Author

Data Party THU

Official platform of Tsinghua Big Data Research Center, sharing the team's latest research, teaching updates, and big data news.

507

Articles

Likes

2.4k

Views

Comments

Latest from Data Party THU

100 recent articles max

Data Party THU

Jul 21, 2026 · Artificial Intelligence

Task Decomposition with Multi‑Agent Systems: Boosting Complex AI Workflows

This article reviews a Berkeley PhD thesis that argues powerful foundation models still need task decomposition, detailing six contributions—including LLM‑grounded diffusion, video diffusion, self‑correcting loops, detailed local description, adaptive parallel reasoning, and ThreadWeaver—to organize computation across multiple agents for more controllable, reliable AI systems.

AI systemsLLMmulti-agent systems

0 likes · 16 min read

Task Decomposition with Multi‑Agent Systems: Boosting Complex AI Workflows

Data Party THU

Jul 20, 2026 · Artificial Intelligence

Unleashing Large Language Models for Graph Continual Learning: The UNIT Framework

The paper introduces UNIT, a three‑step framework that leverages a single‑task‑tuned LLM as a stable semantic encoder, combines uncertainty‑aware semantic anchors with explicit structural anchors, and achieves state‑of‑the‑art performance on multiple text‑attributed graph continual‑learning benchmarks, even in few‑shot settings.

Few-Shot LearningGraph Continual LearningSemantic Anchors

0 likes · 15 min read

Unleashing Large Language Models for Graph Continual Learning: The UNIT Framework

Data Party THU

Jul 20, 2026 · Artificial Intelligence

What I Learned After Six Months Building Production AI Agents: The Five Costly Mistakes

The article analyzes why AI agents that shine in demos often fail in production, identifies five common mistakes—including over‑reliance on prompts, manual evaluation, unchecked cost and latency, fragile tool integrations, and missing safety guards—and introduces a five‑layer Harness Engineering framework with a practical four‑week rollout plan to make agents reliable at scale.

AI agentCost ManagementEvaluation

0 likes · 23 min read

What I Learned After Six Months Building Production AI Agents: The Five Costly Mistakes

Data Party THU

Jul 19, 2026 · Artificial Intelligence

Biomni Integrates 105 Tools and 59 Databases to Enable AI‑Driven End‑to‑End Life‑Science Discovery

Biomni is a general biomedical AI agent that unifies 105 bioinformatics software packages and 59 curated databases, dynamically selects resources, uses code as a universal action language, and plans experiments, achieving 57% average accuracy on a 443‑question benchmark and dramatically speeding up expert‑level analyses.

AIAutomationBiomedical

0 likes · 8 min read

Biomni Integrates 105 Tools and 59 Databases to Enable AI‑Driven End‑to‑End Life‑Science Discovery

Data Party THU

Jul 18, 2026 · Artificial Intelligence

Smart Cellular Bricks: 3D Neural Cellular Automata for Life‑Like Modular Robots

The study introduces Smart Cellular Bricks, a modular robot system that uses 3D Neural Cellular Automata to enable identical cubes to exchange minimal local information, achieve 98.97% shape‑classification accuracy, 94.8% damage‑detection precision, and self‑repair within 60 update cycles, demonstrating scalable, life‑like collective intelligence.

Distributed SystemsSakana AImodular robotics

0 likes · 7 min read

Smart Cellular Bricks: 3D Neural Cellular Automata for Life‑Like Modular Robots

Data Party THU

Jul 18, 2026 · Industry Insights

Tsinghua’s Wang Jianmin Calls for a Global Open‑Source AI Ecosystem at UN Side Event

At the 2026 AI for Good Global Summit in Geneva, Professor Wang Jianmin of Tsinghua University urged worldwide collaboration to build an inclusive, affordable, and unified open‑source AI ecosystem, highlighting China’s massive developer base, top‑ranked contributions, emerging trends, and new governance tools such as OpenDigger.

AI GovernanceApache IoTDBChina Open Source

0 likes · 4 min read

Tsinghua’s Wang Jianmin Calls for a Global Open‑Source AI Ecosystem at UN Side Event

Data Party THU

Jul 17, 2026 · Industry Insights

Autonomous Robotic and Microrobotic Surgery: Progress and Roadmaps in Five Surgical Areas

The article reviews the state of autonomous robotic‑assisted and microrobotic surgery across vascular, lumen, laparoscopic, ophthalmic, and orthopedic fields, outlining recent advances, technical challenges such as energy supply, onboard computing, and safety, and proposing mixed‑control pathways toward fully autonomous procedures.

autonomous systemsendoscopymicrorobotics

0 likes · 9 min read

Autonomous Robotic and Microrobotic Surgery: Progress and Roadmaps in Five Surgical Areas

Data Party THU

Jul 17, 2026 · Industry Insights

Fei-Fei Li Predicts Only Two Types of Workers Will Remain in Ten Years

In a recent interview, Fei‑Fei Li argues that AI will not make intelligence cheap, warns against mistaking capability for quality, and forecasts that a decade from now the workforce will split into elite experts and self‑sufficient tool builders, while the middle tier fades away.

AI impactFei-Fei LiFuture of work

0 likes · 10 min read

Fei-Fei Li Predicts Only Two Types of Workers Will Remain in Ten Years

Data Party THU

Jul 16, 2026 · Artificial Intelligence

Can Overthinking in Large Language Reasoning Models Trigger DoS Attacks? A New Risk Unveiled

Researchers from Zhejiang University and Alibaba Security reveal that large language reasoning models can be forced into excessive, self‑correcting reasoning—'overthinking'—by crafted inputs, dramatically inflating token output and computation cost, enabling a novel black‑box DoS attack demonstrated via a Hierarchical Genetic Algorithm.

DoS attackblack-box attackhierarchical genetic algorithm

0 likes · 10 min read

Can Overthinking in Large Language Reasoning Models Trigger DoS Attacks? A New Risk Unveiled

Data Party THU

Jul 15, 2026 · Artificial Intelligence

How Autoregressive Boltzmann Generators Are Redefining Molecular Sampling

The paper introduces Autoregressive Boltzmann Generators (ArBG) that replace flow‑based models with autoregressive modeling, enabling exact likelihood computation, efficient importance‑sampling correction, and scalable transfer learning, and demonstrates superior performance on peptide benchmarks compared with prior Boltzmann generators.

Autoregressive ModelBoltzmann GeneratorImportance Sampling

0 likes · 15 min read

How Autoregressive Boltzmann Generators Are Redefining Molecular Sampling