Search

Vision-Language Models (VLMs), built on pre-trained vision encoders and large language models (LLMs), have shown exceptional multi-modal understanding and dialog capabilities, positioning them as catalysts for the next technological revolution. However, while most VLM research focuses on enhancing multi-modal interaction, the risks of data misuse and leakage have been largely unexplored. This prompts the need for a comprehensive investigation of such risks in VLMs. In this paper, we conduct the first analysis of misuse and leakage detection in VLMs through the lens of membership inference attack (MIA). In specific, we focus on the instruction tuning data of VLMs, which is more likely to contain sensitive or unauthorized information. To address the limitation of existing MIA methods, we introduce a novel approach that infers membership based on a set of samples and their sensitivity to temperature, a unique parameter in VLMs. Based on this, we propose four membership inference methods, each tailored to different levels of background knowledge, ultimately arriving at the most challenging scenario. Our comprehensive evaluations show that these methods can accurately determine membership status, e.g., achieving an AUC greater than 0.8 targeting a small set consisting of only 5 samples on LLaVA., Comment: Accepted by USENIX'25; 22 pages, 28 figures

It has long been a problem to arrange and execute irregular workloads on massively parallel devices. We propose a general framework for statically batching irregular workloads into a single kernel with a runtime task mapping mechanism on GPUs. We further apply this framework to Mixture-of-Experts (MoE) model inference and implement an optimized and efficient CUDA kernel. Our MoE kernel achieves up to 91% of the peak Tensor Core throughput on NVIDIA H800 GPU and 95% on NVIDIA H20 GPU., Comment: 11 pages

It is well known that the intrinsic anomalous Hall effect (AHE) arises from the integration of the non-zero Berry curvature (BC), conventionally observed in the Dirac/Weyl and nodal-line semimetals. Moreover, nodal surface semimetals are expected to exhibit more significant BC under the prevalence of degenerate points near the Fermi level. In this work, we report the detection of a giant AHE in the Kagome magnet Fe$_3$Ge with a two-dimensional (2D) nodal surface (NS) at $k_{z}=\pi$ plane, exhibiting an anomalous Hall conductivity (AHC) of 1500 $\Omega^{-1}$cm$^{-1}$ at 160 K, the highest among all reported Kagome topological materials. This finding suggests a new platform for searching large AHC materials and facilitates potential room-temperature applications in spintronic devices and quantum computing.

This paper presents the development of a comprehensive dataset capturing interactions between Autonomous Vehicles (AVs) and traffic control devices, specifically traffic lights and stop signs. Derived from the Waymo Motion dataset, our work addresses a critical gap in the existing literature by providing real-world trajectory data on how AVs navigate these traffic control devices. We propose a methodology for identifying and extracting relevant interaction trajectory data from the Waymo Motion dataset, incorporating over 37,000 instances with traffic lights and 44,000 with stop signs. Our methodology includes defining rules to identify various interaction types, extracting trajectory data, and applying a wavelet-based denoising method to smooth the acceleration and speed profiles and eliminate anomalous values, thereby enhancing the trajectory quality. Quality assessment metrics indicate that trajectories obtained in this study have anomaly proportions in acceleration and jerk profiles reduced to near-zero levels across all interaction categories. By making this dataset publicly available, we aim to address the current gap in datasets containing AV interaction behaviors with traffic lights and signs. Based on the organized and published dataset, we can gain a more in-depth understanding of AVs' behavior when interacting with traffic lights and signs. This will facilitate research on AV integration into existing transportation infrastructures and networks, supporting the development of more accurate behavioral models and simulation tools.

The Markov property serves as a foundational assumption in most existing work on vehicle driving behavior, positing that future states depend solely on the current state, not the series of preceding states. This study validates the Markov properties of vehicle trajectories for both Autonomous Vehicles (AVs) and Human-driven Vehicles (HVs). A statistical method used to test whether time series data exhibits Markov properties is applied to examine whether the trajectory data possesses Markov characteristics. t test and F test are additionally introduced to characterize the differences in Markov properties between AVs and HVs. Based on two public trajectory datasets, we investigate the presence and order of the Markov property of different types of vehicles through rigorous statistical tests. Our findings reveal that AV trajectories generally exhibit stronger Markov properties compared to HV trajectories, with a higher percentage conforming to the Markov property and lower Markov orders. In contrast, HV trajectories display greater variability and heterogeneity in decision-making processes, reflecting the complex perception and information processing involved in human driving. These results have significant implications for the development of driving behavior models, AV controllers, and traffic simulation systems. Our study also demonstrates the feasibility of using statistical methods to test the presence of Markov properties in driving trajectory data.

In this work, we carry out a systematic investigation of the coherence peak in unconventional superconductors as they transition into the superconducting phase at $T_c$. Using $d$-wave cuprates as an example, we reveal the presence of a coherence peak below $T_c$ in the charge channel. The nuclear quadrupole relaxation rate is shown to be an effective method for detecting this unconventional coherence peak, with the superconducting coherence factor playing a pivotal role in its emergence. Additionally, we explore the influence of correlation effects, which further enhance this phenomenon. Extending our analysis, we demonstrate the existence of a similar coherence peak in ultrasonic attenuation and iron-based superconductors. Our findings offer a fresh perspective on probing superconducting gap symmetry in unconventional superconductors., Comment: 5 pages, 3 figures

The Hebel-Slichter coherence peak, observed in the spin-lattice relaxation rate $1/T_1$ just below the critical temperature $T_{\rm c}$, serves as a crucial experimental validation of the Bardeen-Cooper-Schrieffer pairing symmetry in conventional superconductors. However, no coherence peak in $1/T_1$ has been observed in unconventional superconductors like cuprates. In this study, an unconventional coherence peak is identified for the first time using nuclear quadrupole resonance on YBa$_2$Cu$_4$O$_8$, pointing to a distinctive pairing symmetry. The spin-lattice relaxation rate in nuclear quadrupole resonance and nuclear magnetic resonance with nuclear spin $I>1/2$ comprises the magnetic relaxation rate $1/T_{1}^{\rm mag}$, which probes magnetic fluctuations, and the quadrupole relaxation rate $1/T_{1}^{\rm quad}$, which probes charge fluctuations. By utilizing $^{63}$Cu and $^{65}$Cu isotopes, we successfully distinguish $1/T_{1}^{\rm mag}$ and $1/T_{1 }^{\rm quad}$ of YBa$_2$Cu$_4$O$_8$ and reveal the presence of the coherence peak in $1/T_{1 }^{\rm quad}$ but not in $1/T_{1}^{\rm mag}$, in contrast to conventional superconductors. Our finding demonstrates that unconventional superconductors do not exhibit a coherence peak in $1/T_{1}$ when the relaxation is due to fluctuations of the hyperfine field. Conversely, a coherence peak is expected when the relaxation is caused by electric field gradient fluctuations, due to the different coherence factors between charge and magnetic fluctuations. Our successful measurements of $1/T_{1}$ for the chains of YBa$_2$Cu$_4$O$_8$ suggest that, should the conditions for predominant quadrupole relaxation be satisfied, this phenomenon could provide a novel approach to exploring the unconventional nature of the pairing mechanism in other superconductors.

Recently, preference optimization methods such as DPO have significantly enhanced large language models (LLMs) in wide tasks including dialogue and question-answering. However, current methods fail to account for the varying difficulty levels of training samples during preference optimization, leading to mediocre performance in tasks with high accuracy requirements, particularly in mathematical reasoning. To address this limitation, we propose a novel training framework, which employs multiple sampling to analyze output distributions, assign different weights to samples, and incorporate these weights into the preference optimization process. This plug-and-play approach enables LLMs to prioritize challenging examples during training, improving learning efficiency. Experimental results demonstrate that our framework integrates seamlessly with various preference optimization methods and achieves consistent improvements in mathematical reasoning tasks., Comment: 12 pages, 9 figures

The co-existence of terrestrial and non-terrestrial networks (NTNs) is essential for achieving comprehensive global coverage in sixth-generation cellular networks. Given the escalating demand for spectrum, there is an ongoing global discourse on the feasibility of sharing certain frequencies currently utilized by terrestrial networks (TNs) with NTNs. However, this sharing leads to co-channel interference and subsequent performance degradation. This paper specifically investigates the interference caused by NTNs on TNs in the S-band and its relationship with the relative position between satellite and TN user equipment. We analyzed the transmission mechanisms of satellite signals and employed the ITU two-state model for our interference analysis. Through simulations, we evaluated the interference intensity at different separation distances and slant ranges. Our findings reveal that the angle between the user equipment direction and the sub-satellite point direction from the beam center significantly influences the interference level. Furthermore, we determine the minimum separation distance needed to keep the interference-to-noise ratio of NTN interference below 0 dB., Comment: 9 pages

Can Large Language Models (LLMs) accurately predict election outcomes? While LLMs have demonstrated impressive performance in healthcare, legal analysis, and creative applications, their capabilities in election forecasting remain uncertain. Notably, election prediction poses unique challenges: limited voter-level data, evolving political contexts, and the complexity of modeling human behavior. In the first part of this paper, we explore and introduce a multi-step reasoning framework for election prediction, which systematically integrates demographic, ideological, and time-sensitive factors. Validated on 2016 and 2020 real-world data and extensive synthetic personas, our approach adapts to changing political landscapes, reducing bias and significantly improving predictive accuracy. We further apply our pipeline to the 2024 U.S. presidential election, illustrating its ability to generalize beyond observed historical data. Beyond enhancing accuracy, the second part of the paper provides insights into the broader implications of LLM-based election forecasting. We identify potential political biases embedded in pretrained corpora, examine how demographic patterns can become exaggerated, and suggest strategies for mitigating these issues. Together, this project, a large-scale LLM empirical study, advances the accuracy of election predictions and establishes directions for more balanced, transparent, and context-aware modeling in political science research and practice., Comment: arXiv admin note: substantial text overlap with arXiv:2411.03321

The 3D silicon sensor has demonstrated excellent performances (signal collection, detection efficiency, power consumption, etc.) comparable or even better with respect to the traditional planar sensor of the ATLAS Detector at the Large Hadron Collider (LHC), especially after the high irradiation fluence, mainly due to the shorter drift length of the generated carriers. These characteristics have made it the most attractive technology for the detection and track reconstruction of charged particles for the High Energy Physics (HEP). In addition, its application is also being explored in astronomy, microdosimetry and medical imaging. This paper will present the design and fabrication of a novel 3D-Trench sensor which features an enclosed deep trench surrounding the central columnar cathode. This novel sensor has been fabricated on the 8-inch COMS pilot line at the Institute of Microelectronics of the Chinese Academy of Sciences (IMECAS) where ultra-narrow etch width of 0.5 {\mu}m and the ultra-high depth-to-width ratio (aspect ratio) (>70) have been achieved. Its preliminary simulation and characterization results including electrostatic potential, electric field, Current-Voltage (IV), Capacitance-Voltage (CV), Charge Collection Efficiency (CCE) and Timing Performance before irradiation will be presented in this paper.

Extensive research has been conducted to explore the capability of Large Language Models (LLMs) for table reasoning and has significantly improved the performance on existing benchmarks. However, tables and user questions in real-world applications are more complex and diverse, presenting an unignorable gap compared to the existing benchmarks. To fill the gap, we propose a \textbf{M}ult\textbf{i}-scale spreadsheet benchmark with \textbf{M}eta \textbf{o}perations for \textbf{Table} reasoning, named as MiMoTable. Specifically, MiMoTable incorporates two key features. First, the tables in MiMoTable are all spreadsheets used in real-world scenarios, which cover seven domains and contain different types. Second, we define a new criterion with six categories of meta operations for measuring the difficulty of each question in MiMoTable, simultaneously as a new perspective for measuring the difficulty of the existing benchmarks. Experimental results show that Claude-3.5-Sonnet achieves the best performance with 77.4\% accuracy, indicating that there is still significant room to improve for LLMs on MiMoTable. Furthermore, we grade the difficulty of existing benchmarks according to our new criteria. Experiments have shown that the performance of LLMs decreases as the difficulty of benchmarks increases, thereby proving the effectiveness of our proposed new criterion., Comment: Accepted by COLING 2025

Textual-based prompt learning methods primarily employ multiple learnable soft prompts and hard class tokens in a cascading manner as text prompt inputs, aiming to align image and text (category) spaces for downstream tasks. However, current training is restricted to aligning images with predefined known categories and cannot be associated with unknown categories. In this work, we propose utilizing universal attributes as a bridge to enhance the alignment between images and unknown categories. Specifically, we introduce an Attribute-embedded Textual Prompt learning method for vision-language models, named ATPrompt. This approach expands the learning space of soft prompts from the original one-dimensional category level into the multi-dimensional attribute level by incorporating multiple universal attribute tokens into the learnable soft prompts. Through this modification, we transform the text prompt from a category-centric form to an attribute-category hybrid form. To finalize the attributes for downstream tasks, we propose a differentiable attribute search method that learns to identify representative and suitable attributes from a candidate pool summarized by a large language model. As an easy-to-use plug-in technique, ATPrompt can seamlessly replace the existing prompt format of textual-based methods, offering general improvements at a negligible computational cost. Extensive experiments on 11 datasets demonstrate the effectiveness of our method., Comment: Technical Report. Project Page: https://zhengli97.github.io/ATPrompt/

Upon absorbing a photon, the ionized electron sails through the target force field in attoseconds to reach free space. This navigation probes details of the potential landscape that get imprinted into the phase of the ionization amplitude. The Eisenbud-Wigner-Smith (EWS) time delay, the energy derivative of this phase, provides the navigation time relative to the time of the electron's ``free'' exit. This time is influenced by the diffraction of the electron from the potential landscape, offering structural and dynamical information about interactions. If the potential has an intrinsic symmetry, a regular pattern in the time delay, including subpatterns of delays and advances, may occur from the diffraction process. The recent synthesis of a polyhedral fluorocarbon instigates the current study of photoionization from a cubic molecule. Our simulation of the EWS delay unravels rich diffraction motifs within $\pm$100 attoseconds in both energy and angular distributions. Averaging over the Euler angles from the laboratory to the molecular frame and over the photoelectron azimuthal direction indicates that the pattern should be discernible in ultrafast chronoscopy. The study benchmarks diffraction in molecular photoionization as a fundamental process which can be experimentally accessed through ultrafast time delay.

Optical resonators have shown outstanding abilities to tailor chemical landscapes through enhanced light-matter interaction between confined optical modes and molecule vibrations. We propose a theoretical model to study cooperative vibrational strong coupling in an open quantum system. The non-equilibrium stochastic molecular dynamics in an optical cavity with an auxiliary ensemble is investigated. It shows that coupling with a cavity mode introduces an additional colored noise and a negative feedback, both of which enable control over thermalization rates (i.e. the lifetime of excitation) of reactive molecules. Our work offers a pathway to steer stability of chemical bonds for chemical reactivity under cooperative vibrational strong coupling.

This paper presents a novel framework for modeling and conditional generation of 3D articulated objects. Troubled by flexibility-quality tradeoffs, existing methods are often limited to using predefined structures or retrieving shapes from static datasets. To address these challenges, we parameterize an articulated object as a tree of tokens and employ a transformer to generate both the object's high-level geometry code and its kinematic relations. Subsequently, each sub-part's geometry is further decoded using a signed-distance-function (SDF) shape prior, facilitating the synthesis of high-quality 3D shapes. Our approach enables the generation of diverse objects with high-quality geometry and varying number of parts. Comprehensive experiments on conditional generation from text descriptions demonstrate the effectiveness and flexibility of our method., Comment: impl. repo: https://github.com/ShuYuMo2003/ArtFormer

Physics-Informed Neural Networks (PINN) has evolved into a powerful tool for solving partial differential equations, which has been applied to various fields such as energy, environment, en-gineering, etc. When utilizing PINN to solve partial differential equations, it is common to rely on Automatic Differentiation (AD) to compute the residuals of the governing equations. This can lead to certain precision losses, thus affecting the accuracy of the network prediction. This paper pro-poses a Finite Volume Physics-Informed Neural Network (FV-PINN), designed to address steady-state problems of incompressible flow. This method divides the solution domain into mul-tiple grids. Instead of calculating the residuals of the Navier-Stokes equations at collocation points within the grid, as is common in traditional PINNs, this approach evaluates them at Gaussian in-tegral points on the grid boundaries using Gauss's theorem. The loss function is constructed using the Gaussian integral method, and the differentiation order for velocity is reduced. To validate the effectiveness of this approach, we predict the velocity and pressure fields for two typical examples in fluid topology optimization. The results are compared with commercial software COMSOL, which indicates that FVI-PINN significantly improves the prediction accuracy of both the velocity and pressure fields while accelerating the training speed of the network.

Estimating causal effects from nonexperimental data is a fundamental problem in many fields of science. A key component of this task is selecting an appropriate set of covariates for confounding adjustment to avoid bias. Most existing methods for covariate selection often assume the absence of latent variables and rely on learning the global network structure among variables. However, identifying the global structure can be unnecessary and inefficient, especially when our primary interest lies in estimating the effect of a treatment variable on an outcome variable. To address this limitation, we propose a novel local learning approach for covariate selection in nonparametric causal effect estimation, which accounts for the presence of latent variables. Our approach leverages testable independence and dependence relationships among observed variables to identify a valid adjustment set for a target causal relationship, ensuring both soundness and completeness under standard assumptions. We validate the effectiveness of our algorithm through extensive experiments on both synthetic and real-world data.

We address the issue of the testability of instrumental variables derived from observational data. Most existing testable implications are centered on scenarios where the treatment is a discrete variable, e.g., instrumental inequality (Pearl, 1995), or where the effect is assumed to be constant, e.g., instrumental variables condition based on the principle of independent mechanisms (Burauel, 2023). However, treatments can often be continuous variables, such as drug dosages or nutritional content levels, and non-constant effects may occur in many real-world scenarios. In this paper, we consider an additive nonlinear, non-constant effects model with unmeasured confounders, in which treatments can be either discrete or continuous, and propose an Auxiliary-based Independence Test (AIT) condition to test whether a variable is a valid instrument. We first show that if the candidate instrument is valid, then the AIT condition holds. Moreover, we illustrate the implications of the AIT condition and demonstrate that, in certain conditions, AIT conditions are necessary and sufficient to detect all invalid IVs. We also extend the AIT condition to include covariates and introduce a practical testing algorithm. Experimental results on both synthetic and three different real-world datasets show the effectiveness of our proposed condition.

The $tW\bar{b}$ production contributes to the real corrections to the $tW$ cross section. It would interfere with the top quark pair production, causing difficulties in a clear definition of the $tW{\bar b}$ events. The subtraction of the $t\bar{t}$ contributions has been performed in the diagram removal or diagram subtraction schemes for the tree-level processes. However, these schemes rely on the ability to identify the double resonant diagrams and thus can not be extended to loop diagrams. We propose a new scheme to subtract the $t\bar{t}$ contributions by power expansion of the squared amplitude in the resonant region. In order to cancel the infra-red divergences of the loop amplitudes, a widely used method is to introduce the dipole counter-terms, an ingredient in calculations of the full next-to-leading order QCD corrections. In our scheme, these counter-terms are also power-expanded. As a proof of principle, we calculate the one-loop correction to the $d\bar{d}\to \bar{b}Wt$ process, and present the invariant mass distribution of the $W\bar{b}$ system., Comment: 21 pages, 8 figures

In the past two years, large language models (LLMs) have achieved rapid development and demonstrated remarkable emerging capabilities. Concurrently, with powerful semantic understanding and reasoning capabilities, LLMs have significantly empowered the rapid advancement of the recommendation system field. Specifically, in news recommendation (NR), systems must comprehend and process a vast amount of clicked news text to infer the probability of candidate news clicks. This requirement exceeds the capabilities of traditional NR models but aligns well with the strengths of LLMs. In this paper, we propose a novel NR algorithm to reshape the news model via LLM Embedding and Co-Occurrence Pattern (LECOP). On one hand, we fintuned LLM by contrastive learning using large-scale datasets to encode news, which can fully explore the semantic information of news to thoroughly identify user preferences. On the other hand, we explored multiple co-occurrence patterns to mine collaborative information. Those patterns include news ID co-occurrence, Item-Item keywords co-occurrence and Intra-Item keywords co-occurrence. The keywords mentioned above are all generated by LLM. As far as we know, this is the first time that constructing such detailed Co-Occurrence Patterns via LLM to capture collaboration. Extensive experiments demonstrate the superior performance of our proposed novel method

Online shopping is a complex multi-task, few-shot learning problem with a wide and evolving range of entities, relations, and tasks. However, existing models and benchmarks are commonly tailored to specific tasks, falling short of capturing the full complexity of online shopping. Large Language Models (LLMs), with their multi-task and few-shot learning abilities, have the potential to profoundly transform online shopping by alleviating task-specific engineering efforts and by providing users with interactive conversations. Despite the potential, LLMs face unique challenges in online shopping, such as domain-specific concepts, implicit knowledge, and heterogeneous user behaviors. Motivated by the potential and challenges, we propose Shopping MMLU, a diverse multi-task online shopping benchmark derived from real-world Amazon data. Shopping MMLU consists of 57 tasks covering 4 major shopping skills: concept understanding, knowledge reasoning, user behavior alignment, and multi-linguality, and can thus comprehensively evaluate the abilities of LLMs as general shop assistants. With Shopping MMLU, we benchmark over 20 existing LLMs and uncover valuable insights about practices and prospects of building versatile LLM-based shop assistants. Shopping MMLU can be publicly accessed at https://github.com/KL4805/ShoppingMMLU. In addition, with Shopping MMLU, we host a competition in KDD Cup 2024 with over 500 participating teams. The winning solutions and the associated workshop can be accessed at our website https://amazon-kddcup24.github.io/., Comment: NeurIPS 2024 Datasets and Benchmarks Track Accepted. Modified typos in Figure 9

Short-form video content has become increasingly popular and influential in recent years. Its concise yet engaging format aligns well with todays' fast-paced and on-the-go lifestyles, making it a dominating trend in the digital world. As one of the front runners in the short video platform space, ByteDance has been highly successful in delivering a one-of-a-kind short video experience and attracting billions of users worldwide. One key contributing factor is its advanced end-to-end personalized short video playback technology, where we pioneered and developed the new technical field over the past five years to optimize user experience. This paper introduces the major concepts and methodologies of this personalized video playback technology that distinguish it from traditional multimedia technologies. More details, including goal setting, iterative process, modeling, experimental methods and required supporting systems, are also provided to encourage deeper research in this area.

Can Large Language Models (LLMs) accurately predict election outcomes? While LLMs have demonstrated impressive performance in various domains, including healthcare, legal analysis, and creative tasks, their ability to forecast elections remains unknown. Election prediction poses unique challenges, such as limited voter-level data, rapidly changing political landscapes, and the need to model complex human behavior. To address these challenges, we introduce a multi-step reasoning framework designed for political analysis. Our approach is validated on real-world data from the American National Election Studies (ANES) 2016 and 2020, as well as synthetic personas generated by the leading machine learning framework, offering scalable datasets for voter behavior modeling. To capture temporal dynamics, we incorporate candidates' policy positions and biographical details, ensuring that the model adapts to evolving political contexts. Drawing on Chain of Thought prompting, our multi-step reasoning pipeline systematically integrates demographic, ideological, and time-dependent factors, enhancing the model's predictive power., Comment: This research is ongoing work. Xiyang Hu and Yue Zhao are the corresponding authors

The ability of large language models (LLMs) to execute complex instructions is essential for their real-world applications. However, several recent studies indicate that LLMs struggle with challenging instructions. In this paper, we propose Evolutionary Contrastive Distillation (ECD), a novel method for generating high-quality synthetic preference data designed to enhance the complex instruction-following capability of language models. ECD generates data that specifically illustrates the difference between a response that successfully follows a set of complex instructions and a response that is high-quality, but nevertheless makes some subtle mistakes. This is done by prompting LLMs to progressively evolve simple instructions to more complex instructions. When the complexity of an instruction is increased, the original successful response to the original instruction becomes a "hard negative" response for the new instruction, mostly meeting requirements of the new instruction, but barely missing one or two. By pairing a good response with such a hard negative response, and employing contrastive learning algorithms such as DPO, we improve language models' ability to follow complex instructions. Empirically, we observe that our method yields a 7B model that exceeds the complex instruction-following performance of current SOTA 7B models and is competitive even with open-source 70B models.

Large vision-language models (LVLMs) have been rapidly developed and widely used in various fields, but the (potential) stereotypical bias in the model is largely unexplored. In this study, we present a pioneering measurement framework, $\texttt{ModSCAN}$, to $\underline{SCAN}$ the stereotypical bias within LVLMs from both vision and language $\underline{Mod}$alities. $\texttt{ModSCAN}$ examines stereotypical biases with respect to two typical stereotypical attributes (gender and race) across three kinds of scenarios: occupations, descriptors, and persona traits. Our findings suggest that 1) the currently popular LVLMs show significant stereotype biases, with CogVLM emerging as the most biased model; 2) these stereotypical biases may stem from the inherent biases in the training dataset and pre-trained models; 3) the utilization of specific prompt prefixes (from both vision and language modalities) performs well in reducing stereotypical biases. We believe our work can serve as the foundation for understanding and addressing stereotypical bias in LVLMs., Comment: Accepted in EMNLP 2024. 29 pages, 22 figures

We present the first complete next-to-next-to-next-to-leading-order calculation of the matching coefficients that link unpolarized flavor non-singlet parton distribution functions with lattice QCD computable correlation functions. By using this high-order result, we notice a reduction in theoretical uncertainties compared to relying solely on previously known lower-order matching coefficients. Furthermore, based on this result we have extracted the three-loop unpolarized flavor non-singlet splitting function, which is in agreement with the state-of-the-art result. Due to the simplicity of our method, it has the potential to advance the calculation of splitting functions to the desired four-loop order., Comment: 12 pages, 2 figures

In modern healthcare, the demand for autonomous robotic assistants has grown significantly, particularly in the operating room, where surgical tasks require precision and reliability. Robotic scrub nurses have emerged as a promising solution to improve efficiency and reduce human error during surgery. However, challenges remain in terms of accurately grasping and handing over surgical instruments, especially when dealing with complex or difficult objects in dynamic environments. In this work, we introduce a novel robotic scrub nurse system, RoboNurse-VLA, built on a Vision-Language-Action (VLA) model by integrating the Segment Anything Model 2 (SAM 2) and the Llama 2 language model. The proposed RoboNurse-VLA system enables highly precise grasping and handover of surgical instruments in real-time based on voice commands from the surgeon. Leveraging state-of-the-art vision and language models, the system can address key challenges for object detection, pose optimization, and the handling of complex and difficult-to-grasp instruments. Through extensive evaluations, RoboNurse-VLA demonstrates superior performance compared to existing models, achieving high success rates in surgical instrument handovers, even with unseen tools and challenging items. This work presents a significant step forward in autonomous surgical assistance, showcasing the potential of integrating VLA models for real-world medical applications. More details can be found at https://robonurse-vla.github.io.

Prompt learning has surfaced as an effective approach to enhance the performance of Vision-Language Models (VLMs) like CLIP when applied to downstream tasks. However, current learnable prompt tokens are primarily used for the single phase of adapting to tasks (i.e., adapting prompt), easily leading to overfitting risks. In this work, we propose a novel Cascade Prompt Learning CasPL framework to enable prompt learning to serve both generic and specific expertise (i.e., boosting and adapting prompt) simultaneously. Specifically, CasPL is a new learning paradigm comprising two distinct phases of learnable prompts: the first boosting prompt is crafted to extract domain-general knowledge from a senior larger CLIP teacher model by aligning their predicted logits using extensive unlabeled domain images. The second adapting prompt is then cascaded with the frozen first set to fine-tune the downstream tasks, following the approaches employed in prior research. In this manner, CasPL can effectively capture both domain-general and task-specific representations into explicitly different gradual groups of prompts, thus potentially alleviating overfitting issues in the target domain. It's worth noting that CasPL serves as a plug-and-play module that can seamlessly integrate into any existing prompt learning approach. CasPL achieves a significantly better balance between performance and inference speed, which is especially beneficial for deploying smaller VLM models in resource-constrained environments. Compared to the previous state-of-the-art method PromptSRC, CasPL shows an average improvement of 1.85% for base classes, 3.44% for novel classes, and 2.72% for the harmonic mean over 11 image classification datasets. Code is publicly available at: https://github.com/megvii-research/CasPL., Comment: ECCV2024

Tooth segmentation is a critical technology in the field of medical image segmentation, with applications ranging from orthodontic treatment to human body identification and dental pathology assessment. Despite the development of numerous tooth image segmentation models by researchers, a common shortcoming is the failure to account for the challenges of blurred tooth boundaries. Dental diagnostics require precise delineation of tooth boundaries. This paper introduces an innovative tooth segmentation network that integrates boundary information to address the issue of indistinct boundaries between teeth and adjacent tissues. This network's core is its boundary feature extraction module, which is designed to extract detailed boundary information from high-level features. Concurrently, the feature cross-fusion module merges detailed boundary and global semantic information in a synergistic way, allowing for stepwise layer transfer of feature information. This method results in precise tooth segmentation. In the most recent STS Data Challenge, our methodology was rigorously tested and received a commendable overall score of 0.91. When compared to other existing approaches, this score demonstrates our method's significant superiority in segmenting tooth boundaries., Comment: MICCAI workshop,see https://link.springer.com/book/9783031723957

Dimension reduction techniques, such as Sufficient Dimension Reduction (SDR), are indispensable for analyzing high-dimensional datasets. This paper introduces a novel SDR method named Principal Square Response Forward Regression (PSRFR) for estimating the central subspace of the response variable Y, given the vector of predictor variables $\bm{X}$. We provide a computational algorithm for implementing PSRFR and establish its consistency and asymptotic properties. Monte Carlo simulations are conducted to assess the performance, efficiency, and robustness of the proposed method. Notably, PSRFR exhibits commendable performance in scenarios where the variance of each component becomes increasingly dissimilar, particularly when the predictor variables follow an elliptical distribution. Furthermore, we illustrate and validate the effectiveness of PSRFR using a real-world dataset concerning wine quality. Our findings underscore the utility and reliability of the PSRFR method in practical applications of dimension reduction for high-dimensional data analysis.

Adapting Large Language Models (LLMs) to specific tasks introduces concerns about computational efficiency, prompting an exploration of efficient methods such as In-Context Learning (ICL). However, the vulnerability of ICL to privacy attacks under realistic assumptions remains largely unexplored. In this work, we present the first membership inference attack tailored for ICL, relying solely on generated texts without their associated probabilities. We propose four attack strategies tailored to various constrained scenarios and conduct extensive experiments on four popular large language models. Empirical results show that our attacks can accurately determine membership status in most cases, e.g., 95\% accuracy advantage against LLaMA, indicating that the associated risks are much higher than those shown by existing probability-based attacks. Additionally, we propose a hybrid attack that synthesizes the strengths of the aforementioned strategies, achieving an accuracy advantage of over 95\% in most cases. Furthermore, we investigate three potential defenses targeting data, instruction, and output. Results demonstrate combining defenses from orthogonal dimensions significantly reduces privacy leakage and offers enhanced privacy assurances., Comment: To Appear in the ACM Conference on Computer and Communications Security, October 14-18, 2024

The transfer of population between two intersecting quantum states is the most fundamental dynamical event that governs a broad variety of processes in physics, chemistry, biology and material science. Whereas any two-state description implies that population leaving one state instantaneously appears in the other state, we show that coupling to additional states, present in all real-world systems, can cause a measurable delay in population transfer. Using attosecond spectroscopy supported by advanced quantum-chemical calculations, we measure a delay of 1.46$\pm$0.41 fs at a charge-transfer state crossing in CF$_3$I$^+$, where an electron hole moves from the fluorine atoms to iodine. Our measurements also fully resolve the other fundamental quantum-dynamical processes involved in the charge-transfer reaction: a vibrational rearrangement time of 9.38$\pm$0.21 fs (during which the vibrational wave packet travels to the state crossing) and a population-transfer time of 2.3-2.4 fs. Our experimental results and theoretical simulations show that delays in population transfer readily appear in otherwise-adiabatic reactions and are typically on the order of 1 fs for intersecting molecular valence states. These results have implications for many research areas, such as atomic and molecular physics, charge transfer or light harvesting., Comment: 18 pages, 5 figures, article

The co-existence of terrestrial and non-terrestrial networks (NTNs) is essential for achieving global coverage in sixth-generation cellular networks. Due to increasing spectrum demand, there is discussion in the world level to share some frequencies used in terrestrial Networks (TNs) with NTNs, resulting in co-channel interference and performance degradation. This paper analyzes the interference caused by satellite networks on TN in the S-band. We examined the transmission mechanisms of satellite signals and conducted simulations to evaluate interference intensity across varying slant ranges. Our findings indicate that the angle between the user equipment direction and the sub-satellite point direction from the beam center significantly impacts the interference level., Comment: 3 pages, 6 figures, JWOC conference

Despite being prevalent in the general field of Natural Language Processing (NLP), pre-trained language models inherently carry privacy and copyright concerns due to their nature of training on large-scale web-scraped data. In this paper, we pioneer a systematic exploration of such risks associated with pre-trained language encoders, specifically focusing on the membership leakage of pre-training data exposed through downstream models adapted from pre-trained language encoders-an aspect largely overlooked in existing literature. Our study encompasses comprehensive experiments across four types of pre-trained encoder architectures, three representative downstream tasks, and five benchmark datasets. Intriguingly, our evaluations reveal, for the first time, the existence of membership leakage even when only the black-box output of the downstream model is exposed, highlighting a privacy risk far greater than previously assumed. Alongside, we present in-depth analysis and insights toward guiding future researchers and practitioners in addressing the privacy considerations in developing pre-trained language models., Comment: ECAI24

Fine-tuning pre-trained models for downstream tasks has led to a proliferation of open-sourced task-specific models. Recently, Model Merging (MM) has emerged as an effective approach to facilitate knowledge transfer among these independently fine-tuned models. MM directly combines multiple fine-tuned task-specific models into a merged model without additional training, and the resulting model shows enhanced capabilities in multiple tasks. Although MM provides great utility, it may come with security risks because an adversary can exploit MM to affect multiple downstream tasks. However, the security risks of MM have barely been studied. In this paper, we first find that MM, as a new learning paradigm, introduces unique challenges for existing backdoor attacks due to the merging process. To address these challenges, we introduce BadMerging, the first backdoor attack specifically designed for MM. Notably, BadMerging allows an adversary to compromise the entire merged model by contributing as few as one backdoored task-specific model. BadMerging comprises a two-stage attack mechanism and a novel feature-interpolation-based loss to enhance the robustness of embedded backdoors against the changes of different merging parameters. Considering that a merged model may incorporate tasks from different domains, BadMerging can jointly compromise the tasks provided by the adversary (on-task attack) and other contributors (off-task attack) and solve the corresponding unique challenges with novel attack designs. Extensive experiments show that BadMerging achieves remarkable attacks against various MM algorithms. Our ablation study demonstrates that the proposed attack designs can progressively contribute to the attack performance. Finally, we show that prior defense mechanisms fail to defend against our attacks, highlighting the need for more advanced defense., Comment: To appear in ACM Conference on Computer and Communications Security (CCS), 2024

Masked Image Modeling (MIM) has achieved significant success in the realm of self-supervised learning (SSL) for visual recognition. The image encoder pre-trained through MIM, involving the masking and subsequent reconstruction of input images, attains state-of-the-art performance in various downstream vision tasks. However, most existing works focus on improving the performance of MIM.In this work, we take a different angle by studying the pre-training data privacy of MIM. Specifically, we propose the first membership inference attack against image encoders pre-trained by MIM, which aims to determine whether an image is part of the MIM pre-training dataset. The key design is to simulate the pre-training paradigm of MIM, i.e., image masking and subsequent reconstruction, and then obtain reconstruction errors. These reconstruction errors can serve as membership signals for achieving attack goals, as the encoder is more capable of reconstructing the input image in its training set with lower errors. Extensive evaluations are conducted on three model architectures and three benchmark datasets. Empirical results show that our attack outperforms baseline methods. Additionally, we undertake intricate ablation studies to analyze multiple factors that could influence the performance of the attack.

Machine learning (ML), driven by prominent paradigms such as centralized and federated learning, has made significant progress in various critical applications ranging from autonomous driving to face recognition. However, its remarkable success has been accompanied by various attacks. Recently, the model hijacking attack has shown that ML models can be hijacked to execute tasks different from their original tasks, which increases both accountability and parasitic computational risks. Nevertheless, thus far, this attack has only focused on centralized learning. In this work, we broaden the scope of this attack to the federated learning domain, where multiple clients collaboratively train a global model without sharing their data. Specifically, we present HijackFL, the first-of-its-kind hijacking attack against the global model in federated learning. The adversary aims to force the global model to perform a different task (called hijacking task) from its original task without the server or benign client noticing. To accomplish this, unlike existing methods that use data poisoning to modify the target model's parameters, HijackFL searches for pixel-level perturbations based on their local model (without modifications) to align hijacking samples with the original ones in the feature space. When performing the hijacking task, the adversary applies these cloaks to the hijacking samples, compelling the global model to identify them as original samples and predict them accordingly. We conduct extensive experiments on four benchmark datasets and three popular models. Empirical results demonstrate that its attack performance outperforms baselines. We further investigate the factors that affect its performance and discuss possible defenses to mitigate its impact.

Multi-turn dialogues are a key interaction method between humans and Large Language Models (LLMs), as conversations extend over multiple rounds, keeping LLMs' high generation quality and low latency is a challenge. Mainstream LLMs can be grouped into two categories based on masking strategy: causal LLM and prefix LLM. Several works have demonstrated that prefix LLMs tend to outperform causal ones in scenarios that heavily depend on historical context such as multi-turn dialogues or in-context learning, thanks to their bidirectional attention on prefix sequences. However, prefix LLMs have an inherent inefficient training problem in multi-turn dialogue datasets. In addition, the attention mechanism of prefix LLM makes it unable to reuse Key-Value Cache (KV Cache) across dialogue rounds to reduce generation latency. In this paper, we propose a novel masking scheme called Intermittent Semi-working Mask (ISM) to address these problems. Specifically, we apply alternate bidirectional and unidirectional attention on queries and answers in the dialogue history. In this way, ISM is able to maintain the high quality of prefix LLM and low generation latency of causal LLM, simultaneously. Extensive experiments illustrate that our ISM achieves significant performance.

Recent advancements have significantly improved automated task-solving capabilities using autonomous agents powered by large language models (LLMs). However, most LLM-based agents focus on dialogue, programming, or specialized domains, leaving their potential for addressing generative AI safety tasks largely unexplored. In this paper, we propose Atlas, an advanced LLM-based multi-agent framework targeting generative AI models, specifically focusing on jailbreak attacks against text-to-image (T2I) models with built-in safety filters. Atlas consists of two agents, namely the mutation agent and the selection agent, each comprising four key modules: a vision-language model (VLM) or LLM brain, planning, memory, and tool usage. The mutation agent uses its VLM brain to determine whether a prompt triggers the T2I model's safety filter. It then collaborates iteratively with the LLM brain of the selection agent to generate new candidate jailbreak prompts with the highest potential to bypass the filter. In addition to multi-agent communication, we leverage in-context learning (ICL) memory mechanisms and the chain-of-thought (COT) approach to learn from past successes and failures, thereby enhancing Atlas's performance. Our evaluation demonstrates that Atlas successfully jailbreaks several state-of-the-art T2I models equipped with multi-modal safety filters in a black-box setting. Additionally, Atlas outperforms existing methods in both query efficiency and the quality of generated images. This work convincingly demonstrates the successful application of LLM-based agents in studying the safety vulnerabilities of popular text-to-image generation models. We urge the community to consider advanced techniques like ours in response to the rapidly evolving text-to-image generation field.

Beyond conventional ferromagnetism and antiferromagnetism, altermagnetism is a recently discovered unconventional magnetic phase characterized by time-reversal symmetry breaking and spin-split band structures in materials with zero net magnetization. This distinct magnetic phase not only enriches the understanding of fundamental physical concepts but also has profound impacts on condense-matter physics research and practical device applications. Spin-polarized band structures have been recently observed in semiconductors MnTe and MnTe2 with vanishing net magnetization, confirming the existence of this unconventional magnetic order. Metallic altermagnets have unique advantages for exploring novel physical phenomena related to low-energy quasiparticle excitations and for applications in spintronics as electrical conductivity in metals allows the direct manipulation of spin current through electric field. Here, through comprehensive characterization and analysis of the magnetic and electronic structures of KV2Se2O, we have unambiguously demonstrated a metallic room-temperature altermaget with d-wave spin-momentum locking. The highly anisotropic spin-polarized Fermi surfaces and the spin-density-wave order emerging in the altermagnetic phase make it an extraordinary platform for designing high-performance spintronic devices and studying many-body effects coupled with the unconventional magnetism., Comment: 25 pages, 7 figures

Document-level relation extraction (DocRE) aims to extract relations between entities from unstructured document text. Compared to sentence-level relation extraction, it requires more complex semantic understanding from a broader text context. Currently, some studies are utilizing logical rules within evidence sentences to enhance the performance of DocRE. However, in the data without provided evidence sentences, researchers often obtain a list of evidence sentences for the entire document through evidence retrieval (ER). Therefore, DocRE suffers from two challenges: firstly, the relevance between evidence and entity pairs is weak; secondly, there is insufficient extraction of complex cross-relations between long-distance multi-entities. To overcome these challenges, we propose GEGA, a novel model for DocRE. The model leverages graph neural networks to construct multiple weight matrices, guiding attention allocation to evidence sentences. It also employs multi-scale representation aggregation to enhance ER. Subsequently, we integrate the most efficient evidence information to implement both fully supervised and weakly supervised training processes for the model. We evaluate the GEGA model on three widely used benchmark datasets: DocRED, Re-DocRED, and Revisit-DocRED. The experimental results indicate that our model has achieved comprehensive improvements compared to the existing SOTA model.

Reasoning encompasses two typical types: deductive reasoning and inductive reasoning. Despite extensive research into the reasoning capabilities of Large Language Models (LLMs), most studies have failed to rigorously differentiate between inductive and deductive reasoning, leading to a blending of the two. This raises an essential question: In LLM reasoning, which poses a greater challenge - deductive or inductive reasoning? While the deductive reasoning capabilities of LLMs, (i.e. their capacity to follow instructions in reasoning tasks), have received considerable attention, their abilities in true inductive reasoning remain largely unexplored. To investigate into the true inductive reasoning capabilities of LLMs, we propose a novel framework, SolverLearner. This framework enables LLMs to learn the underlying function (i.e., $y = f_w(x)$), that maps input data points $(x)$ to their corresponding output values $(y)$, using only in-context examples. By focusing on inductive reasoning and separating it from LLM-based deductive reasoning, we can isolate and investigate inductive reasoning of LLMs in its pure form via SolverLearner. Our observations reveal that LLMs demonstrate remarkable inductive reasoning capabilities through SolverLearner, achieving near-perfect performance with ACC of 1 in most cases. Surprisingly, despite their strong inductive reasoning abilities, LLMs tend to relatively lack deductive reasoning capabilities, particularly in tasks involving ``counterfactual'' reasoning.

Sufficient dimension reduction (SDR) is a valuable approach for handling high-dimensional data. Outer Product Gradient (OPG) is an popular approach. However, because of focusing the mean regression function, OPG may ignore some directions of central subspace (CS) when the distribution of errors is symmetric about zero. The mode of a distribution can provide an important summary of data. A Local Modal OPG (LMOPG) and its algorithm through mode regression are proposed to estimate the basis of CS with skew errors distribution. The estimator shows the consistent and asymptotic normal distribution under some mild conditions. Monte Carlo simulation is used to evaluate the performance and demonstrate the efficiency and robustness of the proposed method.

Most existing membership inference attacks (MIAs) utilize metrics (e.g., loss) calculated on the model's final state, while recent advanced attacks leverage metrics computed at various stages, including both intermediate and final stages, throughout the model training. Nevertheless, these attacks often process multiple intermediate states of the metric independently, ignoring their time-dependent patterns. Consequently, they struggle to effectively distinguish between members and non-members who exhibit similar metric values, particularly resulting in a high false-positive rate. In this study, we delve deeper into the new membership signals in the black-box scenario. We identify a new, more integrated membership signal: the Pattern of Metric Sequence, derived from the various stages of model training. We contend that current signals provide only partial perspectives of this new signal: the new one encompasses both the model's multiple intermediate and final states, with a greater emphasis on temporal patterns among them. Building upon this signal, we introduce a novel attack method called Sequential-metric based Membership Inference Attack (SeqMIA). Specifically, we utilize knowledge distillation to obtain a set of distilled models representing various stages of the target model's training. We then assess multiple metrics on these distilled models in chronological order, creating distilled metric sequence. We finally integrate distilled multi-metric sequences as a sequential multiformat and employ an attention-based RNN attack model for inference. Empirical results show SeqMIA outperforms all baselines, especially can achieve an order of magnitude improvement in terms of TPR @ 0.1% FPR. Furthermore, we delve into the reasons why this signal contributes to SeqMIA's high attack performance, and assess various defense mechanisms against SeqMIA., Comment: Accepted by ACM CCS 2024

Personality is a fundamental construct in psychology, reflecting an individual's behavior, thinking, and emotional patterns. Previous researches have made some progress in personality detection, primarily by utilizing the whole text to predict personality. However, these studies generally tend to overlook psychological knowledge: they rarely apply the well-established correlations between emotion regulation and personality. Based on this, we propose a new personality detection method called EERPD. This method introduces the use of emotion regulation, a psychological concept highly correlated with personality, for personality prediction. By combining this feature with emotion features, it retrieves few-shot examples and provides process CoTs for inferring labels from text. This approach enhances the understanding of LLM for personality within text and improves the performance in personality detection. Experimental results demonstrate that EERPD significantly enhances the accuracy and robustness of personality detection, outperforming previous SOTA by 15.05/4.29 in average F1 on the two benchmark datasets.

Improving user experience and providing personalized search results in E-commerce platforms heavily rely on understanding purchase intention. However, existing methods for acquiring large-scale intentions bank on distilling large language models with human annotation for verification. Such an approach tends to generate product-centric intentions, overlook valuable visual information from product images, and incurs high costs for scalability. To address these issues, we introduce MIND, a multimodal framework that allows Large Vision-Language Models (LVLMs) to infer purchase intentions from multimodal product metadata and prioritize human-centric ones. Using Amazon Review data, we apply MIND and create a multimodal intention knowledge base, which contains 1,264,441 million intentions derived from 126,142 co-buy shopping records across 107,215 products. Extensive human evaluations demonstrate the high plausibility and typicality of our obtained intentions and validate the effectiveness of our distillation framework and filtering mechanism. Additional experiments reveal that our obtained intentions significantly enhance large language models in two intention comprehension tasks., Comment: EMNLP 2024 main conference

Enhancing Language Models' (LMs) ability to understand purchase intentions in E-commerce scenarios is crucial for their effective assistance in various downstream tasks. However, previous approaches that distill intentions from LMs often fail to generate meaningful and human-centric intentions applicable in real-world E-commerce contexts. This raises concerns about the true comprehension and utilization of purchase intentions by LMs. In this paper, we present IntentionQA, a double-task multiple-choice question answering benchmark to evaluate LMs' comprehension of purchase intentions in E-commerce. Specifically, LMs are tasked to infer intentions based on purchased products and utilize them to predict additional purchases. IntentionQA consists of 4,360 carefully curated problems across three difficulty levels, constructed using an automated pipeline to ensure scalability on large E-commerce platforms. Human evaluations demonstrate the high quality and low false-negative rate of our benchmark. Extensive experiments across 19 language models show that they still struggle with certain scenarios, such as understanding products and intentions accurately, jointly reasoning with products and intentions, and more, in which they fall far behind human performances. Our code and data are publicly available at https://github.com/HKUST-KnowComp/IntentionQA., Comment: Findings of EMNLP 2024

This work investigates efficient score-based black-box adversarial attacks with a high Attack Success Rate (\textbf{ASR}) and good generalizability. We design a novel attack method based on a hierarchical DIsentangled Feature space, called \textbf{DifAttack++}, which differs significantly from the existing ones operating over the entire feature space. Specifically, DifAttack++ firstly disentangles an image's latent feature into an Adversarial Feature (\textbf{AF}) and a Visual Feature (\textbf{VF}) via an autoencoder equipped with our specially designed Hierarchical Decouple-Fusion (\textbf{HDF}) module, where the AF dominates the adversarial capability of an image, while the VF largely determines its visual appearance. We train such two autoencoders for the clean and adversarial image domains (i.e., cross-domain) respectively to achieve image reconstructions and feature disentanglement, by using pairs of clean images and their Adversarial Examples (\textbf{AE}s) generated from available surrogate models via white-box attack methods. Eventually, in the black-box attack stage, DifAttack++ iteratively optimizes the AF according to the query feedback from the victim model until a successful AE is generated, while keeping the VF unaltered. Extensive experimental results demonstrate that our DifAttack++ leads to superior ASR and query efficiency than state-of-the-art methods, meanwhile exhibiting much better visual quality of AEs. The code is available at https://github.com/csjunjun/DifAttack.git., Comment: arXiv admin note: substantial text overlap with arXiv:2309.14585 An extension of the AAAI24 paper "DifAttack: Query-Efficient Black-Box Attack via Disentangled Feature Space."