Large Vision-Language Models (LVLMs) have demonstrated remarkable performance across diverse tasks.Despite great success, recent studies show that LVLMs encounter substantial limitations when engaging with visual graphs.To study the reason behind these limitations, we propose VGCURE, a comprehensive benchmark covering 22 tasks for examining the fundamental graph understanding and reasoning capacities of LVLMs.Extensive evaluations conducted on 14 LVLMs reveal that LVLMs are weak in basic graph understanding and reasoning tasks, particularly those concerning relational or structurally complex information.Based on this observation, we propose a structure-aware fine-tuning framework to enhance LVLMs with structure learning abilities through three self-supervised learning tasks.Experiments validate the effectiveness of our method in improving LVLMs' performance on fundamental and downstream graph learning tasks, as well as enhancing their robustness against complex visual graphs.

The remarkable prompting capability of large language models (LLMs) offers substantial convenience to users across diverse backgrounds. Nevertheless, as the sensitive information within prompts is inevitably exposed to LLMs, caution must be exercised to preserve privacy. Among various studies, text anonymization is considered an effective approach to preventing privacy leakage in prompts through text substitution. However, existing works overemphasize privacy while overlooks preserving contextual integrity, degrading semantic consistency. To address these concerns, this paper introduces a context-sensitive prompt privacy-preserving framework, namely Adaptive Linguistic Sanitization and Anonymization (ALSA). In specific, ALSA incorporates a three-dimensional scoring mechanism to dynamically quantify the substitutability of each word within a prompt by integrating the Privacy Leakage Risk Score (PLRS), the Contextual Information Importance Score (CIIS), and the Task Relevance Score (TRS). Subsequently, a clustering technique is adopted to dynamically determine the threshold for assigning an anonymization action (i.e., Retain, Replace, Encrypt, or Delete) by balancing privacy, semantics, and task relevance. Extensive experiments on five benchmark datasets validate the superiority of ALSA over state-of-the-art baselines in terms of accuracy, privacy preservation, and semantic integrity.

Detecting small moving objects in complex backgrounds from an overhead perspective is a highly challenging task for machine vision systems. As an inspiration from nature, the avian visual system is capable of processing motion information in various complex aerial scenes, and the Retina-OT-Rt visual circuit of birds is highly sensitive to capturing the motion information of small objects from high altitudes. However, more needs to be done on small object motion detection algorithms based on the avian visual system. In this paper, we conducted mathematical description based on extensive studies of the biological mechanisms of the Retina-OT-Rt visual circuit. Based on this, we proposed a novel tectum small object motion detection neural network (TSOM). The TSOM neural network includes the retina, SGC dendritic, SGC Soma, and Rt layers, each corresponding to neurons in the visual pathway for precise topographic projection, spatial-temporal encoding, motion feature selection, and multi-directional motion integration. Extensive experiments on pigeon neurophysiological experiments and image sequence data showed that the TSOM is biologically interpretable and effective in extracting reliable small object motion features from complex high-altitude backgrounds.

Depth weighting plays an essential role in 3D potential-field inversion to counteract the decay of the kernel function of gravity or magnetic potential-fields, making the recovered results more consistent with the actual situation rather than being overly distributed near the surface. We propose a modified generalized depth weighting (MGDW) to improve the horizontal and vertical resolutions of the inversion with compact recovered results, removing the tendency to easily have a long “tail” with low-amplitudes at depth for inversions using the traditional sensitivity matrix as the depth weighting when counteracting the tendency to concentrate near the surface. The improved process is achieved by applying central burial depths of the targets to analyze the decay features of the kernel functions from the initial forward modeling theory of gravity and magnetic anomalies, according to which a generalized depth weighting based on the sensitivity matrix is constructed and then horizontal and vertical correction factors are developed to create the modified one. Here, the central burial depths are estimated by extracting extrema of the depth from extreme points (DEXP) for the potential field along the profiles, which are determined based on the delineated edges by the normalized vertical derivatives of the total horizontal derivative (NVDR-THDR) and the delineated centers by the normalized vertical derivatives of the analytic signal amplitude (NVDR-ASA). The horizontal correction factor is determined based on the normalized potential field and its normalized vertical derivative, which helps to narrow the horizontal range for inversion. The vertical correction factor is constructed based on the decay features of the field source in the upper and lower half-space regions. Synthetic examples with noisy gravity or magnetic anomalies, as well as the gravity inversion of Pako Guyot, demonstrate the robustness, effectiveness, and higher accuracy of the inversion using MGDW. By fully exploring the information of the potential field for self-constraints, the improved depth weighting for 3D gravity and magnetic inversion can be extended to recover the deep structures of seamounts, ore deposits, etc., especially in cases where limited known constraint information is available.

With the growing complexity of fact verification tasks, the concern with "thoughtful" reasoning capabilities is increasing.However, recent fact verification benchmarks mainly focus on checking a narrow scope of semantic factoids within claims and lack an explicit logical reasoning process.In this paper, we introduce CHECKWHY, a challenging dataset tailored to a novel causal fact verification task: checking the truthfulness of the causal relation within claims through rigorous reasoning steps.CHECKWHY consists of over 19K "why" claimevidence-argument structure triplets with supports, refutes, and not enough info labels.Each argument structure is composed of connected evidence, representing the reasoning process that begins with foundational evidence and progresses toward claim establishment.Through extensive experiments on state-of-the-art models, we validate the importance of incorporating the argument structure for causal fact verification.Moreover, the automated and human evaluation of argument structure generation reveals the difficulty in producing satisfying argument structure by fine-tuned models or Chainof-Thought prompted LLMs, leaving considerable room for future improvements 1 .

The opaqueness of the multi-hop fact verification model imposes imperative requirements for explainability. One feasible way is to extract rationales, a subset of inputs, where the performance of prediction drops dramatically when being removed. Though being explainable, most rationale extraction methods for multi-hop fact verification explore the semantic information within each piece of evidence individually, while ignoring the topological information interaction among different pieces of evidence. Intuitively, a faithful rationale bears complementary information being able to extract other rationales through the multi-hop reasoning process. To tackle such disadvantages, we cast explainable multi-hop fact verification as subgraph extraction, which can be solved based on graph convolutional network (GCN) with salience-aware graph learning. In specific, GCN is utilized to incorporate the topological interaction information among multiple pieces of evidence for learning evidence representation. Meanwhile, to alleviate the influence of noisy evidence, the salience-aware graph perturbation is induced into the message passing of GCN. Moreover, the multi-task model with three diagnostic properties of rationale is elaborately designed to improve the quality of an explanation without any explicit annotations. Experimental results on the FEVEROUS benchmark show significant gains over previous state-of-the-art methods for both rationale extraction and fact verification.

Automatic multi-hop fact verification task has gained significant attention in recent years.Despite impressive results, these well-designed models perform poorly on out-of-domain data.One possible solution is to augment the training data with counterfactuals, which are generated by minimally altering the causal features of the original data.However, current counterfactual data augmentation techniques fail to handle multi-hop fact verification due to their incapability to preserve the complex logical relationships within multiple correlated texts.In this paper, we overcome this limitation by developing a rationale-sensitive method to generate linguistically diverse and label-flipping counterfactuals while preserving logical relationships.In specific, the diverse and fluent counterfactuals are generated via an Explain-Edit-Generate architecture.Moreover, the checking and filtering modules are proposed to regularize the counterfactual data with logical relations and flipped labels.Experimental results show that the proposed approach outperforms the SOTA baselines and can generate linguistically diverse counterfactual data without disrupting their logical relationships 1 .

Most person re-identification methods are researched under various assumptions. However, viewpoint variations or occlusions are often encountered in practical scenarios. These are prone to intra-class variance. In this paper, we propose a multiscale global-aware channel attention (MGCA) model to solve this problem. It imitates the process of human visual perception, which tends to observe things from coarse to fine. The core of our approach is a multiscale structure containing two key elements: the global-aware channel attention (GCA) module for capturing the global structural information and the adaptive selection feature fusion (ASFF) module for highlighting discriminative features. Moreover, we introduce a bidirectional guided pairwise metric triplet (BPM) loss to reduce the effect of outliers. Extensive experiments on Market-1501, DukeMTMC-reID, and MSMT17, and achieve the state-of-the-art results on mAP. Especially, our approach exceeds the current best method by 2.0% on the most challenging MSMT17 dataset.