Skip to content

Author

Zi-Hao Zheng

We have 9 of 16 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Nov 2026

When Should Agents Remember? Falsification-Gated Self-Evolution for LLM Agents

LLM agents increasingly operate in long-horizon, verifier-rich environments, but current self-improvement mechanisms often write episode-derived lessons directly into memory, making local success an unreliable signal for reusable and safe knowledge. This paper addresses the core question of when an experience-derived b...

Run-Xuan Tang, Haoyu Gao, Yu-Yan Ding et al. · 0 citations
#artificial intelligence Preprint Oct 2026

Auditing Escalation Reliability in Healthcare LLM Agents: A Perturbation-Controlled Benchmark of Workflow-State Failures

Trustworthy biomedical AI requires reliable action selection when evidence is missing, stale, conflicting, or late, not only accurate answers on static inputs. We instantiate an episode-level evaluation protocol as 400 synthetic healthcare workflow episodes: 40 hand-authored semantic seeds across four task classes, two...

Zi-Hao Zheng, Jun-Yi Yao, Bai-Chuan Li et al. · 2 citations
#artificial intelligence Preprint Aug 2026

Stale Does Not Mean Unsafe: Guard Precision for Tool-Using LLM Agents under Infrastructure State Races

This work evaluates three commit-time guard granularities (global epoch, read-set version, semantic commit predicate), multi-level verification, and model-side gates on three locally hosted quantized model families, and investigates how precisely runtime guards distinguish invalidating races.

Zi-Hao Zheng, Jia-Yu Long, Bai-Chuan Li et al. · 0 citations
Preprint Sep 2026

Engineering Reliable Commit Gates for Agentic AI: Cost-Aware Verification Portfolios under Common-Mode Data Failures

VP-CONTROL, a runtime-assurance design and deterministic benchmark for cost-aware commit gates, is presented, motivating explicit evidence lineage, cost-aware selection, and commit-time enforcement, while exposing the limits of approximate calibration and local-tool generalization.

Zi-Hao Zheng, Bai-Chuan Li, Jun-Yi Yao et al. · 1 citation
Preprint Aug 2026

Reliable Financial Named Entity Recognition Under Domain Shift: Confidence Estimation and Selective Prediction

confidence estimation and selective prediction for financial named entity recognition (NER) on a three-tier stress test spanning SEC filings, financial news, and general-topic social media as an extreme out-of-domain condition, evaluating a BERT tagger and LoRA-tuned Qwen2.

Zi-Hao Zheng, Bai-Chuan Li, Jun-Yi Yao et al. · 3 citations
#natural language process... Preprint Aug 2026

Toward Workflow-Aware Benchmarking for Healthcare NLP Agents

An episode-level evaluation protocol for healthcare NLP agents is introduced, supplying a reproducible intermediate evaluation layer between static benchmarks and prospective workflow studies, with an explicit cost-sensitive treatment of missed versus unnecessary escalation.

Jun-Yi Yao, Bai-Chuan Li, Zi-Hao Zheng et al. · 3 citations · ⚡1
Preprint Aug 2026

Remember, Verify, or Ask? Cross-Family Evaluation of Memory Commitment in LLM Agents

The memory-clarification boundary is studied: whether interaction-derived information should be persisted, used only in the current context, re-verified, or clarified with the user, as well as across Claude and Qwen.

Bai-Chuan Li, Jun-Yi Yao, Zi-Hao Zheng · 3 citations
#artificial intelligence Preprint Jun 2026

Beyond Helpfulness: A Teaching-over-Solving Diagnostic for Measuring Educational Impact in LLM Tutors

It is argued that public tutoring benchmarks can better support positive-impact evaluation by reporting solving-oriented and pedagogy-oriented scores separately and by making disclosure-sensitive, student-agency-preserving criteria more explicit.

Jun-Yi Yao, Zi-Hao Zheng, Baichuan Li · 3 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.