Imaginative Walks: Generative Random Walk Deviation Loss for Improved Unseen Learning Representation
Not on the map. Why?
Abstract (arXiv)
We propose a novel loss for generative models, dubbed as GRaWD (Generative Random Walk Deviation), to improve learning representations of unexplored visual spaces. Quality learning representation of unseen classes (or styles) is critical to facilitate novel image generation and better generative understanding of unseen visual classes, i.e., zero-shot learning (ZSL). By generating representations of unseen classes based on their semantic descriptions, e.g., attributes or text, generative ZSL attempts to differentiate unseen from seen categories. The proposed GRaWD loss is defined by constructing a dynamic graph that includes the seen class/style centers and generated samples in the current minibatch. Our loss initiates a random walk probability from each center through visual generations produced from hallucinated unseen classes. As a deviation signal, we encourage the random walk to eventually land after t steps in a feature representation that is difficult to classify as any of the seen classes. We demonstrate that the proposed loss can improve unseen class representation quality inductively on text-based ZSL benchmarks on CUB and NABirds datasets and attribute-based ZSL benchmarks on AWA2, SUN, and aPY datasets. In addition, we investigate the ability of the proposed loss to generate meaningful novel visual art on the WikiArt dataset. The results of experiments and human evaluations demonstrate that the proposed GRaWD loss can improve StyleGAN1 and StyleGAN2 generation quality and create novel art that is significantly more preferable. Our code is made publicly available at https://github.com/Vision-CAIR/GRaWD.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 51 of its 79 references that are in the map. First posted on arXiv 20 Apr 2021.
Builds on
51 of 79 references.
References · 51
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,510
- 2019Analyzing and Improving the Image Quality of StyleGANcited by 3,580
- 2017Progressive Growing of GANs for Improved Quality, Stability, and Variationcited by 2,967
- 2016Perceptual Losses for Real-Time Style Transfer and Super-Resolutioncited by 2,678
- 2017Improved Training of Wasserstein GANscited by 2,223
- 2014Conditional Generative Adversarial Netscited by 2,152
- 2015Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networkscited by 2,048
- 2017Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networkscited by 707
- 2016StackGAN: Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Networkscited by 575
- 2016Conditional Image Synthesis With Auxiliary Classifier GANscited by 562
38 more
- 2017Zero-Shot Learning -- A Comprehensive Evaluation of the Good, the Bad and the Uglycited by 536
- 2018Meta-Learning for Semi-Supervised Few-Shot Classificationcited by 456
- 2016A Learned Representation For Artistic Stylecited by 340
- 2017A Neural Representation of Sketch Drawingscited by 311
- 2017Feature Generating Networks for Zero-Shot Learningcited by 257
- 2013Zero-Shot Learning Through Cross-Modal Transfercited by 237
- 2018Generalized Zero- and Few-Shot Learning via Aligned Variational Autoencoderscited by 206
- 2014Evaluation of Output Embeddings for Fine-Grained Image Classificationcited by 166
- 2015Label-Embedding for Image Classificationcited by 164
- 2016Learning a Deep Embedding Model for Zero-Shot Learningcited by 153
- 2016An Empirical Study and Analysis of Generalized Zero-Shot Learning for Object Recognition in the Wildcited by 146
- 2013Zero-Shot Learning by Convex Combination of Semantic Embeddingscited by 135
- 2017Semantic Autoencoder for Zero-Shot Learningcited by 133
- 2016Latent Embeddings for Zero-shot Classificationcited by 129
- 2017Generalized Zero-Shot Learning via Synthesized Examplescited by 114
- 2016Synthesized Classifiers for Zero-Shot Learningcited by 113
- 2016Learning What and Where to Drawcited by 105
- 2017CAN: Creative Adversarial Networks, Generating "Art" by Learning About Styles and Deviating from Style Normscited by 96
- 2017A Generative Adversarial Approach for Zero-Shot Learning from Noisy Textscited by 96
- 2021Contrastive Embedding for Generalized Zero-Shot Learningcited by 94
- 2018Multi-modal Cycle-consistent Generalized Zero-Shot Learningcited by 87
- 2019Learning to Self-Train for Semi-Supervised Few-Shot Classificationcited by 83
- 2015Inside-Outside Net: Detecting Objects in Context with Skip Pooling and Recurrent Neural Networkscited by 80
- 2015Zero-Shot Learning via Semantic Similarity Embeddingcited by 74
- 2015Predicting Deep Zero-Shot Convolutional Neural Networks using Textual Descriptionscited by 64
- 2020Leveraging Seen and Unseen Semantic Relationships for Generative Zero-Shot Learningcited by 64
- 2017Deep Learning Techniques for Music Generation -- A Surveycited by 49
- 2017Link the head to the "beak": Zero Shot Learning from Noisy Text Description at Part Precisioncited by 46
- 2016Less is more: zero-shot learning from online textual documents with noise suppressioncited by 40
- 2018Large-Scale Visual Relationship Understandingcited by 33
- 2019Creativity Inspired Zero-Shot Learningcited by 27
- 2021CreativeGAN: Editing Generative Adversarial Networks for Creative Design Synthesiscited by 24
- 2017From Zero-shot Learning to Conventional Supervised Classification: Unseen Visual Data Synthesiscited by 21
- 2018DeSIGN: Design Inspiration from Generative Networkscited by 20
- 2017Learning by Association - A versatile semi-supervised training method for neural networkscited by 19
- 2018Can Computers Create Art?cited by 18
- 2019Trick or TReAT: Thematic Reinforcement for Artistic Typographycited by 11
- 2015Write a Classifier: Predicting Visual Classifiers from Unstructured Textcited by 9
50 most cited of 51 All 51 (JSON)
Built on by
Citing papers per year: 2022: 2.
2 papers.