Technology Category
Computer Vision: Diffusion Models for VisionNatural Language Processing: Code Generation / Program Synthesis from Natural LanguageHumans and AI: Game Design — Procedural Content Generation & Storytelling
Quantitative and Qualitative Comparison of Generative Models for Subject-Specific Gaze Synthesis: Diffusion vs GAN
🤖 AI Summary
This work addresses the lack of individual specificity in existing generative gaze data methods by proposing a subject-specific personalized eye movement sequence synthesis framework. Methodologically, it integrates diffusion models with generative adversarial networks, incorporating a compact user embedding and a subject-aware generation module to accurately model individual oculomotor patterns; it also presents the first systematic comparative evaluation of these two generative paradigms for personalized gaze synthesis. Contributions include: (1) the first end-to-end personalized gaze generation architecture supporting user-conditional input; (2) a quantitative evaluation framework grounded in oculomotor signal quality; and (3) significant improvements over baselines in key metrics—including spatial accuracy and sequential fidelity—demonstrating superior realism and preservation of inter-individual variability.
Technology Category
Application Category
📝 Abstract
Recent advances in deep learning demonstrate the ability to generate synthetic gaze data. However, most approaches have primarily focused on generating data from random noise distributions or global, predefined latent embeddings, whereas individualized gaze sequence generation has been less explored. To address this gap, we revisit two recent approaches based on diffusion and generative adversarial networks (GANs) and introduce modifications that make both models explicitly subject-aware while improving accuracy and effectiveness. For the diffusion-based approach, we utilize compact user embeddings that emphasize per-subject traits. Moreover, for the GAN-based approach, we propose a subject-specific synthesis module that conditioned the generator to retain better idiosyncratic gaze information. Finally, we conduct a comprehensive assessment of these modified approaches utilizing standard eye-tracking signal quality metrics, including spatial accuracy and precision. This work helps define synthetic signal quality, realism, and subject specificity, thereby contributing to the potential development of gaze-based applications.
Problem
Research questions and friction points this paper is trying to address.
Generating subject-specific synthetic gaze sequences from data
Improving accuracy of diffusion and GAN models for gaze synthesis
Assessing synthetic gaze quality using spatial accuracy metrics
Innovation
Methods, ideas, or system contributions that make the work stand out.
Subject-aware diffusion models with user embeddings
GAN-based subject-specific synthesis module
Comprehensive assessment using eye-tracking metrics
🔎 Similar Papers
2024-09-02arXiv.orgCitations: 1
