arXiv subjects

Haoran Li

Publications and source records attributed to Haoran Li.

At least 217 records · Page 12Linked to original sources

arXiv2025

PrivaCI-Bench: Evaluating Privacy with Contextual Integrity and Legal Compliance

Recent advancements in generative large language models (LLMs) have enabled wider applicability, accessibility, and flexibility. However, their reliability and trustworthiness are still in doubt, especially for concerns regarding individuals' data privacy. Great efforts have been made on privacy by building various evaluation benchmarks to study LLMs' privacy awareness and robustness from their generated outputs to their hidden representations. Unfortunately, most of these works adopt a narrow formulation of privacy and only investigate personally identifiable information (PII). In this paper, we follow the merit of the Contextual Integrity (CI) theory, which posits that privacy evaluation should not only cover the transmitted attributes but also encompass the whole relevant social context through private information flows. We present PrivaCI-Bench, a comprehensive contextual privacy evaluation benchmark targeted at legal compliance to cover well-annotated privacy and safety regulations, real court cases, privacy policies, and synthetic data built from the official toolkit to study LLMs' privacy and safety compliance. We evaluate the latest LLMs, including the recent reasoner models QwQ-32B and Deepseek R1. Our experimental results suggest that though LLMs can effectively capture key CI parameters inside a given context, they still require further advancements for privacy compliance.

cs.CL↗

arXiv2025

ExARNN: An Environment-Driven Adaptive RNN for Learning Non-Stationary Power Dynamics

Non-stationary power system dynamics, influenced by renewable energy variability, evolving demand patterns, and climate change, are becoming increasingly complex. Accurately capturing these dynamics requires a model capable of adapting to environmental factors. Traditional models, including Recurrent Neural Networks (RNNs), lack efficient mechanisms to encode external factors, such as time or environmental data, for dynamic adaptation. To address this, we propose the External Adaptive RNN (ExARNN), a novel framework that integrates external data (e.g., weather, time) to continuously adjust the parameters of a base RNN. ExARNN achieves this through a hierarchical hypernetwork design, using Neural Controlled Differential Equations (NCDE) to process external data and generate RNN parameters adaptively. This approach enables ExARNN to handle inconsistent timestamps between power and external measurements, ensuring continuous adaptation. Extensive forecasting tests demonstrate ExARNN's superiority over established baseline models.

cs.LG↗

arXiv2025

Regularized Fingerprinting with Linearly Optimal Weight Matrix in Detection and Attribution of Climate Change

Climate change detection and attribution play a central role in establishing the causal influence of human activities on global warming. The dominant framework, optimal fingerprinting, is a linear errors-in-variables model in which each covariate is subject to measurement error with covariance proportional to that of the regression error. The reliability of such analyses depends critically on accurate inference of the regression coefficients. The optimal weight matrix for estimating these coefficients is the precision matrix of the regression error, which is typically unknown and must be estimated from climate model simulations. However, existing regularized optimal fingerprinting approaches often yield underestimated uncertainties and overly narrow confidence intervals that fail to attain nominal coverage, thereby compromising the reliability of analysis. In this paper, we first propose consistent variance estimators for the regression coefficients within the class of linear shrinkage weight matrices, addressing undercoverage in conventional methods. Building on this, we derive a linearly optimal weight matrix that directly minimizes the asymptotic variances of the estimated scaling factors. Numerical studies confirm improved empirical coverage and shorter interval lengths. When applied to annual mean temperature data, the proposed method produces narrower, more reliable intervals and provides new insights into detection and attribution across different regions.

stat.ME↗

arXiv2025

Purity Law for Generalizable Neural TSP Solvers

Achieving generalization in neural approaches across different scales and distributions remains a significant challenge for the Traveling Salesman Problem~(TSP). A key obstacle is that neural networks often fail to learn robust principles for identifying universal patterns and deriving optimal solutions from diverse instances. In this paper, we first uncover Purity Law (PuLa), a fundamental structural principle for optimal TSP solutions, defining that edge prevalence grows exponentially with the sparsity of surrounding vertices. Statistically validated across diverse instances, PuLa reveals a consistent bias toward local sparsity in global optima. Building on this insight, we propose Purity Policy Optimization~(PUPO), a novel training paradigm that explicitly aligns characteristics of neural solutions with PuLa during the solution construction process to enhance generalization. Extensive experiments demonstrate that PUPO can be seamlessly integrated with popular neural solvers, significantly enhancing their generalization performance without incurring additional computational overhead during inference.

cs.LG↗

arXiv2025

Nature's Insight: A Novel Framework and Comprehensive Analysis of Agentic Reasoning Through the Lens of Neuroscience

Autonomous AI is no longer a hard-to-reach concept, it enables the agents to move beyond executing tasks to independently addressing complex problems, adapting to change while handling the uncertainty of the environment. However, what makes the agents truly autonomous? It is agentic reasoning, that is crucial for foundation models to develop symbolic logic, statistical correlations, or large-scale pattern recognition to process information, draw inferences, and make decisions. However, it remains unclear why and how existing agentic reasoning approaches work, in comparison to biological reasoning, which instead is deeply rooted in neural mechanisms involving hierarchical cognition, multimodal integration, and dynamic interactions. In this work, we propose a novel neuroscience-inspired framework for agentic reasoning. Grounded in three neuroscience-based definitions and supported by mathematical and biological foundations, we propose a unified framework modeling reasoning from perception to action, encompassing four core types, perceptual, dimensional, logical, and interactive, inspired by distinct functional roles observed in the human brain. We apply this framework to systematically classify and analyze existing AI reasoning methods, evaluating their theoretical foundations, computational designs, and practical limitations. We also explore its implications for building more generalizable, cognitively aligned agents in physical and virtual environments. Finally, building on our framework, we outline future directions and propose new neural-inspired reasoning methods, analogous to chain-of-thought prompting. By bridging cognitive neuroscience and AI, this work offers a theoretical foundation and practical roadmap for advancing agentic reasoning in intelligent systems. The associated project can be found at: https://github.com/BioRAILab/Awesome-Neuroscience-Agent-Reasoning .

q-bio.NC↗

arXiv2025

Observation of Intrinsic and LED Light-Enhanced Memristor Performance in In-Plane Ferroelectric NbOI2

Two-dimensional (2D) layered ferroelectrics, as an emerging area of research, have attracted extensive attention, while memristors based on new 2D ferroelectric materials have yet to be fully explored, thereby limiting their applications in modern nanoelectronics. In this work, we report the observation of intrinsic memristive behavior in a newly discovered 2D in-plane ferroelectric material, NbOI2, and the giant enhancement of the memristive performance using LED visible light. The results show that NbOI2 exhibits intrinsically strong memristive response with a current on/off ratio of up to 10^4 and stable switching cycles, which is independent of back-gate voltage. Under LED visible light illumination, the current on/off ratio in NbOI2 is over one order of magnitude higher than that without light, meanwhile, the coercive field is significantly reduced to less than 1.22 kVcm-1, much lower than other 2D ferroelectric material-based memristors. Interestingly, both the intrinsic and the light-enhanced resistive switching phenomena only occur along the in-plane b-axis direction, indicating that the memristive behavior in NbOI2 is driven by electric field-induced and optical field-enhanced ferroelectric polarization switching mechanisms, as evidenced by a combined orientation-dependent electrical/optoelectrical measurement and sweep cycle-induced structural evolution analysis. Our study not only provides a materials strategy based on new 2D ferroelectrics for designing memristor applications, but also offers a simple optical method to enhance its performance, paving the path for its implementation in novel nanoelectronics and optoelectronics.

cond-mat.mtrl-sci↗

arXiv2025

Transient Chirality in the Gelation of Adhesive Spinner Monolayers

Active systems of self-rotating elements inherently exhibit chirality, making them of fundamental interest due to parity violation. Using large-scale hydrodynamic simulations, we investigate the gelation of adhesive spinners confined to quasi-2D monolayers at low Reynolds numbers. Unlike the coarsening dynamics of passive colloids, spinner gelation follows a different pathway, displaying structural chirality during the early stages of aggregation. However, this chirality dissipates upon dynamical arrest, resulting in a final gel structure that resembles a conventional colloidal gel. As a result, we find no sign of odd mechanical responses. Nonetheless, the elastic modulus and gelation time remain tunable through spinning activity, providing a new avenue for the bottom-up design of programmable soft materials.

cond-mat.soft↗

arXiv2025

BaThe: Defense against the Jailbreak Attack in Multimodal Large Language Models by Treating Harmful Instruction as Backdoor Trigger

Multimodal Large Language Models (MLLMs) have showcased impressive performance in a variety of multimodal tasks. On the other hand, the integration of additional image modality may allow the malicious users to inject harmful content inside the images for jailbreaking. Unlike text-based LLMs, where adversaries need to select discrete tokens to conceal their malicious intent using specific algorithms, the continuous nature of image signals provides a direct opportunity for adversaries to inject harmful intentions. In this work, we propose $\textbf{BaThe}$ ($\textbf{Ba}$ckdoor $\textbf{T}$rigger S$\textbf{h}$i$\textbf{e}$ld), a simple yet effective jailbreak defense mechanism. Our work is motivated by recent research on jailbreak backdoor attack and virtual prompt backdoor attack in generative language models. Jailbreak backdoor attack uses harmful instructions combined with manually crafted strings as triggers to make the backdoored model generate prohibited responses. We assume that harmful instructions can function as triggers, and if we alternatively set rejection responses as the triggered response, the backdoored model then can defend against jailbreak attacks. We achieve this by utilizing virtual rejection prompt, similar to the virtual prompt backdoor attack. We embed the virtual rejection prompt into the soft text embeddings, which we call ``wedge''. Our comprehensive experiments demonstrate that BaThe effectively mitigates various types of jailbreak attacks and is adaptable to defend against unseen attacks, with minimal impact on MLLMs' performance.

cs.CR↗

arXiv2025

ConRFT: A Reinforced Fine-tuning Method for VLA Models via Consistency Policy

Vision-Language-Action (VLA) models have shown substantial potential in real-world robotic manipulation. However, fine-tuning these models through supervised learning struggles to achieve robust performance due to limited, inconsistent demonstrations, especially in contact-rich environments. In this paper, we propose a reinforced fine-tuning approach for VLA models, named ConRFT, which consists of offline and online fine-tuning with a unified consistency-based training objective, to address these challenges. In the offline stage, our method integrates behavior cloning and Q-learning to effectively extract policy from a small set of demonstrations and stabilize value estimating. In the online stage, the VLA model is further fine-tuned via consistency policy, with human interventions to ensure safe exploration and high sample efficiency. We evaluate our approach on eight diverse real-world manipulation tasks. It achieves an average success rate of 96.3% within 45-90 minutes of online fine-tuning, outperforming prior supervised methods with a 144% improvement in success rate and 1.9x shorter episode length. This work highlights the potential of integrating reinforcement learning to enhance the performance of VLA models for real-world robotic applications. Videos and code are available at our project website https://cccedric.github.io/conrft/.

cs.RO↗

arXiv2025

Sample-efficient Unsupervised Policy Cloning from Ensemble Self-supervised Labeled Videos

Current advanced policy learning methodologies have demonstrated the ability to develop expert-level strategies when provided enough information. However, their requirements, including task-specific rewards, action-labeled expert trajectories, and huge environmental interactions, can be expensive or even unavailable in many scenarios. In contrast, humans can efficiently acquire skills within a few trials and errors by imitating easily accessible internet videos, in the absence of any other supervision. In this paper, we try to let machines replicate this efficient watching-and-learning process through Unsupervised Policy from Ensemble Self-supervised labeled Videos (UPESV), a novel framework to efficiently learn policies from action-free videos without rewards and any other expert supervision. UPESV trains a video labeling model to infer the expert actions in expert videos through several organically combined self-supervised tasks. Each task performs its duties, and they together enable the model to make full use of both action-free videos and reward-free interactions for robust dynamics understanding and advanced action prediction. Simultaneously, UPESV clones a policy from the labeled expert videos, in turn collecting environmental interactions for self-supervised tasks. After a sample-efficient, unsupervised, and iterative training process, UPESV obtains an advanced policy based on a robust video labeling model. Extensive experiments in sixteen challenging procedurally generated environments demonstrate that the proposed UPESV achieves state-of-the-art interaction-limited policy learning performance (outperforming five current advanced baselines on 12/16 tasks) without exposure to any other supervision except for videos.

cs.CV↗

arXiv2025

Memory-Statistics Tradeoff in Continual Learning with Structural Regularization

We study the statistical performance of a continual learning problem with two linear regression tasks in a well-specified random design setting. We consider a structural regularization algorithm that incorporates a generalized $\ell_2$-regularization tailored to the Hessian of the previous task for mitigating catastrophic forgetting. We establish upper and lower bounds on the joint excess risk for this algorithm. Our analysis reveals a fundamental trade-off between memory complexity and statistical efficiency, where memory complexity is measured by the number of vectors needed to define the structural regularization. Specifically, increasing the number of vectors in structural regularization leads to a worse memory complexity but an improved excess risk, and vice versa. Furthermore, our theory suggests that naive continual learning without regularization suffers from catastrophic forgetting, while structural regularization mitigates this issue. Notably, structural regularization achieves comparable performance to joint training with access to both tasks simultaneously. These results highlight the critical role of curvature-aware regularization for continual learning.

cs.LG↗

arXiv2025

Cross-domain Random Pre-training with Prototypes for Reinforcement Learning

This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. Unsupervised cross-domain Reinforcement Learning (RL) pre-training shows great potential for challenging continuous visual control but poses a big challenge. In this paper, we propose \textbf{C}ross-domain \textbf{R}andom \textbf{P}re-\textbf{T}raining with \textbf{pro}totypes (CRPTpro), a novel, efficient, and effective self-supervised cross-domain RL pre-training framework. CRPTpro decouples data sampling from encoder pre-training, proposing decoupled random collection to easily and quickly generate a qualified cross-domain pre-training dataset. Moreover, a novel prototypical self-supervised algorithm is proposed to pre-train an effective visual encoder that is generic across different domains. Without finetuning, the cross-domain encoder can be implemented for challenging downstream tasks defined in different domains, either seen or unseen. Compared with recent advanced methods, CRPTpro achieves better performance on downstream policy learning without extra training on exploration agents for data collection, greatly reducing the burden of pre-training. We conduct extensive experiments across eight challenging continuous visual-control domains, including balance control, robot locomotion, and manipulation. CRPTpro significantly outperforms the next best Proto-RL(C) on 11/12 cross-domain downstream tasks with only 54.5\% wall-clock pre-training time, exhibiting state-of-the-art pre-training performance with greatly improved pre-training efficiency.

cs.LG↗

arXiv2025

Shear-based Grasp Control for Multi-fingered Underactuated Tactile Robotic Hands

This paper presents a shear-based control scheme for grasping and manipulating delicate objects with a Pisa/IIT anthropomorphic SoftHand equipped with soft biomimetic tactile sensors on all five fingertips. These `microTac' tactile sensors are miniature versions of the TacTip vision-based tactile sensor, and can extract precise contact geometry and force information at each fingertip for use as feedback into a controller to modulate the grasp while a held object is manipulated. Using a parallel processing pipeline, we asynchronously capture tactile images and predict contact pose and force from multiple tactile sensors. Consistent pose and force models across all sensors are developed using supervised deep learning with transfer learning techniques. We then develop a grasp control framework that uses contact force feedback from all fingertip sensors simultaneously, allowing the hand to safely handle delicate objects even under external disturbances. This control framework is applied to several grasp-manipulation experiments: first, retaining a flexible cup in a grasp without crushing it under changes in object weight; second, a pouring task where the center of mass of the cup changes dynamically; and third, a tactile-driven leader-follower task where a human guides a held object. These manipulation tasks demonstrate more human-like dexterity with underactuated robotic hands by using fast reflexive control from tactile sensing.

cs.RO↗

arXiv2025

Spectroscopy of electric dipole and quadrupole transitions in $^{224}$Ra$^+$

We report on spectroscopy of the low-lying electronic transitions in $^{224}$Ra$^+$. The ion's low charge to mass ratio and convenient wavelengths make $^{224}$Ra$^+$ a promising optical clock candidate. We measured the frequencies of the the $^2{S}_{1/2} \ $$\leftrightarrow$$\ ^2{P}_{1/2}$ cooling transition, the $^2{S}_{1/2}\ $$\leftrightarrow$$\ ^2{D}_{5/2}$ clock transition, the $^2{D}_{3/2} \ $$\leftrightarrow$$\ ^2{P}_{3/2}$ electric dipole transition, and the $^2{D}_{5/2} \ $$\leftrightarrow$$\ ^2{P}_{3/2}$ cleanout transition. From these measurements we calculate the frequencies of the $^2{D}_{3/2}\ $$\leftrightarrow$$\ ^2{P}_{1/2}$ repump transition, the $^2{S}_{1/2} \ $$\leftrightarrow$$\ ^2{D}_{3/2}$ electric quadrupole transition, and the $^2{S}_{1/2} \ $$\leftrightarrow$$\ ^2{P}_{3/2}$ electric dipole transition.

physics.atom-ph↗

arXiv2025

Text-to-3D Generation by 2D Editing

Distilling 3D representations from pretrained 2D diffusion models is essential for 3D creative applications across gaming, film, and interior design. Current SDS-based methods are hindered by inefficient information distillation from diffusion models, which prevents the creation of photorealistic 3D contents. In this paper, we first reevaluate the SDS approach by analyzing its fundamental nature as a basic image editing process that commonly results in over-saturation, over-smoothing, lack of rich content and diversity due to the poor-quality single-step denoising. In light of this, we then propose a novel method called 3D Generation by Editing (GE3D). Each iteration of GE3D utilizes a 2D editing framework that combines a noising trajectory to preserve the information of the input image, alongside a text-guided denoising trajectory. We optimize the process by aligning the latents across both trajectories. This approach fully exploits pretrained diffusion models to distill multi-granularity information through multiple denoising steps, resulting in photorealistic 3D outputs. Both theoretical and experimental results confirm the effectiveness of our approach, which not only advances 3D generation technology but also establishes a novel connection between 3D generation and 2D editing. This could potentially inspire further research in the field. Code and demos are released at https://jahnsonblack.github.io/GE3D/.

cs.CV↗

arXiv2025

KSOD: Knowledge Supplement for LLMs On Demand

Large Language Models (LLMs) have demonstrated remarkable capabilities in various tasks, yet still produce errors in domain-specific tasks. To further improve their performance, we propose KSOD (Knowledge Supplement for LLMs On Demand), a novel framework that empowers LLMs to improve their capabilities with knowledge-based supervised fine-tuning (SFT). KSOD analyzes the causes of errors from the perspective of knowledge deficiency by identifying potential missing knowledge in LLM that may lead to the errors. Subsequently, KSOD tunes a knowledge module on knowledge dataset and verifies whether the LLM lacks the identified knowledge based on it. If the knowledge is verified, KSOD supplements the LLM with the identified knowledge using the knowledge module. Tuning LLMs on specific knowledge instead of specific task decouples task and knowledge and our experiments on two domain-specific benchmarks and four general benchmarks empirically demonstrate that KSOD enhances the performance of LLMs on tasks requiring the supplemented knowledge while preserving their performance on other tasks. Our findings shed light on the potential of improving the capabilities of LLMs with knowledge-based SFT.

cs.CL↗

arXiv2025

Interpersonal Memory Matters: A New Task for Proactive Dialogue Utilizing Conversational History

Proactive dialogue systems aim to empower chatbots with the capability of leading conversations towards specific targets, thereby enhancing user engagement and service autonomy. Existing systems typically target pre-defined keywords or entities, neglecting user attributes and preferences implicit in dialogue history, hindering the development of long-term user intimacy. To address these challenges, we take a radical step towards building a more human-like conversational agent by integrating proactive dialogue systems with long-term memory into a unified framework. Specifically, we define a novel task named Memory-aware Proactive Dialogue (MapDia). By decomposing the task, we then propose an automatic data construction method and create the first Chinese Memory-aware Proactive Dataset (ChMapData). Furthermore, we introduce a joint framework based on Retrieval Augmented Generation (RAG), featuring three modules: Topic Summarization, Topic Retrieval, and Proactive Topic-shifting Detection and Generation, designed to steer dialogues towards relevant historical topics at the right time. The effectiveness of our dataset and models is validated through both automatic and human evaluations. We release the open-source framework and dataset at https://github.com/FrontierLabs/MapDia.

cs.CL↗

arXiv2025

MMed-RAG: Versatile Multimodal RAG System for Medical Vision Language Models

Artificial Intelligence (AI) has demonstrated significant potential in healthcare, particularly in disease diagnosis and treatment planning. Recent progress in Medical Large Vision-Language Models (Med-LVLMs) has opened up new possibilities for interactive diagnostic tools. However, these models often suffer from factual hallucination, which can lead to incorrect diagnoses. Fine-tuning and retrieval-augmented generation (RAG) have emerged as methods to address these issues. However, the amount of high-quality data and distribution shifts between training data and deployment data limit the application of fine-tuning methods. Although RAG is lightweight and effective, existing RAG-based approaches are not sufficiently general to different medical domains and can potentially cause misalignment issues, both between modalities and between the model and the ground truth. In this paper, we propose a versatile multimodal RAG system, MMed-RAG, designed to enhance the factuality of Med-LVLMs. Our approach introduces a domain-aware retrieval mechanism, an adaptive retrieved contexts selection method, and a provable RAG-based preference fine-tuning strategy. These innovations make the RAG process sufficiently general and reliable, significantly improving alignment when introducing retrieved contexts. Experimental results across five medical datasets (involving radiology, ophthalmology, pathology) on medical VQA and report generation demonstrate that MMed-RAG can achieve an average improvement of 43.8% in the factual accuracy of Med-LVLMs. Our data and code are available in https://github.com/richard-peng-xia/MMed-RAG.

cs.LG↗