paper-with-me

Papers

Decoupling Global and Local Representations via Invertible Generative Flows

2020-04-12 · ICLR 2021 1 · Xuezhe Ma, Xiang Kong, Shanghang Zhang, Eduard Hovy

In this work, we propose a new generative model that is capable of automatically decoupling global and local representations of images in an entirely unsupervised setting, by embedding a generative flow in the VAE framework to model the decoder. Specifically, the proposed model utilizes the variational auto-encoding framework to learn a (low-dimensional) vector of latent variables to capture the global information of an image, which is fed as a conditional input to a flow-based invertible decoder with architecture borrowed from style transfer literature. Experimental results on standard image benchmarks demonstrate the effectiveness of our model in terms of density estimation, image generation and unsupervised representation learning. Importantly, this work demonstrates that with only architectural inductive biases, a generative model with a likelihood-based objective is capable of learning decoupled representations, requiring no explicit supervision. The code for our model is available at https://github.com/XuezheMax/wolf.

📄 PDF Abstract BibTeX arXiv:2004.11820

Code (1)

XuezheMax/wolf 공식 구현 pytorch

Tasks

DecoderDensity EstimationImage GenerationRepresentation LearningStyle Transfer

Methods 이 논문이 사용한 방법론

USD Coin Customer Service Number +1-833-534-1729 설명 없음

Similar Papers 제목 키워드 기반

Decoupling Local and Global Representations of Time Series

2022-02-04 · Sana Tonekaboni, Chun-Liang Li, Sercan Arik, Anna Goldenberg 외

Real-world time series data are often generated from several sources of variation. Learning representations that capture the factors contributing to this variability enables a better understanding of the data via its und…

counterfactualTime SeriesTime Series Analysis

Faithful Extreme Rescaling via Generative Prior Reciprocated Invertible Representations

2022-01-01 · CVPR 2022 1 · Zhixuan Zhong, Liangyu Chai, Yang Zhou, Bailin Deng 외

This paper presents a Generative prior ReciprocAted Invertible rescaling Network (GRAIN) for generating faithful high-resolution (HR) images from low-resolution (LR) invertible images with an extreme upscaling factor…

Image Rescaling

GLOWin: A Flow-based Invertible Generative Framework for Learning Disentangled Feature Representations in Medical Images

2021-03-19 · Aadhithya Sankar, Matthias Keicher, Rami Eisawy, Abhijeet Parida 외

Disentangled representations can be useful in many downstream tasks, help to make deep learning models more interpretable, and allow for control over features of synthetically generated images that can be useful in train…

Disentanglement

MintNet: Building Invertible Neural Networks with Masked Convolutions

2019-07-18 · NeurIPS 2019 12 · Yang Song, Chenlin Meng, Stefano Ermon

We propose a new way of constructing invertible neural networks by combining simple building blocks with a novel set of composition rules. This leads to a rich set of invertible architectures, including those similar to …

Image Generation

Understanding and Mitigating Exploding Inverses in Invertible Neural Networks

2020-06-16 · Jens Behrmann, Paul Vicol, Kuan-Chieh Wang, Roger Grosse 외

Invertible neural networks (INNs) have been used to design generative models, implement memory-saving gradient computation, and solve inverse problems. In this work, we show that commonly-used INN architectures suffer fr…