Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FedStyle: Style-Based Federated Learning Crowdsourcing Framework for Art Commissions

Changjuan Ran, Yeting Guo, Fang Liu, Shenglan Cui, Yunfan Ye

TL;DR

FedStyle addresses privacy-aware retrieval of artistic styles by enabling style-based federated learning among artists who keep their artworks locally. It learns per-artist style representations and aligns them with a server using a contrastive objective to mitigate non-IID drift, then aggregates via a small public dataset. Empirical results on Artiststyle and Conllustration show FedStyle achieves higher accuracy and F1-scores than baselines and is perceived as privacy-preserving and practically useful by artists and buyers. The approach offers a scalable, privacy-preserving pathway for style-guided artist discovery in crowdsourcing platforms.

Abstract

The unique artistic style is crucial to artists' occupational competitiveness, yet prevailing Art Commission Platforms rarely support style-based retrieval. Meanwhile, the fast-growing generative AI techniques aggravate artists' concerns about releasing personal artworks to public platforms. To achieve artistic style-based retrieval without exposing personal artworks, we propose FedStyle, a style-based federated learning crowdsourcing framework. It allows artists to train local style models and share model parameters rather than artworks for collaboration. However, most artists possess a unique artistic style, resulting in severe model drift among them. FedStyle addresses such extreme data heterogeneity by having artists learn their abstract style representations and align with the server, rather than merely aggregating model parameters lacking semantics. Besides, we introduce contrastive learning to meticulously construct the style representation space, pulling artworks with similar styles closer and keeping different ones apart in the embedding space. Extensive experiments on the proposed datasets demonstrate the superiority of FedStyle.

FedStyle: Style-Based Federated Learning Crowdsourcing Framework for Art Commissions

TL;DR

FedStyle addresses privacy-aware retrieval of artistic styles by enabling style-based federated learning among artists who keep their artworks locally. It learns per-artist style representations and aligns them with a server using a contrastive objective to mitigate non-IID drift, then aggregates via a small public dataset. Empirical results on Artiststyle and Conllustration show FedStyle achieves higher accuracy and F1-scores than baselines and is perceived as privacy-preserving and practically useful by artists and buyers. The approach offers a scalable, privacy-preserving pathway for style-guided artist discovery in crowdsourcing platforms.

Abstract

The unique artistic style is crucial to artists' occupational competitiveness, yet prevailing Art Commission Platforms rarely support style-based retrieval. Meanwhile, the fast-growing generative AI techniques aggravate artists' concerns about releasing personal artworks to public platforms. To achieve artistic style-based retrieval without exposing personal artworks, we propose FedStyle, a style-based federated learning crowdsourcing framework. It allows artists to train local style models and share model parameters rather than artworks for collaboration. However, most artists possess a unique artistic style, resulting in severe model drift among them. FedStyle addresses such extreme data heterogeneity by having artists learn their abstract style representations and align with the server, rather than merely aggregating model parameters lacking semantics. Besides, we introduce contrastive learning to meticulously construct the style representation space, pulling artworks with similar styles closer and keeping different ones apart in the embedding space. Extensive experiments on the proposed datasets demonstrate the superiority of FedStyle.
Paper Structure (12 sections, 6 equations, 4 figures, 2 tables, 2 algorithms)

This paper contains 12 sections, 6 equations, 4 figures, 2 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Motivation
  4. Methodology
  5. System Framework
  6. Local Style Training
  7. Global Style Aggregation
  8. Experiment and Evaluation
  9. Experiment Setup
  10. Quantitative Evaluation
  11. Qualitative Assessment
  12. Conclusion

Figures (4)

  • Figure 1: Accuracy under different proportion of cloaked images and different heterogeneity settings.
  • Figure 2: System framework overiew.
  • Figure 3: Comparison of accuracy and F1-score with baselines
  • Figure 4: Ablation study results