Research map / Image generation / Autoregressive and latent generation
Designed to Spread: A Generative Approach to Enhance Information Diffusion
Ziqing Qian, Jiaying Lei, Shengqi Dang, Nan Cao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add semantic similarity C to the reward (reward sqrt(dL*C) vs influence gain only) (ablation) | RL information editor, text · Olympics text editing | Consistency: 0.9471 -> 0.9510 | Improved |
| add semantic similarity C to the reward (reward sqrt(dL*C) vs influence gain only) (ablation) | RL information editor, text · Olympics text editing | Diffusion Gain: 21.98% -> 20.69% | Hurt |
| add semantic similarity C to the reward (reward sqrt(dL*C) vs influence gain only) (ablation) | RL information editor, image · Olympics image editing | Consistency: 0.8707 -> 0.8730 | Improved |
| add semantic similarity C to the reward (reward sqrt(dL*C) vs influence gain only) (ablation) | RL information editor, image · Olympics image editing | Diffusion Gain: 12.43% -> 11.20% | Hurt |
| learn an RL policy over editing actions instead of greedy search over sampled actions | text editing · Olympics text editing | Diffusion Gain: 17.93% -> 20.69% | Improved |
| learn an RL policy over editing actions instead of greedy search over sampled actions | image editing · Olympics image editing | Diffusion Gain: 7.93% -> 11.20% | Improved |
| use audience-aware RL editing instead of direct LLM rewriting | text editing · Olympics text editing | Diffusion Gain: 14.03% -> 20.69% | Improved |
| add in-context exemplars of high and low influence messages (IC-L) to LLM rewriting | text editing · Olympics text editing | Diffusion Gain: 14.03% -> 13.43% | No clear effect |
| use audience-aware RL editing instead of direct LLM rewriting | image editing · Olympics image editing | Diffusion Gain: 1.36% -> 11.20% | Improved |
| match message with its real audience instead of permuted audience in the influence indicator | influence score L_U(c) · Olympics and Movie | mean influence score: 0.53 -> 0.76 (Olympics); 0.45 -> 0.79 (Movie) | Improved |
Closest papers by meaning
- Self-Play Fine-tuning of Diffusion Models for Text-to-image Generation (NeurIPS 2024 · similarity 0.84)
- Efficient Fine-Tuning and Concept Suppression for Pruned Diffusion Models (CVPR 2025 · similarity 0.83)
- Enhancing Multi-Scale Diffusion Prediction via Sequential Hypergraphs and Adversarial Learning (AAAI 2024 · similarity 0.83)
- Expressive and Flexible Simulation of Information Spread Strategies in Social Networks Using Planning (AAAI 2024 · similarity 0.83)
- Diffusion PID: Interpreting Diffusion via Partial Information Decomposition (NeurIPS 2024 · similarity 0.83)
- Detect-and-Guide: Self-regulation of Diffusion Models for Safe Text-to-Image Generation via Guideline Token Optimization (CVPR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
