Unsupervised Discovery of Disentangled Manifolds in GANs

• URL: http://arxiv.org/abs/2011.11842v2
• Date: Sun, 29 Nov 2020 18:56:50 GMT
• Title: Unsupervised Discovery of Disentangled Manifolds in GANs
• Authors: Yu-Ding Lu, Hsin-Ying Lee, Hung-Yu Tseng, Ming-Hsuan Yang
• Abstract summary: Interpretable generation process is beneficial to various image editing applications. We propose a framework to discover interpretable directions in the latent space given arbitrary pre-trained generative adversarial networks.
• Score: 74.24771216154105
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As recent generative models can generate photo-realistic images, people seek to understand the mechanism behind the generation process. Interpretable generation process is beneficial to various image editing applications. In this work, we propose a framework to discover interpretable directions in the latent space given arbitrary pre-trained generative adversarial networks. We propose to learn the transformation from prior one-hot vectors representing different attributes to the latent space used by pre-trained models. Furthermore, we apply a centroid loss function to improve consistency and smoothness while traversing through different directions. We demonstrate the efficacy of the proposed framework on a wide range of datasets. The discovered direction vectors are shown to be visually corresponding to various distinct attributes and thus enable attribute editing.

Related papers

• Varying Manifolds in Diffusion: From Time-varying Geometries to Visual Saliency [25.632973225129728]
  We study the geometric properties of the diffusion model, whose forward diffusion process and reverse generation process construct a series of distributions on a manifold. We show that the generation rate is highly correlated with intuitive visual properties, such as visual saliency, of the image component. We propose an efficient and differentiable scheme to estimate the generation rate for a given image component over time, giving rise to a generation curve.
  arXiv  Detail & Related papers  (2024-06-07T07:32:41Z)
• Spatial Steerability of GANs via Self-Supervision from Discriminator [123.27117057804732]
  We propose a self-supervised approach to improve the spatial steerability of GANs without searching for steerable directions in the latent space. Specifically, we design randomly sampled Gaussian heatmaps to be encoded into the intermediate layers of generative models as spatial inductive bias. During inference, users can interact with the spatial heatmaps in an intuitive manner, enabling them to edit the output image by adjusting the scene layout, moving, or removing objects.
  arXiv  Detail & Related papers  (2023-01-20T07:36:29Z)
• Latent Transformations via NeuralODEs for GAN-based Image Editing [25.272389610447856]
  We show that nonlinear latent code manipulations realized as flows of a trainable Neural ODE are beneficial for many practical non-face image domains. In particular, we investigate a large number of datasets with known attributes and demonstrate that certain attribute manipulations are challenging to obtain with linear shifts only.
  arXiv  Detail & Related papers  (2021-11-29T18:59:54Z)
• Finding an Unsupervised Image Segmenter in Each of Your Deep Generative Models [92.92095626286223]
  We develop an automatic procedure for finding directions that lead to foreground-background image separation. We use these directions to train an image segmentation model without human supervision.
  arXiv  Detail & Related papers  (2021-05-17T19:34:24Z)
• Diamond in the rough: Improving image realism by traversing the GAN latent space [0.0]
  We present an unsupervised method to find a direction in the latent space that aligns with improved photo-realism. Our approach leaves the network unchanged while enhancing the fidelity of the generated image. We use a simple generator inversion to find the direction in the latent space that results in the smallest change in the image space.
  arXiv  Detail & Related papers  (2021-04-12T14:45:29Z)
• Encoding Robustness to Image Style via Adversarial Feature Perturbations [72.81911076841408]
  We adapt adversarial training by directly perturbing feature statistics, rather than image pixels, to produce robust models. Our proposed method, Adversarial Batch Normalization (AdvBN), is a single network layer that generates worst-case feature perturbations during training.
  arXiv  Detail & Related papers  (2020-09-18T17:52:34Z)
• Unsupervised Controllable Generation with Self-Training [90.04287577605723]
  controllable generation with GANs remains a challenging research problem. We propose an unsupervised framework to learn a distribution of latent codes that control the generator through self-training. Our framework exhibits better disentanglement compared to other variants such as the variational autoencoder.
  arXiv  Detail & Related papers  (2020-07-17T21:50:35Z)
• Closed-Form Factorization of Latent Semantics in GANs [65.42778970898534]
  A rich set of interpretable dimensions has been shown to emerge in the latent space of the Generative Adversarial Networks (GANs) trained for synthesizing images. In this work, we examine the internal representation learned by GANs to reveal the underlying variation factors in an unsupervised manner. We propose a closed-form factorization algorithm for latent semantic discovery by directly decomposing the pre-trained weights.
  arXiv  Detail & Related papers  (2020-07-13T18:05:36Z)
• Network Bending: Expressive Manipulation of Deep Generative Models [0.2062593640149624]
  We introduce a new framework for manipulating and interacting with deep generative models that we call network bending. We show how it allows for the direct manipulation of semantically meaningful aspects of the generative process as well as allowing for a broad range of expressive outcomes.
  arXiv  Detail & Related papers  (2020-05-25T21:48:45Z)
• Controlling generative models with continuous factors of variations [1.7188280334580197]
  We introduce a new method to find meaningful directions in the latent space of any generative model. Our method does not require human annotations and is well suited for the search of directions encoding simple transformations of the generated image.
  arXiv  Detail & Related papers  (2020-01-28T10:04:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.