/really-awesome-gan

A list of papers on Generative Adversarial (Neural) Networks

really-awesome-gan

A list of papers and other resources on Generative Adversarial (Neural) Networks. This site is maintained by Holger Caesar. To complement or correct it, please contact me at holger-at-it-caesar.com or visit it-caesar.com. Also checkout really-awesome-semantic-segmentation and our COCO-Stuff dataset.

NOTE: Despite the enormous interest in this cite (~3000 visitors per month), I will no longer add new papers starting from November 2017. I feel that GANs have come from an exotic topic to the mainstream and an exhaustive list of all GAN papers is no more feasible or useful. However, I invite other people to continue this effort and reuse my list.

Contents

Recommendations

  • Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis [arXiv] Beyond face rotation
  • Pose Guided Person Image Generation [arXiv] Pose guided person
  • Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks [arXiv] Cycle GAN

Tutorials & Workshops & Blogs

  • Columbia Advanced Machine Learning Seminar
    • New Progress on GAN Theory and Practice [Blog]
    • Implicit Generative Models — What are you GAN-na do? [Blog]
  • How to Train a GAN? Tips and tricks to make GANs work [Blog]
  • NIPS 2016 Tutorial: Generative Adversarial Networks [arXiv]
  • NIPS 2016 Workshop on Adversarial Training [Web] [Blog]
  • On the intuition behind deep learning & GANs — towards a fundamental understanding [Blog]
  • OpenAI - Generative Models [Blog]
  • SimGANs - a game changer in unsupervised learning, self driving cars, and more [Blog]

Videos

  • Generative Adversarial Networks by Ian Goodfellow [Video]
  • Tutorial on Generative Adversarial Networks by Mark Chang [Video]

Code

  • Cleverhans: A library for benchmarking vulnerability to adversarial examples [Code] [Blog]
  • Generative Adversarial Networks (GANs) in 50 lines of code (PyTorch) [Blog] [Code]
  • Generative Models: Collection of generative models, e.g. GAN, VAE in Pytorch and Tensorflow [Code]

Papers

Overview

  • Generative Adversarial Networks: An Overview [arXiv]

Theory & Machine Learning

  • A Classification-Based Perspective on GAN Distributions [arXiv]
  • A Connection between Generative Adversarial Networks, Inverse Reinforcement Learning, and Energy-Based Models [arXiv]
  • A General Retraining Framework for Scalable Adversarial Classification [Paper]
  • Activation Maximization Generative Adversarial Nets [arXiv]
  • AdaGAN: Boosting Generative Models [arXiv]
  • Adversarial Autoencoders [arXiv]
  • Adversarial Discriminative Domain Adaptation [arXiv]
  • Adversarial Generator-Encoder Networks [arXiv]
  • Adversarial Feature Learning [arXiv] [Code]
  • Adversarially Learned Inference [arXiv] [Code]
  • AE-GAN: adversarial eliminating with GAN [arXiv]
  • An Adversarial Regularisation for Semi-Supervised Training of Structured Output Neural Networks [arXiv]
  • APE-GAN: Adversarial Perturbation Elimination with GAN [arXiv]
  • Associative Adversarial Networks [arXiv]
  • Autoencoding beyond pixels using a learned similarity metric [arXiv]
  • Bayesian Conditional Generative Adverserial Networks [arXiv]
  • Bayesian GAN [arXiv]
  • BEGAN: Boundary Equilibrium Generative Adversarial Networks [Paper] [arXiv] [Code]
  • Binary Generative Adversarial Networks for Image Retrieval [arXiv]
  • Boundary-Seeking Generative Adversarial Networks [arXiv] [Code]
  • CausalGAN: Learning Causal Implicit Generative Models with Adversarial Training [arXiv]
  • Class-Splitting Generative Adversarial Networks [arXiv]
  • Comparison of Maximum Likelihood and GAN-based training of Real NVPs [arXiv]
  • Conditional CycleGAN for Attribute Guided Face Image Generation [arXiv]
  • Conditional Generative Adversarial Nets [arXiv] [Code]
  • Connecting Generative Adversarial Networks and Actor-Critic Methods [Paper]
  • Continual Learning in Generative Adversarial Nets [arXiv]
  • C-RNN-GAN: Continuous recurrent neural networks with adversarial training [arXiv]
  • CM-GANs: Cross-modal Generative Adversarial Networks for Common Representation Learning [arXiv]
  • Cooperative Training of Descriptor and Generator Networks [arXiv]
  • Coupled Generative Adversarial Networks [arXiv] [Code]
  • Dualing GANs [arXiv]
  • Deep and Hierarchical Implicit Models [arXiv]
  • Energy-based Generative Adversarial Network [arXiv] [Code]
  • Explaining and Harnessing Adversarial Examples [arXiv]
  • Flow-GAN: Bridging implicit and prescribed learning in generative models [arXiv]
  • f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization [arXiv] [Code]
  • Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking [arXiv]
  • Generalization and Equilibrium in Generative Adversarial Nets (GANs) [arXiv]
  • Generating images with recurrent adversarial networks [arXiv]
  • Generative Adversarial Networks [arXiv] [Code] [Code]
  • Generative Adversarial Networks as Variational Training of Energy Based Models [arXiv]
  • Generative Adversarial Networks with Inverse Transformation Unit [arXiv]
  • Generative Adversarial Parallelization [arXiv] [Code]
  • Generative Adversarial Residual Pairwise Networks for One Shot Learning [arXiv]
  • Generative Adversarial Structured Networks [Paper]
  • Generative Cooperative Net for Image Generation and Data Augmentation [arXiv]
  • Generative Moment Matching Networks [arXiv] [Code]
  • Generative Semantic Manipulation with Contrasting GAN [arXiv]
  • Geometric GAN [arXiv]
  • Good Semi-supervised Learning that Requires a Bad GAN [arXiv]
  • Gradient descent GAN optimization is locally stable [arXiv]
  • How to Train Your DRAGAN [arXiv]
  • Image Quality Assessment Techniques Show Improved Training and Evaluation of Autoencoder Generative Adversarial Networks [arXiv]
  • Improved Semi-supervised Learning with GANs using Manifold Invariances [arXiv]
  • Improved Techniques for Training GANs [arXiv] [Code]
  • Improved Training of Wasserstein GANs [arXiv] [Code]
  • InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets [arXiv] [Code]
  • Inverting The Generator Of A Generative Adversarial Network [Paper]
  • It Takes (Only) Two: Adversarial Generator-Encoder Networks [arXiv]
  • KGAN: How to Break The Minimax Game in GAN [arXiv]
  • Learning in Implicit Generative Models [Paper]
  • Learning Loss for Knowledge Distillation with Conditional Adversarial Networks [arXiv]
  • Learning to Discover Cross-Domain Relations with Generative Adversarial Networks [arXiv] [Code]
  • Learning Texture Manifolds with the Periodic Spatial GAN [arXiv]
  • Least Squares Generative Adversarial Networks [arXiv] [Code]
  • Linking Generative Adversarial Learning and Binary Classification [arXiv]
  • Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities [arXiv]
  • LR-GAN: Layered Recursive Generative Adversarial Networks for Image Generation [arXiv]
  • MAGAN: Margin Adaptation for Generative Adversarial Networks [arXiv] [Code]
  • Maximum-Likelihood Augmented Discrete Generative Adversarial Networks [arXiv]
  • McGan: Mean and Covariance Feature Matching GAN [arXiv]
  • Message Passing Multi-Agent GANs [arXiv]
  • MMD GAN: Towards Deeper Understanding of Moment Matching Network [arXiv]
  • Mode Regularized Generative Adversarial Networks [arXiv] [Code]
  • Multi-Agent Diverse Generative Adversarial Networks [arXiv]
  • Multi-Generator Gernerative Adversarial Nets [arXiv]
  • Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models [arXiv]
  • On Convergence and Stability of GANs [arXiv]
  • On the effect of Batch Normalization and Weight Normalization in Generative Adversarial Networks [arXiv]
  • On the Quantitative Analysis of Decoder-Based Generative Models [arXiv]
  • Optimizing the Latent Space of Generative Networks [arXiv]
  • Parametrizing filters of a CNN with a GAN [arXiv]
  • PixelGAN Autoencoders [arXiv]
  • SegAN: Adversarial Network with Multi-scale L1 Loss for Medical Image Segmentation [arXiv]
  • SeqGAN: Sequence Generative Adversarial Nets with Policy Gradient [arXiv]
  • Simple Black-Box Adversarial Perturbations for Deep Networks [Paper]
  • Softmax GAN [arXiv]
  • Stabilizing Training of Generative Adversarial Networks through Regularization [arXiv]
  • Stacked Generative Adversarial Networks [arXiv]
  • Statistics of Deep Generated Images [arXiv]
  • Structured Generative Adversarial Networks [arXiv]
  • Tensorizing Generative Adversarial Nets [arXiv]
  • The Cramer Distance as a Solution to Biased Wasserstein Gradients [arXiv]
  • Towards Understanding Adversarial Learning for Joint Distribution Matching [arXiv]
  • Training generative neural networks via Maximum Mean Discrepancy optimization [arXiv]
  • Triple Generative Adversarial Nets [arXiv]
  • Unrolled Generative Adversarial Networks [arXiv]
  • Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [arXiv] [Code] [Code] [Code] [Code] [Code]
  • Wasserstein GAN [arXiv] [Code] [Code]

Applied Vision

  • 3D Object Reconstruction from a Single Depth View with Adversarial Learning [arXiv]
  • 3D Shape Induction from 2D Views of Multiple Objects [arXiv]
  • A step towards procedural terrain generation with GANs [arXiv] [Code]
  • Abnormal Event Detection in Videos using Generative Adversarial Nets [arXiv]
  • Adversarial Generation of Training Examples for Vehicle License Plate Recognition [arXiv]
  • Adversarial nets with perceptual losses for text-to-image synthesis [arXiv]
  • Adversarial Networks for Spatial Context-Aware Spectral Image Reconstruction from RGB [arXiv]
  • Adversarial Networks for the Detection of Aggressive Prostate Cancer [arXiv]
  • Adversarial PoseNet: A Structure-aware Convolutional Network for Human Pose Estimation [arXiv]
  • Adversarial Training For Sketch Retrieval [arXiv]
  • Aesthetic-Driven Image Enhancement by Adversarial Learning [arXiv]
  • Age Progression / Regression by Conditional Adversarial Autoencoder [arXiv]
  • AlignGAN: Learning to Align Cross-Domain Images with Conditional Generative Adversarial Networks [arXiv]
  • Amortised MAP Inference for Image Super-resolution [arXiv]
  • A Novel Approach to Artistic Textual Visualization via GAN [arXiv]
  • Anti-Makeup: Learning A Bi-Level Adversarial Network for Makeup-Invariant Face Verification [arXiv]
  • ARIGAN: Synthetic Arabidopsis Plants using Generative Adversarial Network [arXiv]
  • ArtGAN: Artwork Synthesis with Conditional Categorial GANs [arXiv]
  • Artificial Generation of Big Data for Improving Image Classification: A Generative Adversarial Network Approach on SAR Data [arXiv]
  • Auto-Encoder Guided GAN for Chinese Calligraphy Synthesis [arXiv]
  • Auto-painter: Cartoon Image Generation from Sketch by Using Conditional Generative Adversarial Networks [arXiv]
  • Automatic Liver Segmentation Using an Adversarial Image-to-Image Network [arXiv]
  • Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis [arXiv]
  • CAN: Creative Adversarial Networks Generating “Art” by Learning About Styles and Deviating from Style Norms [arXiv]
  • Compressed Sensing MRI Reconstruction with Cyclic Loss in Generative Adversarial Networks [arXiv]
  • Conditional Adversarial Network for Semantic Segmentation of Brain Tumor [arXiv]
  • Conditional generative adversarial nets for convolutional face generation [Paper]
  • Conditional Image Synthesis with Auxiliary Classifier GANs [Paper] [arXiv] [Code]
  • Contextual RNN-GANs for Abstract Reasoning Diagram Generation [arXiv]
  • Controllable Generative Adversarial Network [arXiv]
  • Creatism: A deep-learning photographer capable of creating professional work [arXiv]
  • Crossing Nets: Combining GANs and VAEs with a Shared Latent Space for Hand Pose Estimation [arXiv]
  • CVAE-GAN: Fine-Grained Image Generation through Asymmetric Training [arXiv]
  • Data Augmentation in Classification using GAN [arXiv]
  • Deep Generative Adversarial Compression Artifact Removal [arXiv]
  • Deep Generative Adversarial Networks for Compressed Sensing (GANCS) Automates MRI [arXiv]
  • Deep Generative Adversarial Neural Networks for Realistic Prostate Lesion MRI Synthesis [arXiv]
  • Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks [arXiv] [Code] [Blog]
  • Deep multi-scale video prediction beyond mean square error [arXiv] [Code]
  • Deep Unsupervised Representation Learning for Remote Sensing Images [arXiv]
  • DeLiGAN : Generative Adversarial Networks for Diverse and Limited Data [arXiv]
  • Depth Structure Preserving Scene Image Generation [arXiv]
  • DualGAN: Unsupervised Dual Learning for Image-to-Image Translation [arXiv] [Code]
  • Dual Motion GAN for Future-Flow Embedded Video Prediction [arXiv]
  • ExprGAN: Facial Expression Editing with Controllable Expression Intensity [arXiv]
  • Face Aging With Conditional Generative Adversarial Networks [arXiv]
  • Face Transfer with Generative Adversarial Network [arXiv]
  • Filmy Cloud Removal on Satellite Imagery with Multispectral Conditional Generative Adversarial Nets [arXiv]
  • Freehand Ultrasound Image Simulation with Spatially-Conditioned Generative Adversarial Networks [arXiv]
  • From source to target and back: symmetric bi-directional adaptive GAN [arXiv]
  • Full Resolution Image Compression with Recurrent Neural Networks [arXiv]
  • GANs for Biological Image Synthesis [arXiv]
  • GeneGAN: Learning Object Transfiguration and Attribute Subspace from Unpaired Data [arXiv] [Code]
  • Generate Identity-Preserving Faces by Generative Adversarial Networks [arXiv]
  • Generate To Adapt: Aligning Domains using Generative Adversarial Networks [arXiv]
  • Generative Adversarial Models for People Attribute Recognition in Surveillance [arXiv]
  • Generative Adversarial Network based on Resnet for Conditional Image Restoration [arxiv]
  • Generative Adversarial Network-based Synthesis of Visible Faces from Polarimetric Thermal Faces [arXiv]
  • Generative Adversarial Networks for Multimodal Representation Learning in Video Hyperlinking [arXiv]
  • Generative Adversarial Text to Image Synthesis [arXiv] [Code]
  • Generative Visual Manipulation on the Natural Image Manifold [Project] [Youtube] [Paper] [Code]
  • GP-GAN: Gender Preserving GAN for Synthesizing Faces from Landmarks [arXiv]
  • GP-GAN: Towards Realistic High-Resolution Image Blending [arXiv]
  • Guiding InfoGAN with Semi-Supervision [arXiv]
  • How to Fool Radiologists with Generative Adversarial Networks? A Visual Turing Test for Lung Cancer Diagnosis [arXiv]
  • Hierarchical Detail Enhancing Mesh-Based Shape Generation with 3D Generative Adversarial Network [arXiv]
  • High-Quality Face Image SR Using Conditional Generative Adversarial Networks [arXiv]
  • High-Quality Facial Photo-Sketch Synthesis Using Multi-Adversarial Networks [arXiv]
  • Image De-raining Using a Conditional Generative Adversarial Network [arXiv]
  • Image Generation and Editing with Variational Info Generative Adversarial Networks [arXiv]
  • Image-to-Image Translation with Conditional Adversarial Networks [arXiv] [Code]
  • Improved Adversarial Systems for 3D Object Generation and Reconstruction [arXiv] [Code]
  • Improving Heterogeneous Face Recognition with Conditional Adversarial Networks [arXiv]
  • Improving image generative models with human interactions [arXiv]
  • Imitating Driver Behavior with Generative Adversarial Networks [arXiv]
  • Interactive 3D Modeling with a Generative Adversarial Network [arXiv]
  • Intraoperative Organ Motion Models with an Ensemble of Conditional Generative Adversarial Networks [arXiv]
  • Invertible Conditional GANs for image editing [arXiv] [Paper]
  • Label Denoising Adversarial Network (LDAN) for Inverse Lighting of Face Images [arXiv]
  • Learning a Driving Simulator [arXiv]
  • Learning a Generative Adversarial Network for High Resolution Artwork Synthesis [arXiv]
  • Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling [arXiv]
  • Learning from Simulated and Unsupervised Images through Adversarial Training [arXiv]
  • Learning to Discover Cross-Domain Relations with Generative Adversarial Networks [arXiv]
  • Learning to Generate Chairs with Generative Adversarial Nets [arXiv]
  • Learning to Generate Time-Lapse Videos Using Multi-Stage Dynamic Generative Adversarial Networks [arXiv]
  • Low Dose CT Image Denoising Using a Generative Adversarial Network with Wasserstein Distance and Perceptual Loss [arXiv]
  • MARTA GANs: Unsupervised Representation Learning for Remote Sensing Image Classification [arXiv]
  • Megapixel Size Image Creation using Generative Adversarial Networks [arXiv]
  • Microscopy Cell Segmentation via Adversarial Neural Networks [arXiv]
  • MoCoGAN: Decomposing Motion and Content for Video Generation [arXiv]
  • Multi-view Generative Adversarial Networks [Paper]
  • Neural Photo Editing with Introspective Adversarial Networks [Paper] [arXiv]
  • Neural Stain-Style Transfer Learning using GAN for Histopathological Images [arXiv]
  • Outline Colorization through Tandem Adversarial Networks [arXiv]
  • Perceptual Adversarial Networks for Image-to-Image Transformation [arXiv]
  • Perceptual Generative Adversarial Networks for Small Object Detection [arXiv]
  • Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network [arXiv]
  • Pose Guided Person Image Generation [arXiv]
  • Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks [arXiv]
  • Recurrent Topic-Transition GAN for Visual Paragraph Generation [arXiv]
  • RenderGAN: Generating Realistic Labeled Data [arXiv]
  • Representation Learning and Adversarial Generation of 3D Point Clouds [arXiv]
  • Retinal Vasculature Segmentation Using Local Saliency Maps and Generative Adversarial Networks For Image Super Resolution [arXiv]
  • Retinal Vessel Segmentation in Fundoscopic Images with Generative Adversarial Networks [arXiv]
  • SAD-GAN: Synthetic Autonomous Driving using Generative Adversarial Networks [arXiv]
  • SalGAN: Visual Saliency Prediction with Generative Adversarial Networks [arXiv]
  • SegAN: Adversarial Network with Multi-scale L1 Loss for Medical Image Segmentation [arXiv]
  • SeGAN: Segmenting and Generating the Invisible [arXiv]
  • Semantic Image Inpainting with Deep Generative Models [arXiv]
  • Semantic Image Synthesis via Adversarial Learning [arXiv]
  • Semantic Segmentation using Adversarial Networks [arXiv]
  • Semantically Decomposing the Latent Spaces of Generative Adversarial Networks [arXiv]
  • Semi-Latent GAN: Learning to generate and modify facial images from attributes [arXiv]
  • Semi-Supervised Learning with Context-Conditional Generative Adversarial Networks [arXiv]
  • Sharpness-aware Low dose CT denoising using conditional generative adversarial network [arXiv]
  • Simultaneously Color-Depth Super-Resolution with Conditional Generative Adversarial Network [arXiv]
  • Socially-compliant Navigation through Raw Depth Inputs with Generative Adversarial Imitation Learning [arXiv]
  • StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networks [arXiv]
  • StackGAN++: Realistic Image Synthesis with Stacked Generative Adversarial Networks [arXiv]
  • Style Transfer for Sketches with Enhanced Residual U-net and Auxiliary Classifier GAN [arXiv]
  • Supervised Adversarial Networks for Image Saliency Detection [arXiv]
  • Synthesis of Positron Emission Tomography (PET) Images via Multi-channel Generative Adversarial Networks (GANs) [arXiv]
  • Synthesizing Filamentary Structured Images with GANs [arXiv]
  • Synthetic Iris Presentation Attack using iDCGAN [arXiv]
  • Synthetic Medical Images from Dual Generative Adversarial Networks [arXiv]
  • TAC-GAN - Text Conditioned Auxiliary Classifier Generative Adversarial Network [arXiv]
  • Temporal Generative Adversarial Nets with Singular Value Clipping [arXiv]
  • TextureGAN: Controlling Deep Image Synthesis with Texture Patches [arXiv]
  • Texture Synthesis with Spatial Generative Adversarial Networks [arXiv] [Code]
  • The Conditional Analogy GAN: Swapping Fashion Articles on People Images [arXiv]
  • Towards Adversarial Retinal Image Synthesis [arXiv] [Code] [Demo]
  • Towards Diverse and Natural Image Descriptions via a Conditional GAN [arXiv]
  • Towards the Automatic Anime Characters Creation with Generative Adversarial Networks [arXiv]
  • Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro [arXiv]
  • Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks [arXiv]
  • Unsupervised and Semi-supervised Learning with Categorical Generative Adversarial Networks [arXiv]
  • Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery [arXiv]
  • Unsupervised Cross-Domain Image Generation [arXiv]
  • Unsupervised Diverse Colorization via Generative Adversarial Networks [arXiv]
  • Unsupervised Pixel–Level Domain Adaptation with Generative Adversarial Networks [arXiv]
  • Unsupervised Visual Attribute Transfer with Reconfigurable Generative Adversarial Networks [arXiv]
  • VIGAN: Missing View Imputation with Generative Adversarial Networks [arXiv]
  • WaterGAN: Unsupervised Generative Network to Enable Real-time Color Correction of Monocular Underwater Images [arXiv]
  • Weakly Supervised Generative Adversarial Networks for 3D Reconstruction [arXiv]

Applied Other

  • Adversarial Generation of Natural Language [arXiv]
  • Adversarial Ranking for Language Generation [arXiv]
  • Adversarial Training Methods for Semi-Supervised Text Classification [arXiv] [Paper]
  • A Generative Model for Volume Rendering [[arXiv]](A Generative Model for Volume Rendering)
  • ChemGAN challenge for drug discovery: can AI reproduce natural chemical diversity? [arXiv]
  • Generating Adversarial Malware Examples for Black-Box Attacks Based on GAN [arXiv]
  • Generating Multi-label Discrete Electronic Health Records using Generative Adversarial Networks [arXiv]
  • Language Generation with Recurrent Generative Adversarial Networks without Pre-training [arXiv]
  • Learning to Protect Communications with Adversarial Neural Cryptography [arXiv] [Blog]
  • Long Text Generation via Adversarial Training with Leaked Information [arXiv]
  • MidiNet: A Convolutional Generative Adversarial Network for Symbolic-domain Music Generation using 1D and 2D Conditions [arXiv]
  • MuseGAN: Symbolic-domain Music Generation and Accompaniment with Multi-track Sequential Generative Adversarial Networks [arXiv]
  • Reconstruction of three-dimensional porous media using generative adversarial neural networks [arXiv] [Code]
  • SEGAN: Speech Enhancement Generative Adversarial Network [arXiv]
  • Semi-supervised Learning of Compact Document Representations with Deep Networks [Paper]
  • SSGAN: Secure Steganography Based on Generative Adversarial Networks [arXiv]
  • Steganographic Generative Adversarial Networks [arXiv]
  • Towards Grounding Conceptual Spaces in Neural Representations [arXiv]

Humor

  • Stopping GAN Violence: Generative Unadversarial Networks [arXiv]