Implicit Signal Mining for Personalized Decision Systems
Chika Komari
A unified track across structural, semantic, and psychological signal mining
I focus on implicit signal mining in complex information networks for risk-aware and personalized decision systems.
Roadmap Snapshot
View Full RoadmapPsychological Signal Mining
Semantic Signal Mining
Structural Signal Mining
Featured Artifacts
Browse All ArtifactsAlignment Between the Minds (ICML 2026 Submission)
研究具身体在多轮指令交互中对用户潜在偏好的持续推断、维护与利用，聚焦高维心理信号挖掘。
Identifying Critical Nodes with Deep Learning and Reinforcement Learning
研究城市路网场景下的关键节点识别与影响力传播建模，属于低维结构信号挖掘方向的核心成果。
AIRI LocalGUI Task Agent (Open Source; In Progress)
把自然语言指令转换为可执行桌面任务计划，支持自动校验、自动修复与多工具协同执行。
Zanao: LLM-driven Market Archive & Analytics
面向真实市场信息流的端到端系统，覆盖数据采集、语义检索、多轮问答与可视化分析。
Latest Updates
Open Blog讨论：从《DeepSeek-R1 通过强化学习激励LLM中的推理》开始
**关键观点**：LLM 的推理能力可以通过纯强化学习 (RL) 来激励，从而无需人工标记的推理轨迹 <img src="/assets/images/LLM学习/deepseek-r1/DeepSeek-R1训练范式.png" alt="描述文字" width=…
资源整理
LLM与数据科学领域 专用任务LLM(如命名实体识别、向量化等) | 名称 | 功能描述 | 链接 | 拓展部分 | | :--------------: | :--------------
强化学习(reinforcement learning)
强化学习的数学基础 基本概念 强化学习的基础概名词 - **状态(State)**：表示当下环境，是做决策的唯一依据 - **动作(Action)**：做出的决策，动作种数一般用 表示 - **环境(Environment)**：使智能体不断变换状态的位
- Jan 2026: submitted first-author ICML 2026 paper on latent preference discovery
- 2025: accepted at GBCESC 2025 and received Best Paper Award
- Since Apr 2025: building financial knowledge-graph infrastructure at Evolution Asset Management