Trustworthy AI for data-driven risk intelligence and complex real-world systems
About · Research · People · Projects · Maintenance · Publications · News · Connect
The Financial Intelligence & Data Mining Lab (FinD Lab) is a research group at the Institute of Computing Technology, Chinese Academy of Sciences. We are affiliated with the State Key Laboratory of AI Safety and the Key Laboratory of Intelligent Information Processing.
Our mission is to build trustworthy, safe, and deployable artificial intelligence for complex financial and data-intensive systems. We study how AI can understand, model, and safeguard real-world systems through financial risk intelligence, robust graph learning, large language models, user behavior modeling, and data mining.
We value problem-driven research, rigorous evaluation, reproducible systems, and research outcomes that can create lasting academic and practical impact.
| 🛡️ Trustworthy AI & Financial Risk Risk assessment, fraud and anomaly detection, algorithmic safety, robust decision-making, and reliable AI for high-stakes financial scenarios. |
🕸️ Graph Learning & Robustness Robust graph neural networks, graph security, adversarial attacks and defenses, out-of-distribution generalization, and graph model privacy. |
| 📈 Quantitative Intelligence Formulaic alpha discovery, reinforcement learning, symbolic regression, stock trend modeling, and LLM-assisted quantitative research. |
🤖 LLMs, Agents & Personalized Intelligence LLMs for AI research, agentic simulation, personalized generation, user behavior modeling, recommendation, and automated research workflows. |
| Researcher | Role | Research Focus | Links |
|---|---|---|---|
| Xiang Ao (敖翔) | Professor & Ph.D. Supervisor | Intelligent finance, trustworthy and safe AI, data mining, and natural language processing | Homepage · ICT Profile |
| Yang Liu (柳阳) | Research Faculty | Trustworthy graph machine learning, graph security, and financial data mining | Homepage · ICT Profile |
Our work is driven by faculty, postdoctoral researchers, graduate students, research assistants, interns, and open-source contributors. Visit the organization’s People page to discover public members and contributors.
| Project | What it provides | Research / Resources |
|---|---|---|
AlphaGen |
Reinforcement-learning-based generation of synergistic formulaic alpha collections, with extensions for LLM-assisted alpha discovery. | KDD 2023 Paper · FCS 2026 Paper · Code |
| GrAP³ | A data-free model extraction attack framework for graph models under graph pre-training and prompt learning. | CIKM 2026 · Research Page · Code |
| GRASP | Differentially private graph reconstruction defense with structured perturbation. | KDD 2025 Paper · Code |
| SPEAR | A structure-preserving manipulation method for graph backdoor attacks. | WWW 2025 · Research Page · Code |
Awesome Robust Graph Learning |
A maintained collection of papers, benchmarks, datasets, tutorials, conferences, and toolkits for robust graph learning. | Resources · MIT License |
Awesome LLMs for AI Research |
The companion resource hub for AI4AIR, covering LLM-assisted data engineering, model design, evaluation, and closed-loop research automation. | Project Page · Resources |
| SCAPE | Stylistic-content-aware personalized headline generation with panoramic user interests. | WWW 2025 · Code |
Explore all public repositories in the FinD Lab organization.
To improve documentation and reproducibility without modifying author-led repositories, the following personal forks are maintained separately. The FinD Lab repositories above remain the canonical research sources.
| Canonical Project | Community Maintenance Fork | Initial Maintenance Scope |
|---|---|---|
| AlphaGen | Yangmingchi0/alphagen · Maintenance Guide | Reproduction audit, environment/problem index, and verification protocol |
| GrAP³ | Yangmingchi0/GrAP3 | Professional documentation, responsible-use notes, setup/argument guide, and reproducibility checklist |
| Awesome Robust Graph Learning | Yangmingchi0/awesome-robust-graph-learning · Curation Policy | Evidence standards, resource submission format, and link-review cadence |
These forks are community-maintenance workspaces, not replacements for the official projects. Algorithmic changes and upstream contributions require separate review and permission from the original maintainers.
| Year | Research Output | Venue | Links |
|---|---|---|---|
| 2026 | GrAP³: A Data-free Model Extraction Attack Framework under Graph Pre-training and Prompt Learning Paradigm | CIKM | Research Page · Code |
| 2026 | Graph-Agnostic Linear Transformers | Neural Networks | Paper |
| 2026 | A Hybrid Approach to Formulaic Alpha Discovery with Large Language Model Assistance | Frontiers of Computer Science | Paper · Code |
| 2025 | GRASP: Differentially Private Graph Reconstruction Defense with Structured Perturbation | KDD | Paper · Code |
| 2025 | SPEAR: A Structure-Preserving Manipulation Method for Graph Backdoor Attacks | WWW | Research Page · Code |
| 2025 | Panoramic Interests: Stylistic-Content Aware Personalized Headline Generation | WWW | Code |
| 2023 | Generating Synergistic Formulaic Alpha Collections via Reinforcement Learning | KDD | Paper · Code |
For a broader publication list, visit Xiang Ao’s homepage and Yang Liu’s homepage.
- 2026.08 — GrAP³ was accepted by CIKM 2026.
- 2026.08 — LiveSim: Simulating Environment-Shaped Users in Multi-Agent Live-Stream Ecosystems was accepted by EMNLP 2026; two additional works on user representation and multi-persona agentic simulation were accepted by ICDM 2026.
- 2026.05 — Outsmarting the Chameleon: Counterfactual Decoupling for Tactical OOD Shifts in Live Streaming Risk Assessment was accepted by KDD 2026.
- 2026.04 — Two works on retrieval-augmented live-streaming risk assessment and delayed conversion-rate prediction were accepted by SIGIR 2026.
- 2026.04 — The FinD Lab WeChat official account was launched to share research highlights, paper explanations, and open-source updates.
See the latest research updates on the group homepage.
We are progressively standardizing active repositories so that each major research project provides:
- A concise overview and authoritative paper link
- Environment and dependency specifications
- A quick-start example and reproducible experiment commands
- Dataset and checkpoint instructions where redistribution is permitted
- Citation metadata, license information, and release notes
- Issue and contribution guidelines for community feedback
We welcome reproducibility reports, documentation improvements, missing-resource suggestions, and responsible pull requests. Please use each project’s issue tracker for repository-specific questions.
- Research and news: Xiang Ao’s homepage
- Open-source projects: FinD Lab on GitHub
- Public members: Organization People
- WeChat: FinD Lab official account launch article
- Research opportunities: Please consult the faculty homepages and current institutional announcements for up-to-date openings.
- State Key Laboratory of AI Safety
- Key Laboratory of Intelligent Information Processing
- Institute of Computing Technology, Chinese Academy of Sciences