arXiv subjects

Jiang Guo

Publications and source records attributed to Jiang Guo.

At least 19 recordsLinked to original sources

arXiv2026

RayAR: Autoregressive Dominant Path Generation for Low-Overhead MIMO Beam Alignment

Beam alignment in multiple-input multiple-output (MIMO) systems requires selecting suitable transmit and receive beam pairs with limited measurement overhead. Although 3GPP Release 18 investigates beam prediction from partial measurements, collecting these measurements still incurs probing overhead. Geometry can focus this probing through the departure and arrival directions of dominant paths, whose identification requires accounting for coupled diffraction interactions. To address this challenge, we propose RayAR, an autoregressive method that generates dominant propagation paths from urban geometry. RayAR conditions candidate ranking on the path prefix, local diffraction geometry, and receiver reachability, updating this context after each selection until the receiver becomes visible. The completed path's departure and arrival directions rank beam pairs for probing, allowing the pair with the highest received power to be selected from a small candidate set. Under 3GPP Release 18 beam-management metrics, RayAR reduces probing by 96.88\% relative to exhaustive sweeping and achieves the highest optimal-pair hit rate among compared methods at the same budget. Real-world measurements further demonstrate the accuracy of this geometric guidance, with simulation-trained RayAR achieving the lowest departure and arrival angular errors among compared methods without fine-tuning and a mean arrival-direction error of 9.07$^\circ$.

eess.SP↗

arXiv2026

EvoBelief: Falsifiable World Models for Self-Improving LLM Agents

Self-evolving LLM agents typically improve how they act through accumulated experience, reflection, or skills. Yet even procedurally capable agents can fail when their beliefs about environment dynamics are incomplete, incorrect, or outdated. We study a complementary axis of self-evolution: evolving those beliefs during deployment. Such beliefs can support multiple tasks, but incorrect ones can also propagate across them. We introduce EvoBelief, a self-evolving world model that represents environment knowledge as falsifiable hypotheses. Unexpected outcomes and decision-relevant unknowns trigger hypothesis formation. Later observations and controlled experiments provide independent validation before a rule can guide predictions or plans within its supported scope. Counterexamples narrow or retire rules as the environment changes. The language model remains frozen, and each rule remains linked to its evidence and the decisions it changes. Across four benchmarks spanning static dynamics, hidden drift, and scientific discovery, EvoBelief improves task performance and reuses learned rules across tasks. On ALFWorld without actor-side transition memory, it improves success over the matched Frozen World Model by 23.9% and 31.3% with Claude Haiku 4.5 and GPT-5.4, respectively. On MiniGrid, it revises stale rules and recovers after unannounced dynamics changes. In DiscoveryWorld's Reactor Lab, it discovers and validates quantitative laws, completing 83.3% of Normal-tier and 58.3% of Challenge-tier instances, versus 0% for each evaluated baseline. These results demonstrate self-improvement through testable, revisable environment knowledge rather than model-parameter updates.

cs.AI↗

arXiv2026

Plan-and-Patch: Diffusion Language Models for Agentic Planning

Planning is increasingly important for long-horizon agents, where successful execution requires coordinating subgoals, tool use, and intermediate outcomes over many steps. Yet assumptions made during planning may be invalidated by the environment, tools may return unexpected results, or actions may fail. Effective agents must therefore not only generate plans, but also revise them. Such revisions often affect only part of a plan, leaving the preceding and subsequent structure intact. Rather than regenerate the entire plan and risk unnecessary changes, repair can regenerate the affected region conditioned on the preserved prefix and suffix. We introduce Plan-and-Patch, a plan-and-act framework in which a diffusion language model (dLLM) generates a structured, program-like plan through parallel unmasking and repairs it by filling in selected regions while keeping the surrounding steps fixed. We compare DreamReasoner-8B and Qwen3-8B as diffusion and autoregressive (AR) planners. On Natural Plan without task-specific training, diffusion (53.7%) achieves nearly twice the plan repair success rate of AR (27.0%). After task-specific training on agentic benchmarks, ALFWorld and TextCraft, the planners achieve similar observed success in plan generation, while diffusion reduces mean plan-generation latency by 39-46% relative to AR. Our results show that Plan-and-Patch provides a framework for faster plan generation and effective plan repair in long-horizon agents.

cs.AI↗

arXiv2026

Deny Without Disabling: Authorization-Paired Evaluation and Control for Multi-Agent Systems

Multi-agent systems derive their capabilities from sharing evidence, delegating tasks, and combining information across agents. The same process creates a safety problem: contributions that are admissible in isolation can jointly enable a prohibited use. Blocking every sensitive action avoids disclosure but defeats the purpose of collaboration. We introduce authorization-paired evaluation, which makes blocking prohibited uses and completing required authorized uses a joint success criterion, and FlowReview, a framework connecting object resolution, permission ranking, and deterministic enforcement. In controlled composition experiments, reviewing combined artifacts reduces the denied-commit rate from 86.0% to zero with no loss of authorized supply. Our findings show that preserving information and lineage alone does not ensure correct permission attribution. Object identity and permission must remain connected to execution through components whose outputs can be verified. Together, these findings establish a system-level requirement for multi-agent safety: govern composed information flows while preserving the authorized capabilities that make collaboration useful.

cs.MA↗

arXiv2026

Poison Once, Exploit Forever: Environment-Injected Memory Poisoning Attacks on Web Agents

Memory makes LLM-based web agents personalized, powerful, yet exploitable. By storing past interactions to personalize future tasks, agents inadvertently create a persistent attack surface that spans websites and sessions. While existing security research on memory assumes attackers can directly inject into memory storage or exploit shared memory across users, we present a more realistic threat model: contamination through environmental observation alone. We introduce Environment-injected Trajectory-based Agent Memory Poisoning (eTAMP), the first attack to achieve cross-session, cross-site compromise without requiring direct memory access. A single contaminated observation (e.g., viewing a manipulated product page) silently poisons an agent's memory and activates during future tasks on different websites, bypassing permission-based defenses. Our experiments on (Visual)WebArena reveal two key findings. First, eTAMP achieves substantial attack success rates: up to 32.5% on GPT-5-mini, 23.4% on GPT-5.2, and 19.5% on GPT-OSS-120B. Second, we discover Frustration Exploitation: agents under environmental stress become dramatically more susceptible, with ASR increasing up to 8 times when agents struggle with dropped clicks or garbled text. Notably, more capable models are not more secure. GPT-5.2 shows substantial vulnerability despite superior task performance. With the rise of AI browsers like OpenClaw, ChatGPT Atlas, and Perplexity Comet, our findings underscore the urgent need for defenses against environment-injected memory poisoning.

cs.CR↗

arXiv2026

Chimera: Latency- and Performance-Aware Multi-agent Serving for Heterogeneous LLMs

Multi-agent applications often execute complex tasks as multi-stage workflows, where each stage is an LLM call whose output becomes part of context for subsequent steps. Existing LLM serving systems largely assume homogeneous clusters with identical model replicas. This design overlooks the potential of heterogeneous deployments, where models of different sizes and capabilities enable finer trade-offs between latency and performance. However, heterogeneity introduces new challenges in scheduling across models with diverse throughput and performance. We present Chimera, a predictive scheduling system for multi-agent workflow serving on heterogeneous LLM clusters that jointly improves end-to-end latency and task performance. Chimera applies semantic routing to estimate per-model confidence scores for each request, predicts the total remaining output length of the workflow, and estimates per-model congestion using in-flight predicted token volumes for load balancing. We evaluate Chimera on representative agentic workflows for code generation and math reasoning using multiple heterogeneous LLM configurations. Across comparable settings, Chimera traces the best latency-performance frontier, reducing end-to-end latency by 1.2--2.4$\times$ and improving task performance by 8.0-9.5 percentage points on average over competitive baselines including vLLM.

cs.LG↗

arXiv2026

R-WoM: Retrieval-augmented World Model For Computer-use Agents

Large Language Models (LLMs) can serve as world models to enhance agent decision-making in digital environments by simulating future states and predicting action outcomes, potentially eliminating costly trial-and-error exploration. However, this capability is fundamentally limited by LLMs' tendency toward hallucination and their reliance on static training knowledge, which can lead to compounding errors that inhibit long-horizon simulations. To systematically investigate whether LLMs are appropriate for world modeling, we probe two core capabilities of world models--future state prediction and reward estimation--through three tasks: next-state identification, full-procedure planning alignment, and milestone transition recognition. Our analysis shows that while LLMs effectively capture immediate next states and identify meaningful state transitions, their performance rapidly degrades in full-procedure planning. This highlights LLMs' limitations in reliably modeling environment dynamics over long horizons. To address these limitations, we propose the Retrieval-augmented World Model (R-WoM), which grounds LLM simulations by incorporating factual, up-to-date knowledge retrieved from external tutorials. Experiments show that R-WoM achieves relative improvements of up to 23.4% and 16.3% on the subsets of OSWorld and Webarena compared to baselines, with particular advantage in longer-horizon simulations.

cs.CL↗

arXiv2025

Inverse Design of Metamaterials with Manufacturing-Guiding Spectrum-to-Structure Conditional Diffusion Model

Metamaterials are artificially engineered structures that manipulate electromagnetic waves, having optical properties absent in natural materials. Recently, machine learning for the inverse design of metamaterials has drawn attention. However, the highly nonlinear relationship between the metamaterial structures and optical behaviour, coupled with fabrication difficulties, poses challenges for using machine learning to design and manufacture complex metamaterials. Herein, we propose a general framework that implements customised spectrum-to-shape and size parameters to address one-to-many metamaterial inverse design problems using conditional diffusion models. Our method exhibits superior spectral prediction accuracy, generates a diverse range of patterns compared to other typical generative models, and offers valuable prior knowledge for manufacturing through the subsequent analysis of the diverse generated results, thereby facilitating the experimental fabrication of metamaterial designs. We demonstrate the efficacy of the proposed method by successfully designing and fabricating a free-form metamaterial with a tailored selective emission spectrum for thermal camouflage applications.

physics.optics↗

arXiv2025

CRYSIM: Prediction of Symmetric Structures of Large Crystals with GPU-based Ising Machines

Solving black-box optimization problems with Ising machines is increasingly common in materials science. However, their application to crystal structure prediction (CSP) is still ineffective due to symmetry agnostic encoding of atomic coordinates. We introduce CRYSIM, an algorithm that encodes the space group, the Wyckoff positions combination, and coordinates of independent atomic sites as separate variables. This encoding reduces the search space substantially by exploiting the symmetry in space groups. When CRYSIM is interfaced to Fixstars Amplify, a GPU-based Ising machine, its prediction performance was competitive with CALYPSO and Bayesian optimization for crystals containing more than 150 atoms in a unit cell. Although it is not realistic to interface CRYSIM to current small-scale quantum devices, it has the potential to become the standard CSP algorithm in the coming quantum age.

cond-mat.mtrl-sci↗

arXiv2025

On Synthetic Data Strategies for Domain-Specific Generative Retrieval

This paper investigates synthetic data generation strategies in developing generative retrieval models for domain-specific corpora, thereby addressing the scalability challenges inherent in manually annotating in-domain queries. We study the data strategies for a two-stage training framework: in the first stage, which focuses on learning to decode document identifiers from queries, we investigate LLM-generated queries across multiple granularity (e.g. chunks, sentences) and domain-relevant search constraints that can better capture nuanced relevancy signals. In the second stage, which aims to refine document ranking through preference learning, we explore the strategies for mining hard negatives based on the initial model's predictions. Experiments on public datasets over diverse domains demonstrate the effectiveness of our synthetic data generation and hard negative sampling approach.

cs.CL↗

arXiv2025

Human-AI collaboration for modeling heat conduction in nanostructures

In recent years, materials informatics, which combines data science and artificial intelligence (AI), has garnered significant attention owing to its ability to accelerate material development, reduce costs, and enhance product design. However, despite the widespread use of AI, human involvement is often limited to the initiation and oversight of machine learning processes and rarely includes more substantial roles that capitalize on human intuition or domain expertise. Consequently, true human-AI collaborations, where integrated insights can be maximized, are scarce. This study considers the problem of heat conduction in a two-dimensional nanostructure as a case study. An integrated human-AI collaboration framework is designed and used to construct a model to predict the thermal conductivity. This approach is used to determine the parameters that govern phonon transmission over the full range of frequencies and incidence angles. During operation, the self-learning entropic population annealing technique, which combines entropic sampling with a surrogate machine learning model, generates a small dataset that can be interpreted by a human. Therefore, data-efficient and global modeling is achieved, and parameters with physical interpretations are developed, which can guide nanostructural design to produce materials with specific properties. The proposed framework can leverage the complementary strengths of humans and AI, thereby enhancing the understanding and control of materials.

physics.comp-ph↗

arXiv2024

You Only Read Once (YORO): Learning to Internalize Database Knowledge for Text-to-SQL

While significant progress has been made on the text-to-SQL task, recent solutions repeatedly encode the same database schema for every question, resulting in unnecessary high inference cost and often overlooking crucial database knowledge. To address these issues, we propose You Only Read Once (YORO), a novel paradigm that directly internalizes database knowledge into the parametric knowledge of a text-to-SQL model during training and eliminates the need for schema encoding during inference. YORO significantly reduces the input token length by 66%-98%. Despite its shorter inputs, our empirical results demonstrate YORO's competitive performances with traditional systems on three benchmarks as well as its significant outperformance on large databases. Furthermore, YORO excels in handling questions with challenging value retrievals such as abbreviation.

cs.CL↗

arXiv2024

Boosting the quality factor of Tamm structures to millions by quantum inspired classical annealer with factorization machine

The Tamm structures show high quality (Q) factor property with simple photonic Bragg reflectors composed of alternative high/low refractive index layers and metal reflectors. However, the Q-factor of Tamm structures is inherently limited by the periodic constraints and fixed thicknesses of the Bragg reflector pairs. Removing periodic constraints and adopting aperiodic designs can lead to significantly higher Q-factors. Nevertheless, to fully exploit the potential of Tamm structures, it is essential to thoroughly explore the extensive search space introduced by these aperiodic designs. Herein, we introduce a novel approach that utilizes the quantum-inspired classical annealer combined with a factorization machine to facilitate the design of photonic Bragg reflectors with the aim of achieving ultrahigh Q-factor properties. Our investigation begins by establishing the effectiveness of global optimization in a 15-bit problem setting, showcasing a remarkable efficiency improvement of nearly one order of magnitude compared to random search strategies. Subsequently, we expand our analysis to a 20-bit problem, demonstrating comparable efficiency to state-of-the-art Bayesian optimization methods. Moreover, the proposed method proves its capability in handling extremely large search spaces, exemplified by a 40-bit problem, where we successfully uncover a Tamm structure design exhibiting an extraordinary ultrahigh Q-factor surpassing millions. The designed aperiodic Tamm structure exhibits exceptional high localization and enhancement of electric field, and the power loss is concentrated in the low loss dielectric layer. Hence, the decay rate of power dissipation significantly decreases and light-matter interaction time increases, resulting in ultrahigh Q-factor.

cond-mat.mtrl-sci↗

arXiv2024

Towards a Holistic Evaluation of LLMs on Factual Knowledge Recall

Large language models (LLMs) have shown remarkable performance on a variety of NLP tasks, and are being rapidly adopted in a wide range of use cases. It is therefore of vital importance to holistically evaluate the factuality of their generated outputs, as hallucinations remain a challenging issue. In this work, we focus on assessing LLMs' ability to recall factual knowledge learned from pretraining, and the factors that affect this ability. To that end, we construct FACT-BENCH, a representative benchmark covering 20 domains, 134 property types, 3 answer types, and different knowledge popularity levels. We benchmark 31 models from 10 model families and provide a holistic assessment of their strengths and weaknesses. We observe that instruction-tuning hurts knowledge recall, as pretraining-only models consistently outperform their instruction-tuned counterparts, and positive effects of model scaling, as larger models outperform smaller ones for all model families. However, the best performance from GPT-4 still represents a large gap with the upper-bound. We additionally study the role of in-context exemplars using counterfactual demonstrations, which lead to significant degradation of factual knowledge recall for large models. By further decoupling model known and unknown knowledge, we find the degradation is attributed to exemplars that contradict a model's known knowledge, as well as the number of such exemplars. Lastly, we fine-tune LLaMA-7B in different settings of known and unknown knowledge. In particular, fine-tuning on a model's known knowledge is beneficial, and consistently outperforms fine-tuning on unknown and mixed knowledge. We will make our benchmark publicly available.

cs.CL↗

arXiv2024

Propagation and Pitfalls: Reasoning-based Assessment of Knowledge Editing through Counterfactual Tasks

Current approaches of knowledge editing struggle to effectively propagate updates to interconnected facts. In this work, we delve into the barriers that hinder the appropriate propagation of updated knowledge within these models for accurate reasoning. To support our analysis, we introduce a novel reasoning-based benchmark -- ReCoE (Reasoning-based Counterfactual Editing dataset) -- which covers six common reasoning schemes in real world. We conduct a thorough analysis of existing knowledge editing techniques, including input augmentation, finetuning, and locate-and-edit. We found that all model editing methods show notably low performance on this dataset, especially in certain reasoning schemes. Our analysis over the chain-of-thought generation of edited models further uncover key reasons behind the inadequacy of existing knowledge editing methods from a reasoning standpoint, involving aspects on fact-wise editing, fact recall ability, and coherence in generation. We will make our benchmark publicly available.

cs.CL↗

arXiv2023

UNITE: A Unified Benchmark for Text-to-SQL Evaluation

A practical text-to-SQL system should generalize well on a wide variety of natural language questions, unseen database schemas, and novel SQL query structures. To comprehensively evaluate text-to-SQL systems, we introduce a UNIfied benchmark for Text-to-SQL Evaluation (UNITE). It is composed of publicly available text-to-SQL datasets, containing natural language questions from more than 12 domains, SQL queries from more than 3.9K patterns, and 29K databases. Compared to the widely used Spider benchmark, we introduce $\sim$120K additional examples and a threefold increase in SQL patterns, such as comparative and boolean questions. We conduct a systematic study of six state-of-the-art (SOTA) text-to-SQL parsers on our new benchmark and show that: 1) Codex performs surprisingly well on out-of-domain datasets; 2) specially designed decoding methods (e.g. constrained beam search) can improve performance for both in-domain and out-of-domain settings; 3) explicitly modeling the relationship between questions and schemas further improves the Seq2Seq models. More importantly, our benchmark presents key challenges towards compositional generalization and robustness issues -- which these SOTA models cannot address well. Our code and data processing script are available at https://github.com/awslabs/unified-text2sql-benchmark

cs.CL↗

arXiv2023

RxnScribe: A Sequence Generation Model for Reaction Diagram Parsing

Reaction diagram parsing is the task of extracting reaction schemes from a diagram in the chemistry literature. The reaction diagrams can be arbitrarily complex, thus robustly parsing them into structured data is an open challenge. In this paper, we present RxnScribe, a machine learning model for parsing reaction diagrams of varying styles. We formulate this structured prediction task with a sequence generation approach, which condenses the traditional pipeline into an end-to-end model. We train RxnScribe on a dataset of 1,378 diagrams and evaluate it with cross validation, achieving an 80.0% soft match F1 score, with significant improvements over previous models. Our code and data are publicly available at https://github.com/thomas0809/RxnScribe.

cs.CL↗

arXiv2023

MolScribe: Robust Molecular Structure Recognition with Image-To-Graph Generation

Molecular structure recognition is the task of translating a molecular image into its graph structure. Significant variation in drawing styles and conventions exhibited in chemical literature poses a significant challenge for automating this task. In this paper, we propose MolScribe, a novel image-to-graph generation model that explicitly predicts atoms and bonds, along with their geometric layouts, to construct the molecular structure. Our model flexibly incorporates symbolic chemistry constraints to recognize chirality and expand abbreviated structures. We further develop data augmentation strategies to enhance the model robustness against domain shifts. In experiments on both synthetic and realistic molecular images, MolScribe significantly outperforms previous models, achieving 76-93% accuracy on public benchmarks. Chemists can also easily verify MolScribe's prediction, informed by its confidence estimation and atom-level alignment with the input image. MolScribe is publicly available through Python and web interfaces: https://github.com/thomas0809/MolScribe.

cs.CV↗