arXiv subjects

Ying Su

Publications and source records attributed to Ying Su.

At least 19 recordsLinked to original sources

arXiv2026

Jailbreaking Large Language Models with Morality Attacks

Pluralism alignment with AI has the sophisticated and necessary goal of creating AI that can coexist with and serve morally multifaceted humanity. Research towards pluralism alignment has many efforts in enhancing the learning of large language models (LLMs) to accomplish pluralism. Although this is essential, the robustness of LLMs to produce moral content over pluralistic values is still under exploration.Inspired by the astonishing persuasion abilities via jailbreak prompts, we propose to leverage jailbreak attacks to study LLMs' internal pluralistic values. In detail, we develop a morality dataset with 10.3K instances in two categories: Value Ambiguity and Value Conflict. We further formalize four adversarial attacks with the constructed dataset, to manipulate LLMs' judgment over the morality questions. We evaluate both the large language models and guardrail models which are typically used in generative systems with flexible user input. Our experiment results show that there is a critical vulnerability of LLMs and guardrail models to these subtle and sophisticated moral-aware attacks.

cs.CL↗

arXiv2026

Tilewise Domain-Separated Selective Encryption for Remote Sensing Imagery under Chosen-Plaintext Attacks

Selective image encryption is common in remote sensing systems because it protects sensitive regions of interest (ROI) while limiting computational cost. However, many selective designs enable cross-tile structural leakage under chosen-plaintext attacks when secret-dependent transformations are reused across spatial positions. This paper proposes Tilewise Domain-Separated Selective Encryption (TDS-SE), where per-tile (and optionally per-frame) keys are derived from a master secret via HKDF with explicit domain separation, and ROI masks are treated strictly as external side information. Structural leakage is evaluated using two reconstruction-based distinguishers -- a linear model and a lightweight convolutional neural network -- under multiple attack settings. Experiments on RESISC45 and SEN12MS cover ablation tests, cross-position transferability, cross-sample generalization, and ROI-knowledge asymmetry. Results show that per-tile domain separation reduces position-conditioned transfer for the linear probe, and that adding frame freshness improves robustness to imperfect ROI assumptions for the CNN probe. Cross-sample generalization exhibits mixed behavior across settings, consistent with an empirical evaluation perspective, while selective-encryption functionality is preserved under the same tiling and ROI policy. Beyond the method itself, the paper provides a structured protocol for evaluating selective encryption under realistic attacker capabilities.

cs.CR↗

arXiv2025

Unsupervised Feature Selection via Robust Autoencoder and Adaptive Graph Learning

Effective feature selection is essential for high-dimensional data analysis and machine learning. Unsupervised feature selection (UFS) aims to simultaneously cluster data and identify the most discriminative features. Most existing UFS methods linearly project features into a pseudo-label space for clustering, but they suffer from two critical limitations: (1) an oversimplified linear mapping that fails to capture complex feature relationships, and (2) an assumption of uniform cluster distributions, ignoring outliers prevalent in real-world data. To address these issues, we propose the Robust Autoencoder-based Unsupervised Feature Selection (RAEUFS) model, which leverages a deep autoencoder to learn nonlinear feature representations while inherently improving robustness to outliers. We further develop an efficient optimization algorithm for RAEUFS. Extensive experiments demonstrate that our method outperforms state-of-the-art UFS approaches in both clean and outlier-contaminated data settings.

stat.ML↗

arXiv2025

Interaction-Driven Altermagnetic Magnon Chiral Splitting

Nonrelativistic magnon chiral splitting in altermagnets has garnered significant recent attention. In this work, we demonstrate that nonlinear three-wave mixing -- where magnons split or coalesce -- extends this phenomenon into unprecedented relativistic regimes. Employing a bilayer antiferromagnet with Dzyaloshinskii-Moriya interactions, we identify three distinct classes of chiral splitting, each dictated by specific symmetries, such as $C_4T$, $\sigma_v T$, or their combination. This reveals a novel bosonic mechanism for symmetry-protected chiral splitting, capitalizing on the unique ability of magnons to violate particle-number conservation, a feature absent in low-energy fermionic systems. Our findings pave the way for engineering altermagnetic splitting, with potential applications in advanced magnonic devices and deeper insights into magnon dynamics in complex magnetic systems.

cond-mat.mtrl-sci↗

arXiv2024

ActPlan-1K: Benchmarking the Procedural Planning Ability of Visual Language Models in Household Activities

Large language models~(LLMs) have been adopted to process textual task description and accomplish procedural planning in embodied AI tasks because of their powerful reasoning ability. However, there is still lack of study on how vision language models~(VLMs) behave when multi-modal task inputs are considered. Counterfactual planning that evaluates the model's reasoning ability over alternative task situations are also under exploited. In order to evaluate the planning ability of both multi-modal and counterfactual aspects, we propose ActPlan-1K. ActPlan-1K is a multi-modal planning benchmark constructed based on ChatGPT and household activity simulator iGibson2. The benchmark consists of 153 activities and 1,187 instances. Each instance describing one activity has a natural language task description and multiple environment images from the simulator. The gold plan of each instance is action sequences over the objects in provided scenes. Both the correctness and commonsense satisfaction are evaluated on typical VLMs. It turns out that current VLMs are still struggling at generating human-level procedural plans for both normal activities and counterfactual activities. We further provide automatic evaluation metrics by finetuning over BLEURT model to facilitate future research on our benchmark.

cs.CL↗

arXiv2024

Quantum Nonlinear Acoustic Hall Effect and Inverse Acoustic Faraday Effect in Dirac Insulators

We propose to realize the quantum nonlinear Hall effect and the inverse Faraday effect through the acoustic wave in a time-reversal invariant but inversion broken Dirac insulator. We focus on the acoustic frequency much lower than the Dirac gap such that the interband transition is suppressed and these effects arise solely from the intrinsic valley-contrasting band topology. The corresponding acoustoelectric conductivity and magnetoacoustic susceptibility are both proportional to the quantized valley Chern number and independent of the quasiparticle lifetime. The linear and nonlinear components of the longitudinal and transverse topological currents can be tuned by adjusting the polarization and propagation directions of the surface acoustic wave. The static magnetization generated by a circularly polarized acoustic wave scales linearly with the acoustic frequency as well as the strain-induced charge density. Our results unveil a quantized nonlinear topological acoustoelectric response of gapped Dirac materials, like hBN and transition-metal dichalcogenide, paving the way toward room-temperature acoustoelectric devices due to their large band gaps.

cond-mat.mes-hall↗

arXiv2024

LLMs Assist NLP Researchers: Critique Paper (Meta-)Reviewing

This work is motivated by two key trends. On one hand, large language models (LLMs) have shown remarkable versatility in various generative tasks such as writing, drawing, and question answering, significantly reducing the time required for many routine tasks. On the other hand, researchers, whose work is not only time-consuming but also highly expertise-demanding, face increasing challenges as they have to spend more time reading, writing, and reviewing papers. This raises the question: how can LLMs potentially assist researchers in alleviating their heavy workload? This study focuses on the topic of LLMs assist NLP Researchers, particularly examining the effectiveness of LLM in assisting paper (meta-)reviewing and its recognizability. To address this, we constructed the ReviewCritique dataset, which includes two types of information: (i) NLP papers (initial submissions rather than camera-ready) with both human-written and LLM-generated reviews, and (ii) each review comes with "deficiency" labels and corresponding explanations for individual segments, annotated by experts. Using ReviewCritique, this study explores two threads of research questions: (i) "LLMs as Reviewers", how do reviews generated by LLMs compare with those written by humans in terms of quality and distinguishability? (ii) "LLMs as Metareviewers", how effectively can LLMs identify potential issues, such as Deficient or unprofessional review segments, within individual paper reviews? To our knowledge, this is the first work to provide such a comprehensive analysis.

cs.CL↗

arXiv2024

Gauge theory of giant phonon magnetic moment in doped Dirac semimetals

We present a quantum theory of phonon magnetic moment in doped Dirac semimetals. Our theory is based on an emergent gauge field approach to the electron-phonon coupling, applicable to both gapless and gapped systems. We find that the magnetic moment is directly proportional to the electrical Hall conductivity through the phonon Hall viscosity. Our theory is combined with the first-principles calculations, allowing us to quantitatively implement it to realistic materials. Magnetic moments are found to be on the order of Bohr magneton for certain phonon modes in graphene and $\text{Cd}_3 \text{As}_2$. Our results provide practical guidance for the dynamical generation of large magnetization in the topological quantum materials.

cond-mat.mtrl-sci↗

arXiv2024

DPER: Diffusion Prior Driven Neural Representation for Limited Angle and Sparse View CT Reconstruction

Limited-angle and sparse-view computed tomography (LACT and SVCT) are crucial for expanding the scope of X-ray CT applications. However, they face challenges due to incomplete data acquisition, resulting in diverse artifacts in the reconstructed CT images. Emerging implicit neural representation (INR) techniques, such as NeRF, NeAT, and NeRP, have shown promise in under-determined CT imaging reconstruction tasks. However, the unsupervised nature of INR architecture imposes limited constraints on the solution space, particularly for the highly ill-posed reconstruction task posed by LACT and ultra-SVCT. In this study, we introduce the Diffusion Prior Driven Neural Representation (DPER), an advanced unsupervised framework designed to address the exceptionally ill-posed CT reconstruction inverse problems. DPER adopts the Half Quadratic Splitting (HQS) algorithm to decompose the inverse problem into data fidelity and distribution prior sub-problems. The two sub-problems are respectively addressed by INR reconstruction scheme and pre-trained score-based diffusion model. This combination first injects the implicit image local consistency prior from INR. Additionally, it effectively augments the feasibility of the solution space for the inverse problem through the generative diffusion model, resulting in increased stability and precision in the solutions. We conduct comprehensive experiments to evaluate the performance of DPER on LACT and ultra-SVCT reconstruction with two public datasets (AAPM and LIDC), an in-house clinical COVID-19 dataset and a public raw projection dataset created by Mayo Clinic. The results show that our method outperforms the state-of-the-art reconstruction methods on in-domain datasets, while achieving significant performance improvements on out-of-domain (OOD) datasets.

eess.IV↗

arXiv2024

Rethinking the Bounds of LLM Reasoning: Are Multi-Agent Discussions the Key?

Recent progress in LLMs discussion suggests that multi-agent discussion improves the reasoning abilities of LLMs. In this work, we reevaluate this claim through systematic experiments, where we propose a novel group discussion framework to enrich the set of discussion mechanisms. Interestingly, our results show that a single-agent LLM with strong prompts can achieve almost the same performance as the best existing discussion approach on a wide range of reasoning tasks and backbone LLMs. We observe that the multi-agent discussion performs better than a single agent only when there is no demonstration in the prompt. Further study reveals the common interaction mechanisms of LLMs during the discussion.

cs.CL↗

arXiv2024

EntailE: Introducing Textual Entailment in Commonsense Knowledge Graph Completion

Commonsense knowledge graph completion is a new challenge for commonsense knowledge graph construction and application. In contrast to factual knowledge graphs such as Freebase and YAGO, commonsense knowledge graphs (CSKGs; e.g., ConceptNet) utilize free-form text to represent named entities, short phrases, and events as their nodes. Such a loose structure results in large and sparse CSKGs, which makes the semantic understanding of these nodes more critical for learning rich commonsense knowledge graph embedding. While current methods leverage semantic similarities to increase the graph density, the semantic plausibility of the nodes and their relations are under-explored. Previous works adopt conceptual abstraction to improve the consistency of modeling (event) plausibility, but they are not scalable enough and still suffer from data sparsity. In this paper, we propose to adopt textual entailment to find implicit entailment relations between CSKG nodes, to effectively densify the subgraph connecting nodes within the same conceptual class, which indicates a similar level of plausibility. Each node in CSKG finds its top entailed nodes using a finetuned transformer over natural language inference (NLI) tasks, which sufficiently capture textual entailment signals. The entailment relation between these nodes are further utilized to: 1) build new connections between source triplets and entailed nodes to densify the sparse CSKGs; 2) enrich the generalization ability of node representations by comparing the node embeddings with a contrastive loss. Experiments on two standard CSKGs demonstrate that our proposed framework EntailE can improve the performance of CSKG completion tasks under both transductive and inductive settings.

cs.CL↗

arXiv2024

Contrasting twisted bilayer graphene and transition metal dichalcogenides for fractional Chern insulators: an emergent gauge picture

The recent experimental discovery of the zero-field fractional Chern insulator (FCI) in twisted $\mathrm{MoTe_2}$ moir\'e superlattices has sparked immense interest in this exotic topological quantum state. The FCI has also been observed in previous experiments in magic angle twisted bilayer graphene (TBG) under a finite magnetic field of about 5 Tesla. Generally, the stabilization of FCI requires fine-tuning the topological band to satisfy certain conditions. It would still be helpful to have an intuitive picture to understand the different behaviors in twisted $\mathrm{MoTe_2}$ and TBG. Here, we compare them through the lens of emergent gauge fields. In TBG, the system can be mapped to two Dirac fermions coupled to emergent gauge fields with opposite signs. In contrast, the twisted $\mathrm{MoTe_2}$ reduces to a hole with parabolic dispersion coupled to an emergent gauge field. This contrasting gauge structure provides a new perspective on the observed difference: the zero-field FCI is stable in $\mathrm{MoTe_2}$ but absent in TBG. Based on this understanding, we will explore potential strategies for stabilizing FCI in both moir\'e superlattices.

cond-mat.mes-hall↗

arXiv2024

PipeNet: Question Answering with Semantic Pruning over Knowledge Graphs

It is well acknowledged that incorporating explicit knowledge graphs (KGs) can benefit question answering. Existing approaches typically follow a grounding-reasoning pipeline in which entity nodes are first grounded for the query (question and candidate answers), and then a reasoning module reasons over the matched multi-hop subgraph for answer prediction. Although the pipeline largely alleviates the issue of extracting essential information from giant KGs, efficiency is still an open challenge when scaling up hops in grounding the subgraphs. In this paper, we target at finding semantically related entity nodes in the subgraph to improve the efficiency of graph reasoning with KG. We propose a grounding-pruning-reasoning pipeline to prune noisy nodes, remarkably reducing the computation cost and memory usage while also obtaining decent subgraph representation. In detail, the pruning module first scores concept nodes based on the dependency distance between matched spans and then prunes the nodes according to score ranks. To facilitate the evaluation of pruned subgraphs, we also propose a graph attention network (GAT) based module to reason with the subgraph data. Experimental results on CommonsenseQA and OpenBookQA demonstrate the effectiveness of our method.

cs.CL↗

arXiv2023

On the Discussion of Large Language Models: Symmetry of Agents and Interplay with Prompts

Two ways has been discussed to unlock the reasoning capability of a large language model. The first one is prompt engineering and the second one is to combine the multiple inferences of large language models, or the multi-agent discussion. Theoretically, this paper justifies the multi-agent discussion mechanisms from the symmetry of agents. Empirically, this paper reports the empirical results of the interplay of prompts and discussion mechanisms, revealing the empirical state-of-the-art performance of complex multi-agent mechanisms can be approached by carefully developed prompt engineering. This paper also proposes a scalable discussion mechanism based on conquer and merge, providing a simple multi-agent discussion solution with simple prompts but state-of-the-art performance.

cs.CL↗

arXiv2023

Are LLMs Rigorous Logical Reasoners? Empowering Natural Language Proof Generation by Stepwise Decoding with Contrastive Learning

Logical reasoning is a pivotal component in the field of artificial intelligence. Proof planning, particularly in contexts requiring the validation of explanation accuracy, continues to present challenges. The recent advancement of large language models (LLMs) has led to significant progress in natural language proof planning, evolving from one-stage generators to more complex three-stage systems that include additional searchers or verifiers. While these assisted methods improve the quality of generated results, they also introduce increased search efforts and computational costs. Furthermore, the generative process itself remains underexplored. In this study, we propose a stepwise decoding approach augmented by contrastive learning to address two common errors encountered during the LLM generator's decoding process. We fine-tune the language model using both vanilla and enhanced hard negatives to mitigate these decoding errors. Empirical results demonstrate the effectiveness of our strategy. Additionally, our further analysis reveals that even larger LLMs still struggle to generate rigorous logical chains.

cs.CL↗

arXiv2023

Exciton Superposition across Moir\'e States in a Semiconducting Moir\'e Superlattice

Moir\'e superlattices of semiconducting transition metal dichalcogenides (TMDCs) enable unprecedented spatial control of electron wavefunctions in an artificial lattice with periodicities more than ten times larger than that of atomic crystals, leading to emerging quantum states with fascinating electronic and optical properties. The breaking of translational symmetry further introduces a new degree of freedom inside each moir\'e unit cell: high symmetry points of energy minima called moir\'e sites, behaving as spatially separated quantum dots. The superposition of a quasiparticle wavefunction between different moir\'e sites will enable a new platform for quantum information processing but is hindered by the suppressed electron tunneling between moir\'e sites. Here we demonstrate the superposition between two moir\'e sites by constructing an angle-aligned trilayer WSe2/monolayer WS2 moir\'e heterojunction. The two moir\'e sites with energy minimum allow the formation of two different interlayer excitons, with the hole residing in either moir\'e site of the first WSe2 layer interfacing the WS2 layer and the electron in the third WSe2 layer. An external electric field can drive the hybridization of either of the interlayer excitons with the intralayer excitons in the third WSe2 layer, realizing the continuous tuning of interlayer exciton hopping between two moir\'e sites. Therefore, a superposition of the two interlayer excitons localized at different moir\'e sites can be realized, which can be resolved in the electric-field-dependent optical reflectance spectra, distinctly different from that of the natural trilayer WSe2 in which the moir\'e modulation is absent. Our study illustrates a strategy of harnessing the new moir\'e site degree of freedom for quantum information science, a new direction of twistronics.

cond-mat.mes-hall↗

arXiv2023

Quadrupolar Excitons and Hybridized Interlayer Mott Insulator in a Trilayer Moir\'e Superlattice

Transition metal dichalcogenide (TMDC) moir\'e superlattices, owing to the moir\'e flatbands and strong correlation, can host periodic electron crystals and fascinating correlated physics. The TMDC heterojunctions in the type-II alignment also enable long-lived interlayer excitons that are promising for correlated bosonic states, while the interaction is dictated by the asymmetry of the heterojunction. Here we demonstrate a new excitonic state, quadrupolar exciton, in a symmetric WSe2-WS2-WSe2 trilayer moir\'e superlattice. The quadrupolar excitons exhibit a quadratic dependence on the electric field, distinctively different from the linear Stark shift of the dipolar excitons in heterobilayers. This quadrupolar exciton stems from the hybridization of WSe2 valence moir\'e flatbands. The same mechanism also gives rise to an interlayer Mott insulator state, in which the two WSe2 layers share one hole laterally confined in one moir\'e unit cell. In contrast, the hole occupation probability in each layer can be continuously tuned via an out-of-plane electric field, reaching 100% in the top or bottom WSe2 under a large electric field, accompanying the transition from quadrupolar excitons to dipolar excitons. Our work demonstrates a trilayer moir\'e system as a new exciting playground for realizing novel correlated states and engineering quantum phase transitions.

cond-mat.mes-hall↗

arXiv2023

Many-Body Anderson Metal-Insulator Transition using Kicked Quantum Gases

Understanding the interplay of interactions and disorder in quantum transport poses long-standing scientific challenges, with many-body quantum transport phenomena in high-dimensional disordered systems remaining largely unexplored experimentally. We utilize a momentum space lattice platform using quasi-periodically kicked ultracold atomic gases to experimentally investigate many-body effects on the three-dimensional Anderson metal-insulator transition. We observe interaction-driven sub-diffusion and a divergence of delocalization onset time on approaching the many-body phase boundary. Mean-field numerical simulations are in qualitative agreement with experimental observations.

cond-mat.quant-gas↗