arXiv subjects

Jinyu Liu

Publications and source records attributed to Jinyu Liu.

At least 19 recordsLinked to original sources

arXiv2026

Ask or Answer: A Decision Framework for Multi-Turn Health Misinformation Intervention

Correcting health misinformation in dialogue requires more than producing a factual rebuttal: users differ in what they know, what they believe, and what they need to hear, so an effective intervention often depends on first asking the right clarifying question. Yet existing methods either respond immediately or probe indiscriminately, treating clarification as either unnecessary or always beneficial. We propose Reward-Optimized Probe-and-Respond (RO-PnR), a framework that learns when asking is worth its cost. At each turn, RO-PnR chooses between probing for more information and committing to a final correction, guided by a turn-level reward that weighs the expected gain from probing against its interaction cost. To capture how user heterogeneity affects probing value, we model each simulated user with a latent state along health literacy and belief commitment. Experiments show that RO-PnR achieves the highest cost-adjusted utility across three health-misinformation datasets and three base models, using 30% fewer turns than always-probe baselines.

cs.AI↗

arXiv2026

Unison: Benchmarking Unified Multimodal Models via Synergistic Understanding and Generation

Unified multimodal models capable of both understanding and generation have achieved remarkable strides. However, despite their unified designs, existing evaluations typically assess understanding and generation capabilities in isolation, overlooking the synergy between comprehension and generation. To bridge this gap, we introduce Unison, a comprehensive benchmark comprising 2,169 high-quality unified task samples, designed to evaluate joint understanding and generation in unified multimodal models. Unison offers three key strengths: 1) Comprehensive Dimensions: Unison encompasses internal consistency, understanding-guided generation, generation-guided understanding, and mutual enhancement to enable holistic evaluation. 2) Diagnostic Evaluation: it provides both unified and decoupled tracks for understanding and generation, allowing fine-grained attribution of failure modes and quantitative analysis of the gains from unified modeling. 3) Human Alignment: we also introduce Unison-Judge, an evaluation model well aligned with human judgments to ensure reliable assessment. Based on systematic evaluations of state-of-the-art models on Unison, we uncover critical limitations in current unified multimodal systems and highlight promising directions for future research. Codes, Unison and Unison-Judge are publicly available at https://github.com/FudanCVL/Unison.

cs.CV↗

arXiv2026

Event-Aware Instructed Assistant for Referring Video Segmentation

Existing referring video segmentation methods often treat a video as a single event consisting of multiple images, overlooking the fact that a video typically contains multiple distinct events. Under such a mechanism, the model needs to directly understand all the complex content in the video and text, which can easily lead to confusion and hallucinations. To address this issue, we propose to decompose a video to a set of simple events by learnable Event Query, and understand complex video content in an event-by-event, easy-to-understand manner. This is based on the observation that natural language expressions often divide a video into distinct, text-related segments, each representing a separate event within a compound event. We introduce EVIS, an Event-Aware Video Instructed Segmentation Assistant, which utilizes text-guided Event Queries to partition a video into simple events, extracting event-aware visual-text features to achieve a hierarchical understanding of the video. Additionally, we propose Object-Pixel-Hybrid Learning, which enables the MLLMs to track targets in long-term videos by integrating fine-grained pixel features with prior object queries. Extensive experimental results on 5 public benchmarks demonstrate EVIS's strong performance in addressing the referring video segmentation task.

cs.CV↗

arXiv2026

Real-Time Glottis Detection Framework via Spatial-decoupled Feature Learning for Nasal Transnasal Intubation

Nasotracheal intubation (NTI) is a vital procedure in emergency airway management, where rapid and accurate glottis detection is essential to ensure patient safety. However, existing machine assisted visual detection systems often rely on high performance computational resources and suffer from significant inference delays, which limits their applicability in time critical and resource constrained scenarios. To overcome these limitations, we propose Mobile GlottisNet, a lightweight and efficient glottis detection framework designed for real time inference on embedded and edge devices. The model incorporates structural awareness and spatial alignment mechanisms, enabling robust glottis localization under complex anatomical and visual conditions. We implement a hierarchical dynamic thresholding strategy to enhance sample assignment, and introduce an adaptive feature decoupling module based on deformable convolution to support dynamic spatial reconstruction. A cross layer dynamic weighting scheme further facilitates the fusion of semantic and detail features across multiple scales. Experimental results demonstrate that the model, with a size of only 5MB on both our PID dataset and Clinical datasets, achieves inference speeds of over 62 FPS on devices and 33 FPS on edge platforms, showing great potential in the application of emergency NTI.

cs.CV↗

arXiv2026

Temporal Data and Short-Time Averages Improve Multiphase Mass Flow Metering

Reliable flow measurements are essential in many industries, but current instruments often fail to accurately estimate multiphase flows, which are frequently encountered in real-world operations. Combining machine learning (ML) algorithms with accurate single-phase flowmeters has therefore received extensive research attention in recent years. The Coriolis mass flowmeter is a widely used single-phase meter that provides direct mass flow measurements, which ML models can be trained to correct, thereby reducing measurement errors in multiphase conditions. This paper demonstrates that preserving temporal information significantly improves model performance in such scenarios. We compare a multilayer perceptron, a windowed multilayer perceptron, and a convolutional neural network (CNN) on three-phase air-water-oil flow data from 342 experiments. Whereas prior work typically compresses each experiment into a single averaged sample, we instead compute short-time averages from within each experiment and train models that preserve temporal information at several downsampling intervals. The CNN performed best at 0.25 Hz with approximately 95 % of relative errors below 13 %, a normalized root mean squared error of 0.03, and a mean absolute percentage error of approximately 4.3 %, clearly outperforming the best single-averaged model and demonstrating that short-time averaging within individual experiments is preferable. Results are consistent across multiple data splits and random seeds, demonstrating robustness.

eess.SP↗

arXiv2025

Ref-SAM3D: Bridging SAM3D with Text for Reference 3D Reconstruction

SAM3D has garnered widespread attention for its strong 3D object reconstruction capabilities. However, a key limitation remains: SAM3D cannot reconstruct specific objects referred to by textual descriptions, a capability that is essential for practical applications such as 3D editing, game development, and virtual environments. To address this gap, we introduce Ref-SAM3D, a simple yet effective extension to SAM3D that incorporates textual descriptions as a high-level prior, enabling text-guided 3D reconstruction from a single RGB image. Through extensive qualitative experiments, we show that Ref-SAM3D, guided only by natural language and a single 2D view, delivers competitive and high-fidelity zero-shot reconstruction performance. Our results demonstrate that Ref-SAM3D effectively bridges the gap between 2D visual cues and 3D geometric understanding, offering a more flexible and accessible paradigm for reference-guided 3D reconstruction. Code is available at: https://github.com/FudanCVL/Ref-SAM3D.

cs.CV↗

arXiv2025

CrypTorch: PyTorch-based Auto-tuning Compiler for Machine Learning with Multi-party Computation

MPC-based ML uses multi-party computation (MPC) to run machine learning (ML) workloads across multiple parties without each having to share their private data or model parameters. However, existing frameworks frequently degrade accuracy and performance due to a series of MPC-specific transformations that add errors and overheads. These transformations are mostly opaque to users, making it hard to find and/or optimize problematic transformations. We propose CrypTorch, a modular, extensible, and iteratively-testable compiler framework for MPC-based ML. CrypTorch splits MPC-specific transformations into modular compilation stages, allowing users to easily inspect and optimize them. CrypTorch emits an executable graph after each transformation, allowing iterative testing to pinpoint any problematic transformations. Building on these features, CrypTorch automatically chooses a set of transformations from a pool of choices to balance performance and accuracy during the operator approximation stage, which we identified as the biggest contributor to accuracy/performance degradation. CrypTorch's auto-tuning alone provides 1.21-1.5x speedup without accuracy loss, and 1.33-1.74x speedup when some accuracy degradation is allowed. Combined with better engineering and adoption of state-of-the-art practices (made easier due to CrypTorch's modular design), CrypTorch brings 3.74-8.32x end-to-end speedup compared to the popular CrypTen. CrypTorch is built as an extension to PyTorch 2's compiler.

cs.CR↗

arXiv2025

A Multi-Agent System for Complex Reasoning in Radiology Visual Question Answering

Radiology visual question answering (RVQA) provides precise answers to questions about chest X-ray images, alleviating radiologists' workload. While recent methods based on multimodal large language models (MLLMs) and retrieval-augmented generation (RAG) have shown promising progress in RVQA, they still face challenges in factual accuracy, hallucinations, and cross-modal misalignment. We introduce a multi-agent system (MAS) designed to support complex reasoning in RVQA, with specialized agents for context understanding, multimodal reasoning, and answer validation. We evaluate our system on a challenging RVQA set curated via model disagreement filtering, comprising consistently hard cases across multiple MLLMs. Extensive experiments demonstrate the superiority and effectiveness of our system over strong MLLM baselines, with a case study illustrating its reliability and interpretability. This work highlights the potential of multi-agent approaches to support explainable and trustworthy clinical AI applications that require complex reasoning.

cs.AI↗

arXiv2025

A Hybrid Framework for Subject Analysis: Integrating Embedding-Based Regression Models with Large Language Models

Providing subject access to information resources is an essential function of any library management system. Large language models (LLMs) have been widely used in classification and summarization tasks, but their capability to perform subject analysis is underexplored. Multi-label classification with traditional machine learning (ML) models has been used for subject analysis but struggles with unseen cases. LLMs offer an alternative but often over-generate and hallucinate. Therefore, we propose a hybrid framework that integrates embedding-based ML models with LLMs. This approach uses ML models to (1) predict the optimal number of LCSH labels to guide LLM predictions and (2) post-edit the predicted terms with actual LCSH terms to mitigate hallucinations. We experimented with LLMs and the hybrid framework to predict the subject terms of books using the Library of Congress Subject Headings (LCSH). Experiment results show that providing initial predictions to guide LLM generations and imposing post-edits result in more controlled and vocabulary-aligned outputs.

cs.CL↗

arXiv2025

Spin-Triplet Excitonic Insulator in the Ultra-Quantum Limit of HfTe5

More than fifty years ago, excitonic insulators, formed by the pairing of electrons and holes due to Coulomb interactions, were first predicted. Since then, excitonic insulators have been observed in various classes of materials, including quantum Hall bilayers, graphite, transition metal chalcogenides, and more recently in moire superlattices. In these excitonic insulators, an electron and a hole with the same spin bind together and the resulting exciton is a spin singlet. Here, we report the experimental observation of a spin-triplet exciton insulator in the ultra-quantum limit of a three-dimensional topological material HfTe5. We observe that the spin-polarized zeroth Landau bands, dispersing along the field direction, cross each other beyond a characteristic magnetic field in HfTe5, forming the one-dimensional Weyl mode. Transport measurements reveal the emergence of a gap of about 250 {\mu}eV when the field surpasses a critical threshold. By performing the material-specific modeling, we identify this gap as a consequence of a spin-triplet exciton formation, where electrons and holes with opposite spin form bound states, and the translational symmetry is preserved. The system reaches charge neutrality following the gap opening, as evidenced by the zero Hall conductivity over a wide magnetic field range (10 - 72 T). Our finding of the spin-triplet excitonic insulator paves the way for studying novel spin transport including spin superfluidity, spin Josephson currents, and Coulomb drag of spin currents in analogy to the transport properties associated with the layer pseudospin in quantum Hall bilayers.

cond-mat.str-el↗

arXiv2024

Reinforcement Learning Based Bidding Framework with High-dimensional Bids in Power Markets

Over the past decade, bidding in power markets has attracted widespread attention. Reinforcement Learning (RL) has been widely used for power market bidding as a powerful AI tool to make decisions under real-world uncertainties. However, current RL methods mostly employ low dimensional bids, which significantly diverge from the N price-power pairs commonly used in the current power markets. The N-pair bidding format is denoted as High Dimensional Bids (HDBs), which has not been fully integrated into the existing RL-based bidding methods. The loss of flexibility in current RL bidding methods could greatly limit the bidding profits and make it difficult to tackle the rising uncertainties brought by renewable energy generations. In this paper, we intend to propose a framework to fully utilize HDBs for RL-based bidding methods. First, we employ a special type of neural network called Neural Network Supply Functions (NNSFs) to generate HDBs in the form of N price-power pairs. Second, we embed the NNSF into a Markov Decision Process (MDP) to make it compatible with most existing RL methods. Finally, experiments on Energy Storage Systems (ESSs) in the PJM Real-Time (RT) power market show that the proposed bidding method with HDBs can significantly improve bidding flexibility, thereby improving the profit of the state-of-the-art RL bidding methods.

cs.LG↗

arXiv2024

A Reproducibility Study on Quantifying Language Similarity: The Impact of Missing Values in the URIEL Knowledge Base

In the pursuit of supporting more languages around the world, tools that characterize properties of languages play a key role in expanding the existing multilingual NLP research. In this study, we focus on a widely used typological knowledge base, URIEL, which aggregates linguistic information into numeric vectors. Specifically, we delve into the soundness and reproducibility of the approach taken by URIEL in quantifying language similarity. Our analysis reveals URIEL's ambiguity in calculating language distances and in handling missing values. Moreover, we find that URIEL does not provide any information about typological features for 31\% of the languages it represents, undermining the reliabilility of the database, particularly on low-resource languages. Our literature review suggests URIEL and lang2vec are used in papers on diverse NLP tasks, which motivates us to rigorously verify the database as the effectiveness of these works depends on the reliability of the information the tool provides.

cs.CL↗

arXiv2024

Rydberg exciton states and near-infrared light-emitting diode in monolayer MoTe2 devices

Excitons, or bound electron-hole pairs, play a crucial role in the optical response of monolayer, 2H-phase transition-metal dichalcogenides (TMDs). They hold significant promise for the development of novel quantum opto-electronic devices due to their large binding energies and strong spin-orbit coupling. Among the monolayer TMDs, MoTe2 stands out because of its bandgap in the near-infrared (NIR) regime. Here, we report the experimental observation of NIR Rydberg excitons and conduction band-split charged excitons, in high-quality, boron nitride (BN)-encapsulated monolayer MoTe2 devices, probed by photoluminescence and electroluminescence spectroscopy. By employing a graphite bottom gate, we successfully modulate the emission intensity of various excitonic species. Additionally, our device fabrication process within an argon-filled glove box ensures clean TMD/metal electrode interfaces, enabling the construction of p-n junctions near the electrodes. Our work significantly advances our understanding of excitons in monolayer TMDs and contributes to the application of MoTe2 in NIR quantum opto-electronic devices.

cond-mat.mtrl-sci↗

arXiv2024

Predicting Machine Translation Performance on Low-Resource Languages: The Role of Domain Similarity

Fine-tuning and testing a multilingual large language model is expensive and challenging for low-resource languages (LRLs). While previous studies have predicted the performance of natural language processing (NLP) tasks using machine learning methods, they primarily focus on high-resource languages, overlooking LRLs and shifts across domains. Focusing on LRLs, we investigate three factors: the size of the fine-tuning corpus, the domain similarity between fine-tuning and testing corpora, and the language similarity between source and target languages. We employ classical regression models to assess how these factors impact the model's performance. Our results indicate that domain similarity has the most critical impact on predicting the performance of Machine Translation models.

cs.CL↗

arXiv2023

High-dimensional Bid Learning for Energy Storage Bidding in Energy Markets

With the growing penetration of renewable energy resource, electricity market prices have exhibited greater volatility. Therefore, it is important for Energy Storage Systems(ESSs) to leverage the multidimensional nature of energy market bids to maximize profitability. However, current learning methods cannot fully utilize the high-dimensional price-quantity bids in the energy markets. To address this challenge, we modify the common reinforcement learning(RL) process by proposing a new bid representation method called Neural Network Embedded Bids (NNEBs). NNEBs refer to market bids that are represented by monotonic neural networks with discrete outputs. To achieve effective learning of NNEBs, we first learn a neural network as a strategic mapping from the market price to ESS power output with RL. Then, we re-train the network with two training modifications to make the network output monotonic and discrete. Finally, the neural network is equivalently converted into a high-dimensional bid for bidding. We conducted experiments over real-world market datasets. Our studies show that the proposed method achieves 18% higher profit than the baseline and up to 78% profit of the optimal market bidder.

eess.SY↗

arXiv2023

Controllable Strain-driven Topological Phase Transition and Dominant Surface State Transport in High-Quality HfTe5 Samples

Controlling materials to create and tune topological phases of matter could potentially be used to explore new phases of topological quantum matter and to create novel devices where the carriers are topologically protected. It has been demonstrated that a trivial insulator can be converted into a topological state by modulating the spin-orbit interaction or the crystal lattice. However, there are limited methods to controllably and efficiently tune the crystal lattice and at the same time perform electronic measurements at cryogenic temperatures. Here, we use large controllable strain to demonstrate the topological phase transition from a weak topological insulator phase to a strong topological insulator phase in high-quality HfTe5 samples. After applying high strain to HfTe5 and converting it into a strong topological insulator, we found that the sample's resistivity increased by more than two orders of magnitude (24,000%) and that the electronic transport is dominated by the topological surface states at cryogenic temperatures. Our findings show that HfTe5 is an ideal material for engineering topological properties, and it could be generalized to study topological phase transitions in van der Waals materials and heterostructures. These results can pave the way to create novel devices with applications ranging from spintronics to fault-tolerant topologically protected quantum computers.

cond-mat.mtrl-sci↗

arXiv2023

Two-Stream Joint-Training for Speaker Independent Acoustic-to-Articulatory Inversion

Acoustic-to-articulatory inversion (AAI) aims to estimate the parameters of articulators from speech audio. There are two common challenges in AAI, which are the limited data and the unsatisfactory performance in speaker independent scenario. Most current works focus on extracting features directly from speech and ignoring the importance of phoneme information which may limit the performance of AAI. To this end, we propose a novel network called SPN that uses two different streams to carry out the AAI task. Firstly, to improve the performance of speaker-independent experiment, we propose a new phoneme stream network to estimate the articulatory parameters as the phoneme features. To the best of our knowledge, this is the first work that extracts the speaker-independent features from phonemes to improve the performance of AAI. Secondly, in order to better represent the speech information, we train a speech stream network to combine the local features and the global features. Compared with state-of-the-art (SOTA), the proposed method reduces 0.18mm on RMSE and increases 6.0% on Pearson correlation coefficient in the speaker-independent experiment. The code has been released at https://github.com/liujinyu123/AAINetwork-SPN.

cs.SD↗

arXiv2023

Magnetic tuning of band topology evidenced by exotic quantum oscillations in the Dirac semimetal EuMnSb$_2$

Interplay between magnetism and electronic band topology is of central current interest in topological matter research. We use quantum oscillations as powerful tool to probe the evolution of band topology in the Dirac semimetal EuMnSb$_2$. The Eu local 4f magnetic moments display different antiferromagnetic states below 25 K and a field-polarized phase above 16 T. Upon cooling from 65 K into the field-polarized state, an exotic temperature dependent shift of oscillation peaks arises, accompanied by the development of non-zero Berry phase and a huge unconventional splitting of the oscillations. Band-structure calculations confirm the change from trivial to non-trivial band topology induced by the ferromagnetic Eu state, classifying EuMnSb$_2$ as unique magnetic topological semimetal.

cond-mat.str-el↗