arXiv subjects

Zhihan Guo

Publications and source records attributed to Zhihan Guo.

13 recordsLinked to original sources

arXiv2026

MisEdu-RAG: A Misconception-Aware Dual-Hypergraph RAG for Novice Math Teachers

Novice math teachers often encounter students' mistakes that are difficult to diagnose and remediate. Misconceptions are especially challenging because teachers must explain what went wrong and how to solve them. Although many existing large language model (LLM) platforms can assist in generating instructional feedback, these LLMs loosely connect pedagogical knowledge and student mistakes, which might make the guidance less actionable for teachers. To address this gap, we propose MisEdu-RAG, a dual-hypergraph-based retrieval-augmented generation (RAG) framework that organizes pedagogical knowledge as a concept hypergraph and real student mistake cases as an instance hypergraph. Given a query, MisEdu-RAG performs a two-stage retrieval to gather connected evidence from both layers and generates a response grounded in the retrieved cases and pedagogical principles. We evaluate on \textit{MisstepMath}, a dataset of math mistakes paired with teacher solutions, as a benchmark for misconception-aware retrieval and response generation across topics and error types. Evaluation results on \textit{MisstepMath} show that, compared with baseline models, MisEdu-RAG improves token-F1 by 10.95\% and yields up to 15.3\% higher five-dimension response quality, with the largest gains on \textit{Diversity} and \textit{Empowerment}. To verify its applicability in practical use, we further conduct a pilot study through a questionnaire survey of 221 teachers and interviews with 6 novices. The findings suggest that MisEdu-RAG provides diagnosis results and concrete teaching moves for high-demand misconception scenarios. Overall, MisEdu-RAG demonstrates strong potential for scalable teacher training and AI-assisted instruction for misconception handling. Our code is available on GitHub: https://github.com/GEMLab-HKU/MisEdu-RAG.

cs.IR↗

arXiv2026

TurnGuide: Enhancing Meaningful Full Duplex Spoken Interactions via Dynamic Turn-Level Text-Speech Interleaving

Full-Duplex Speech Language Models (FD-SLMs) are specialized foundation models designed to enable natural, real-time spoken interactions by modeling complex conversational turn-taking such as interruptions, backchannels, and overlapping speech. End-to-end (e2e) FD-SLMs leverage real-world double-channel conversational data to capture nuanced two-speaker dialogue patterns for human-like interactions, but their conversational abilities often degrade compared to pure-text conversation due to prolonged speech sequences and limited high-quality spoken dialogue data. Although interleaved text-speech generation could mitigate this degradation, integrating discrete text tokens into continuous double-channel audio streams could disrupt the precise time alignment required for fluid interaction. To address this, we propose TurnGuide, a novel text-speech interleaved generation approach for e2e FD-SLMs that dynamically segments assistant speech into dialogue turns and interleaves turn-level text and speech generation. This approach allows FD-SLMs to integrate the semantic intelligence of LLMs without compromising the natural acoustic flow. Extensive experiments show that TurnGuide not only significantly improves e2e FD-SLMs to produce semantically meaningful, coherent speech but also achieves state-of-the-art performance on various turn-taking events. Demos are available at https://dreamtheater123.github.io/TurnGuide-Demo/. Code is available at https://github.com/dreamtheater123/TurnGuide.

cs.CL↗

arXiv2026

ADRA-Bank: A Modular Benchmark for Academic Deep Research Agents

A surge in academic publications calls for automated deep research (DR) systems, but accurately evaluating them is still an open problem. First, existing benchmarks often focus narrowly on retrieval while neglecting high-level planning and reasoning. Second, existing benchmarks favor general domains over the academic domains that are the core application for DR agents. To address these gaps, we introduce ADRA-Bank, a modular benchmark for Academic DR Agents. Grounded in academic literature, our benchmark is a human-annotated dataset of 200 instances across 10 academic domains, including both research and review papers. Furthermore, we propose a modular Evaluation Paradigm for Academic DR Agents (ADRA-Eval), which leverages the rich structure of academic papers to assess the core capabilities of planning, retrieval, and reasoning. It employs two complementary modes: an end-to-end evaluation for \task agents and an isolated evaluation for foundational LLMs as potential backbones. Results reveal uneven capabilities: while agents show specialized strengths, they struggle with multi-source retrieval and cross-field consistency. Moreover, improving high-level planning capability is the crucial factor for unlocking the reasoning potential of foundational LLMs as backbones. By exposing these actionable failure modes, ADRA-Bank provides a diagnostic tool to guide the development of more reliable automatic academic research assistants.

cs.CL↗

arXiv2026

A Survey of Audio Reasoning in Multimodal Foundation Models

Reasoning has become a defining capability of modern foundation models, yet its development in the audio modality remains limited. Audio poses challenges that are distinct from those of text and vision. It is continuous, temporally dense, and contains linguistic, paralinguistic, and environmental information at multiple time scales. As a result, audio reasoning models must align acoustic signals with the discrete semantic space of large language models, while still preserving fine-grained information needed for reliable inference. Progress is also limited by three major obstacles: the scarcity of genuinely audio-grounded reasoning data, shortcut learning and modality hallucination, and the tension between reasoning depth and real-time latency in spoken interaction. In this paper, we present the first dedicated survey of audio reasoning. We provide a unified formulation that distinguishes direct predictive modeling from reasoning-augmented generation, review the architectural and training foundations of audio reasoning models, and systematically organize recent advances in Audio-to-Text, Audio-to-Speech, Audio-Visual Reasoning and Agentic Audio Reasoning. We further examine emerging paradigms such as Chain-of-Thought prompting, supervised fine-tuning, reinforcement learning, and latency-aware spoken interaction, and discuss evaluation practices, open challenges, and future directions. Our goal is to offer a coherent roadmap for developing robust, efficient, and natively grounded audio reasoning systems.

eess.AS↗

arXiv2026

Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memory

Long-horizon language agents must operate under limited runtime memory, yet existing memory mechanisms often organize experience around descriptive criteria such as relevance, salience, or summary quality. For an agent, however, memory is valuable not because it faithfully describes the past, but because it preserves the distinctions between histories that must remain separated under a fixed budget to support good decisions. We cast this as a decision-centric rate-distortion problem, measuring memory quality by the loss in achievable decision quality induced by compression. This yields an exact forgetting boundary for what can be safely forgotten, and a memory-distortion frontier characterizing the optimal tradeoff between memory budget and decision quality. Motivated by this decision-centric view of memory, we propose DeMem, an online memory learner that refines its partition only when data certify that a shared state would induce decision conflict, and prove near-minimax regret guarantees. On both controlled synthetic diagnostics and long-horizon conversational benchmarks, DeMem yields consistent gains under the same runtime budget, supporting the principle that memory should preserve the distinctions that matter for decisions, not descriptions.

cs.AI↗

arXiv2026

Simulating Novice Students Using Machine Unlearning and Relearning in Large Language Models

Student simulation can support learning-by-teaching pedagogy where human students (as tutors) teach AI-simulated novice students (as tutees). Recent research often relies on prompt engineering with large language models (LLMs) to simulate novice student behaviour, but it is difficult to keep the AI-simulated student at a stable novice knowledge level. A key reason is that many LLMs are trained to be broadly capable, so even when prompted to "act like a novice," the LLMs can still produce expert-level explanations during the learning-by-teaching interaction process. As a result, the AI-simulated student may drift beyond the intended knowledge level, reducing the credibility of the simulation for studying learning-by-teaching processes. Thus, we propose a knowledge-level simulation approach based on machine unlearning. We investigate this approach using a dataset of multiple-choice questions on Python programming concepts. We apply machine unlearning to transform a knowledgeable LLM into a novice-level AI student (i.e., teachable agent), then evaluate whether the teachable agent can relearn targeted knowledge components through learning-by-teaching dialogue interactions. Finally, we analyse the dialogue logs to characterise how the agent's behaviour changes over time, including its question asking, error patterns, and responsiveness to instruction. The results show that (1) unlearning produces simulated student agents with more novice-like responses than prompt-only baselines, (2) the agents recover a measurable portion of the unlearned knowledge under structured exposure, and (3) dialogue analyses reveal identifiable trajectories of conceptual change and teaching moves that predict learning recovery.

cs.HC↗

arXiv2026

Hierarchical Molecular Representation Learning via Fragment-Based Self-Supervised Embedding Prediction

Graph self-supervised learning (GSSL) has demonstrated strong potential for generating expressive graph embeddings without the need for human annotations, making it particularly valuable in domains with high labeling costs such as molecular graph analysis. However, existing GSSL methods mostly focus on node- or edge-level information, often ignoring chemically relevant substructures which strongly influence molecular properties. In this work, we propose Graph Semantic Predictive Network (GraSPNet), a hierarchical self-supervised framework that explicitly models both atomic-level and fragment-level semantics. GraSPNet decomposes molecular graphs into chemically meaningful fragments without predefined vocabularies and learns node- and fragment-level representations through multi-level message passing with masked semantic prediction at both levels. This hierarchical semantic supervision enables GraSPNet to learn multi-resolution structural information that is both expressive and transferable. Extensive experiments on multiple molecular property prediction benchmarks demonstrate that GraSPNet learns chemically meaningful representations and consistently outperforms state-of-the-art GSSL methods in transfer learning settings.

cs.LG↗

arXiv2026

Insular intracranial activity identifies multiple facial expressions via diverse, intermixed temporal patterns at the single-contact level

How neural representations in the insular cortex support emotional processing remains poorly understood, and the extent to which the insula is specialized for disgust processing remains debated. We recorded stereoelectroencephalography data from the insula while human subjects with implanted electrode contacts performed a facial emotion recognition task involving disgusted, fearful, angry, sad, neutral, and happy expressions. Expression category specificity of insular activity was assessed via pairwise comparisons of within- and between-category pattern similarities, capturing both the shape and scale of event-related potentials (ERPs) and event-related spectral perturbations (ERSPs; theta to high-gamma frequency ranges). Insular activity successfully identified all investigated expressions, mediated by diverse ERP responses intermixed across the insula. In contrast to the marked heterogeneity of insula ERP responses, the fusiform face area exhibited convergent ERP responses across expressions and contacts, with ERSPs also contributing substantially to expression identification. These findings not only elucidate the insula's neural mechanisms underlying facial emotion perception, but also establish a potential single-contact-level neural substrate for how the insula leverages its heterogeneous response profiles to act as a key hub for versatile cognitive and emotional functions.

q-bio.NC↗

arXiv2025

From General to Targeted Rewards: Surpassing GPT-4 in Open-Ended Long-Context Generation

Current research on long-form context in Large Language Models (LLMs) primarily focuses on the understanding of long-contexts, the Open-ended Long Text Generation (Open-LTG) remains insufficiently explored. Training a long-context generation model requires curation of gold standard reference data, which is typically nonexistent for informative Open-LTG tasks. However, previous methods only utilize general assessments as reward signals, which limits accuracy. To bridge this gap, we introduce ProxyReward, an innovative reinforcement learning (RL) based framework, which includes a dataset and a reward signal computation method. Firstly, ProxyReward Dataset generation is accomplished through simple prompts that enables the model to create automatically, obviating extensive labeled data or significant manual effort. Secondly, ProxyReward Signal offers a targeted evaluation of information comprehensiveness and accuracy for specific questions. The experimental results indicate that our method ProxyReward surpasses even GPT-4-Turbo. It can significantly enhance performance by 20% on the Open-LTG task when training widely used open-source models, while also surpassing the LLM-as-a-Judge approach. Our work presents effective methods to enhance the ability of LLMs to address complex open-ended questions posed by human.

cs.CL↗

arXiv2025

A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?

As enthusiasm for scaling computation (data and parameters) in the pretraining era gradually diminished, test-time scaling (TTS), also referred to as ``test-time computing'' has emerged as a prominent research focus. Recent studies demonstrate that TTS can further elicit the problem-solving capabilities of large language models (LLMs), enabling significant breakthroughs not only in specialized reasoning tasks, such as mathematics and coding, but also in general tasks like open-ended Q&A. However, despite the explosion of recent efforts in this area, there remains an urgent need for a comprehensive survey offering a systemic understanding. To fill this gap, we propose a unified, multidimensional framework structured along four core dimensions of TTS research: what to scale, how to scale, where to scale, and how well to scale. Building upon this taxonomy, we conduct an extensive review of methods, application scenarios, and assessment aspects, and present an organized decomposition that highlights the unique functional roles of individual techniques within the broader TTS landscape. From this analysis, we distill the major developmental trajectories of TTS to date and offer hands-on guidelines for practical deployment. Furthermore, we identify several open challenges and offer insights into promising future directions, including further scaling, clarifying the functional essence of techniques, generalizing to more tasks, and more attributions. Our repository is available on https://github.com/testtimescaling/testtimescaling.github.io/

cs.CL↗

arXiv2022

Cornus: Atomic Commit for a Cloud DBMS with Storage Disaggregation (Extended Version)

Two-phase commit (2PC) is widely used in distributed databases to ensure the atomicity of distributed transactions. However, 2PC has two limitations. First, it requires two eager log writes on the critical path, which incurs significant latency. Second, when a coordinator fails, a participant may be blocked waiting for the coordinator's decision, leading to indefinitely long latency and low throughput. 2PC was originally designed for a shared-nothing architecture. We observe that the two problems above can be addressed in an emerging storage disaggregation architecture which provides compare-and-swap capability in the storage layer. We propose Cornus, an optimized 2PC protocol for Cloud DBMS with Storage Disaggregation. We present Cornus in detail with proofs and show how it addresses the two limitations in 2PC. We also deploy it on real storage services including Azure Blob Storage and Redis. Empirical evaluations show that Cornus can achieve up to 1.9x speedup in latency over 2PC.

cs.DB↗

arXiv2021

Releasing Locks As Early As You Can: Reducing Contention of Hotspots by Violating Two-Phase Locking (Extended Version)

Hotspots, a small set of tuples frequently read/written by a large number of transactions, cause contention in a concurrency control protocol. While a hotspot may comprise only a small fraction of a transaction's execution time, conventional strict two-phase locking allows a transaction to release lock only after the transaction completes, which leaves significant parallelism unexploited. Ideally, a concurrency control protocol serializes transactions only for the duration of the hotspots, rather than the duration of transactions. We observe that exploiting such parallelism requires violating two-phase locking. In this paper, we propose Bamboo, a new concurrency control protocol that can enable such parallelism by modifying the conventional two-phase locking, while maintaining the same guarantees in correctness. We thoroughly analyzed the effect of cascading aborts involved in reading uncommitted data and discussed optimizations that can be applied to further improve the performance. Our evaluation on TPC-C shows a performance improvement up to 4x compared to the best of pessimistic and optimistic baseline protocols. On synthetic workloads that contain a single hotspot, Bamboo achieves a speedup up to 19x over baselines.

cs.DB↗

arXiv2019

Learning Functional Dependencies with Sparse Regression

We study the problem of discovering functional dependencies (FD) from a noisy dataset. We focus on FDs that correspond to statistical dependencies in a dataset and draw connections between FD discovery and structure learning in probabilistic graphical models. We show that discovering FDs from a noisy dataset is equivalent to learning the structure of a graphical model over binary random variables, where each random variable corresponds to a functional of the dataset attributes. We build upon this observation to introduce AutoFD a conceptually simple framework in which learning functional dependencies corresponds to solving a sparse regression problem. We show that our methods can recover true functional dependencies across a diverse array of real-world and synthetic datasets, even in the presence of noisy or missing data. We find that AutoFD scales to large data instances with millions of tuples and hundreds of attributes while it yields an average F1 improvement of 2 times against state-of-the-art FD discovery methods.

cs.DB↗