techniques and their effectiveness for speech/acoustic signal processing applications.\n\nIn the recent past years, deep learning based machine learning systems have demonstrated remarkable success for a wide range of learning tasks in multiple domains such as computer vision, speech recognition and other pattern recognition based applications. The purpose of this article is to contribute a timely review and introduction of state-of-the-art deep learning techniques and their effectiveness in speech/acoustic signal processing. Thorough investigations of various deep learning architectures are provided under the categories of discriminative and generative algorithms, including the up-to-date Generative Adversarial Networks (GANs) as an integrated model. A comprehensive overview of applications in audio generation is highlighted. Based on understandings from these approaches, we discuss how deep learning methods can benefit the field of speech/acoustic signal synthesis and the potential issues that need to be addressed for prospective real-world scenarios. We hope this survey provides a valuable reference for practitioners seeking to innovate in the usage of deep learning approaches for speech/acoustic signal generation.\n_____\n\n\n_____\n**[Music Generation Using Deep Learning](https://doi.org/10.1007/978-981-13-3393-4_21)**\n\nDeep learning has recently been used for many art-related activities such as automatic generation of music and pictures. This paper deals with music generation by using raw audio files in the frequency domain using Restricted Boltzmann Machine and Long Short- Term Memory architectures. The work does not use any information about musical structure to aid the learning, instead, it learns from a previous permutation of notes and generates an optimal and pleasant permutation. It also serves as a comparative study for music generation using Long Short-Term Memory and Restricted Boltzmann Machine.\n_____\n\n\n_____\n**[Deep Learning for Expressive Music Generation](https://doi.org/10.1145/3359852.3359898)**\n\n\n> **TL;DR:** We study the expressive potential of musical elements such as note velocity and duration in artificially generated music compositions. We find that these elements are important for musical expression and that including them in the generated pieces improves their quality.\n\nIn the last decade, Deep Learning (DL) algorithms have been increasing its popularity in several fields such as computer vision, speech recognition, natural language processing and many others. DL models, however, are not limited to scientific domains as they have recently been applied to content generation in diverse art forms - both in the generation of novel contents and as co-creative tools. Artificial music generation is one of the fields where DL architectures have been applied. They have been mostly used to create new compositions exhibiting promising results when compared to human compositions. Despite this, the majority of these artificial pieces lack some expression when compared to music compositions performed by humans. In this document, we propose a system capable of artificially generating expressive music compositions. Our main goal is to improve the quality of the musical compositions generated by the artificial system by exploring perceptually relevant musical elements such as note velocity and duration. To assess this hypothesis, we perform user tests. Results suggest that expressive elements such as duration and velocity are key aspects in a music composition expression, making the ones who include these preferable to non-expressive ones.\n_____\n\n\n_____\n**[Deep Learning Based Program Generation From Requirements Text: Are We There Yet?](https://doi.org/10.1109/tse.2020.3018481)**\n\n\n> **TL;DR:** We present a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements\n\nTo release developers from time-consuming software development, many approaches have been proposed to generate source code automatically according to software requirements. With significant advances in deep learning and natural language processing, deep learning-based approaches are proposed to generate source code from natural language descriptions. The key insight is that given a large corpus of software requirements and their corresponding implementations, advanced deep learning techniques may learn how to translate software requirements into source code that fulfill such requirements. Although such approaches are reported to be highly accurate, they are evaluated on datasets that are rather small, lack of diversity, and significantly different from real-world software requirements. To this end, we build a large scale dataset that is composed of longer requirements as well as validated implementations. We evaluate the state-of-the-art approaches on this new dataset, and the results suggest that their performance on our dataset is significantly lower than that on existing datasets concerning the common metrics, i.e., BLEU. Evaluation results also suggest that the generated programs often contain syntactic and semantical errors, and none of them can pass even a single predefined test case. Further analysis reveals that the state-of-the-art approaches learn little from software requirements, and most of the successfully generated statements are popular statements in the training programs. Based on this finding, we propose a popularity-based approach that always generates the most popular statements in training programs regardless of the input (software requirements). Evaluation results suggest that none of the state-of-the-art approaches can outperform this simple statistics-based approach. As a conclusion, deep learning-based program generation requires significant improvement in the future, and our dataset may serve as a basis for future research in this direction.\n_____\n\n\n\n_____\n**[A Review of Text Style Transfer using Deep Learning](https://www.semanticscholar.org/paper/9fecc1d7e0c5ff9c4789a45ff05509079b7c81c1)**\n\nStyle is an integral component of a sentence indicated by the choice of words a person makes. Different people have different ways of expressing themselves, however, they adjust their speaking and writing style to a social context, an audience, an interlocutor or the formality of an occasion. Text style transfer is defined as a task of adapting and/or changing the stylistic manner in which a sentence is written, while preserving the meaning of the original sentence. A systematic review of text style transfer methodologies using deep learning is presented in this paper. We point out the technological advances in deep neural networks that have been the driving force behind current successes in the fields of natural language understanding and generation. The review is structured around two key stages in the text style transfer process, namely, representation learning and sentence generation in a new style. The discussion highlights the commonalities and differences between proposed solutions as well as challenges and opportunities that are expected to direct and foster further research in the field. Impact Statement—Motivated by recent advancements in the field, we have carried out a systematic review of state-of-the-art research to highlight the trends, commonalities and differences across style transfer methodologies using deep learning. The discussion is organized around key stages of the process, namely, representation learning of style and content of a given sentence, and generation of the sentence in a new style. A comprehensive view of methodologies, available datasets and evaluation metrics is compiled to foster further research in the field.\n_____", "1779003": "satoshidatamoto these look interesting, thanks for sharing :)", "1779010": "satoshidatamoto Thank you for sharing" }, "source": "meta" }