Achieving physical-world attacks on object detectors often relies on adversarial camouflage, which applies textures to target surfaces. However, existing methods typically simulate environmental variations through post-rendering image transformations, which do not fully account for the 3D object’s geometry and its interaction with lighting. To address this, we propose the Flexible Physical-camouflage Attack (FPA), a framework that integrates a differentiable 3D renderer with comprehensive, multi-parameter environmental randomization such as lighting, material, and viewpoint directly into the optimization loop. This ensures that the generated textures are robust to real-world variations. For texture generation, FPA leverages a denoising diffusion probabilistic model whose generative prior helps produce structurally coherent and visually realistic patterns. These are jointly optimized with a set of task-oriented loss functions including adversarial, smoothness, non-printability, and concealment constraints within the unified framework. Through systematic ablation and extensive physical experiments on a 1:24 scale model, we demonstrate that FPA achieves a high attack success rate (ASR) and strong transferability to black-box detectors. Crucially, our analysis reveals a controllable trade-off between adversarial effectiveness and visual stealth, validated by perceptual metrics and human evaluation. Our findings highlight the importance of integrated physical modeling and systematic evaluation for advancing physically realizable adversarial camouflage.

Parameter-efficient transfer learning (PETL) has emerged as an effective strategy for fine-tuning large vision–language foundation models because it sharply reduces computational and memory overhead. However, existing PETL techniques underperform on dense prediction tasks that require fine-grained multimodal reasoning, such as referring image segmentation (RIS), owing to the lack of mechanisms that simultaneously strengthen local perception and enforce precise cross-modal alignment. We present a PETL framework with two lightweight and complementary adapters. The Global–Local Integrated Adapter (GLiA) enriches intra-modal features by coupling multi-scale depthwise-separable convolutions with a lightweight self-attention layer, capturing local context without sacrificing global dependencies. The Cross-Modal Alignment Adapter (CAA) explicitly aligns textual phrases with their corresponding visual regions, bridging the semantic gap between vision and language and enhancing multimodal reasoning. Experiments on three mainstream RIS benchmarks show that MiCA achieves the best accuracy while saving numerous updated parameters compared to the full fine-tuning and other PETL methods. Notably, with only 1.93% tunable backbone parameters, MiCA improves average accuracy by 0.8% across the three benchmarks compared to the baseline model.

Black-box adversarial attacks remain challenging due to limited access to model internals. Existing methods often depend on specific network architectures or require numerous queries, resulting in limited cross-architecture transferability and high query costs. To address these limitations, we propose JAD, a latent diffusion model framework for black-box adversarial attacks. JAD generates adversarial examples by leveraging a latent diffusion model guided by attention maps distilled from both a convolutional neural network (CNN) and a Vision Transformer (ViT) models. By focusing on image regions that are commonly sensitive across architectures, this approach crafts adversarial perturbations that transfer effectively between different model types. This joint attention distillation strategy enables JAD to be architecture-agnostic, achieving superior attack generalization across diverse models. Moreover, the generative nature of the diffusion framework yields high adversarial sample generation efficiency by reducing reliance on iterative queries. Experiments demonstrate that JAD attack offers improved attack generalization, generation efficiency, and cross-architecture transferability compared to existing methods, providing a promising and effective paradigm for black-box adversarial attacks.

Unmanned aerial vehicles (UAVs) are increasingly employed to perform high-risk tasks that require minimal human intervention. However, they face escalating cybersecurity threats, particularly from GNSS spoofing attacks. While previous studies have extensively investigated the impacts of GNSS spoofing on UAVs, few have focused on its effects on specific tasks. Moreover, the influence of UAV motion states on the assessment of cybersecurity risks is often overlooked. To address these gaps, we first provide a detailed evaluation of how motion states affect the effectiveness of network attacks. We demonstrate that nonlinear motion states not only enhance the effectiveness of position spoofing in GNSS spoofing attacks but also reduce the probability of detecting speed-related attacks. Building upon this, we propose a state-triggered backdoor attack method (SSD) to deceive GNSS systems and assess its risk to trajectory planning tasks. Extensive validation of SSD’s effectiveness and stealthiness is conducted. Experimental results show that, with appropriately tuned hyperparameters, SSD significantly increases positioning errors and the risk of task failure, while maintaining high stealthy rates across three state-of-the-art detectors.

Weiqi Wang, Limeng Cui, Xin Liu, Sreyashi Nag, Wenju Xu, Chen Luo, Sheikh Muhammad Sarwar, Yang Li, Hansu Gu, Hui Liu, Changlong Yu, Jiaxin Bai, Yifan Gao, Haiyang Zhang, Qi He, Shuiwang Ji, Yangqiu Song. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Ran Xu, Hui Liu, Sreyashi Nag, Zhenwei Dai, Yaochen Xie, Xianfeng Tang, Chen Luo, Yang Li, Joyce C. Ho, Carl Yang, Qi He. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Online shopping platforms, such as Amazon, offer services to billions of people worldwide. Unlike web search or other search engines, product search engines have their unique characteristics, primarily featuring short queries which are mostly a combination of product attributes and structured product search space. The uniqueness of product search underscores the crucial importance of the query understanding component. However, there are limited studies focusing on exploring this impact within real-world product search engines. In this work, we aim to bridge this gap by conducting a comprehensive study and sharing our year-long journey investigating how the query understanding service impacts Amazon Product Search. Firstly, we explore how query understanding-based ranking features influence the ranking process. Next, we delve into how the query understanding system contributes to understanding the performance of a ranking model. Building on the insights gained from our study on the evaluation of the query understanding-based ranking model, we propose a query understanding-based multi-task learning framework for ranking. We present our studies and investigations on Amazon Search.

Abstract Random numbers are very important for the security of computer system. However, generating qualified random numbers is difficult because we cannot always successfully introduce dedicated random number hardware into computer system. Although most operating systems provide random number generation capabilities, the effective entropy supply is still dependent on the hardware platform including memory and clocks etc. However, obtaining hardware events such as clocks requires system privileges, which is not conducive for entropy estimation at the application layer. In contrast, data related to the sensor hardware can be extracted directly at the application layer. These sensor data contain some randomness and may be used as a noise source. In this way, applications can use these sensors to implement their own proprietary random number generators. Before taking these sensors as the noise source, it is necessary to fully evaluate their entropy supply capability. In this paper, 300 Android smartphones and 30 iOS smartphones are selected as samples and their sensor entropy supply capabilities are comprehensively evaluated. Based on the entropy evaluation results, we give some suggestions on how to generate random numbers using these sensor data. We first design a framework for evaluating the entropy supply capability for smartphone sensors, based on the min-entropy estimation method proposed in NIST SP 800-90B. According to this framework, we simulate stationary and mobile working states for each smartphone, and collect sufficient sensor data as the min-entropy estimation dataset. The min-entropy estimation results show that in the stationary working state, each ACCELEROMETER sensor data collection can obtain at least 1.5 bits of entropy in Android, while each GYROSCOPE sensor data collection can obtain at least 20 bits of entropy in iOS. In the mobile working state, each ACCELEROMETER sensor data collection can obtain at least 1.9 bits of entropy, while each GYROSCOPE sensor data acquisition in iOS system can obtain at least 27 bits of entropy. This means that we can still get a stable entropy output from the sensor even when the smartphone is in stationary working state. Statistical analysis of the data using cross correlation methods suggests it is hard for an attacker to guess or predict the random numbers generated by a smartphone through another smartphone put in the similar external environment.

Object detection is a fundamental task in various applications ranging from autonomous driving to intelligent security systems. However, recognition of a person can be hindered when their clothing is decorated with carefully designed graffiti patterns, leading to the failure of object detection. To achieve greater attack potential against unknown black-box models, adversarial patches capable of affecting the outputs of multiple-object detection models are required. While ensemble models have proven effective, current research in the field of object detection typically focuses on the simple fusion of the outputs of all models, with limited attention being given to developing general adversarial patches that can function effectively in the physical world. In this paper, we introduce the concept of energy and treat the adversarial patches generation process as an optimization of the adversarial patches to minimize the total energy of the “person” category. Additionally, by adopting adversarial training, we construct a dynamically optimized ensemble model. During training, the weight parameters of the attacked target models are adjusted to find the balance point at which the generated adversarial patches can effectively attack all target models. We carried out six sets of comparative experiments and tested our algorithm on five mainstream object detection models. The adversarial patches generated by our algorithm can reduce the recognition accuracy of YOLOv2 and YOLOv3 to 13.19% and 29.20%, respectively. In addition, we conducted experiments to test the effectiveness of T-shirts covered with our adversarial patches in the physical world and could achieve that people are not recognized by the object detection model. Finally, leveraging the Grad-CAM tool, we explored the attack mechanism of adversarial patches from an energetic perspective.

Deep reinforcement learning has advanced greatly and applied in many areas. In this paper, we explore the vulnerability of deep reinforcement learning by proposing a novel generative model for creating effective adversarial examples to attack the agent. Our proposed model can achieve both targeted attacks and untargeted attacks. Considering the specificity of deep reinforcement learning, we propose the action consistency ratio as a measure of stealthiness, and a new measurement index of effectiveness and stealthiness. Experiment results show that our method can ensure the effectiveness and stealthiness of attack compared with other algorithms. Moreover, our methods are considerably faster and thus can achieve rapid and efficient verification of the vulnerability of deep reinforcement learning.

Conversational sentiment analysis (CSA) and emotion-cause pair extraction (ECPE) tasks have attracted increasing attention in recent years. The former aims to predict the sentiment states of speakers in a conversation, and the latter is about extracting emotion-cause clauses in a document. However, one drawback of CSA is that it cannot model the causal reasoning among emotion and neutral utterances from different speakers. In this work, we propose a new task: emotion-cause pair extraction in conversations (ECPEC), which aims to extract pairs of emotional utterances and corresponding cause utterances in conversations. The utterance-level ECPEC task is more challenging since the distance between emotion and cause utterances is larger than that of the clause-level ECPE task. To this end, we build a novel dataset ConvECPE and propose a specifically designed two-step framework for the new ECPEC task. Experimental results on ConvECPE dataset demonstrate the feasibility of the ECPEC task as well as the effectiveness of our framework.

The existing sea–land clutter classification task of sky-wave over-the-horizon-radar (OTHR) assumes that the training data and test data are drawn from the same probability distribution. However, there is a distribution discrepancy/domain shift of the collected sea–land clutter under various working conditions of OTHR, which leads to the advanced sea–land clutter classification methods being difficult to achieve effective cross-domain classification. To solve this problem, this article proposes an improved maximum classifier discrepancy (MCD) framework, namely, triple loss adversarial domain adaptation network (TLADAN) for cross-domain sea–land clutter classification, which includes a metric-based feature-level loss, an adversarial-based instance-level loss, and an adversarial-based class-level loss. The proposed TLADAN performs feature-, instance-, and class-level alignments of the sea–land clutter from different domains, so as to learn the domain-invariant features to improve the classification performance in the cross-domain scenario. Our method is evaluated in six sea–land clutter domain adaptation (DA) scenarios. Meanwhile, state-of-the-art DA methods are selected for comparison. The experimental results validate the effectiveness and superiority of TLADAN.

In the sea-land clutter classification of sky-wave over-the-horizon-radar (OTHR), the imbalanced and scarce data leads to a poor performance of the deep learning-based classification model. To solve this problem, this paper proposes an improved auxiliary classifier generative adversarial network (AC-GAN) architecture, namely auxiliary classifier variational autoencoder generative adversarial network (AC-VAEGAN). AC-VAEGAN can synthesize higher quality sea-land clutter samples than AC-GAN and serve as an effective tool for data augmentation. Specifically, a one-dimensional convolutional AC-VAEGAN architecture is designed to synthesize sea-land clutter samples. Additionally, an evaluation method combining both traditional evaluation of GAN domain and statistical evaluation of signal domain is proposed to evaluate the quality of synthetic samples. Using a dataset of OTHR sea-land clutter, both the quality of the synthetic samples and the performance of data augmentation of AC-VAEGAN are verified. Further, the effect of AC-VAEGAN as a data augmentation method on the classification performance of imbalanced and scarce sea-land clutter samples is validated. The experiment results show that the quality of samples synthesized by AC-VAEGAN is better than those synthesized by the state-of-the-art GAN-based methods, and the data augmentation method with AC-VAEGAN is able to improve the classification performance in the case of imbalanced and scarce sea-land clutter samples.

Deep learning has unlocked new paths towards the emulation of the peculiarly-human capability of learning from examples. While this kind of bottom-up learning works well for tasks such as image classification or object detection, it is not as effective when it comes to natural language processing. Communication is much more than learning a sequence of letters and words: it requires a basic understanding of the world and social norms, cultural awareness, commonsense knowledge, etc.; all things that we mostly learn in a top-down manner. In this work, we integrate top-down and bottom-up learning via an ensemble of symbolic and subsymbolic AI tools, which we apply to the interesting problem of polarity detection from text. In particular, we integrate logical reasoning within deep learning architectures to build a new version of SenticNet, a commonsense knowledge base for sentiment analysis.

In the personal house renting scenario, customers usually make quick assessments based on previous customers' reviews, which makes such reviews essential for the business. If the house is assessed as popular, a Matthew effect will be observed as more people will be willing to book it. Due to the lack of definition and quantity assessment measures, however, it is difficult to make a popularity evaluation and prediction. To solve this problem, the concept of house popularity is well defined in this paper. Specifically, the house popularity is decided by inter-event timeand rating score at the same time. To make a more effective prediction over these two correlated variables, a dual-gated recurrent unit (DGRU) is employed. Furthermore, an encoder-decoder framework with DGRU is proposed to perform popularity prediction. Empirical results show the effectiveness of the proposed DGRU and the encoder-decoder framework in two-correlated sequences prediction and popularity prediction, respectively.

Microtext analysis is a crucial task for gauging social media opinion. In this paper, we compare four different deep learning encoder-decoder frameworks to handle microtext normalization problem. The frameworks have been evaluated on four different datasets in three different domains. To understand the impact of microtext normalization, we further integrate the framework into a sentiment classification task. This paper is the first of its kind to incorporate deep learning into a microtext normalization module and improve the sentiment analysis task. We show our models as a sequence to sequence character to word encoder-decoder model. We compare four deep learning models for microtext normalization task which further improve the accuracy of the sentiment analysis. Results show that the attentive LSTM and GRU cell both increase the sentiment analysis accuracy in the range of 4%–7% whereas LSTM and CNN with LSTM improve the accuracy in the range of 2%–4%.

The traditional machine learning algorithms are easily affected by datasets in short text classification tasks, so they have weak generalization ability when confronted with new situations. This paper presents a new method SVMCNN by combining Convolutional Neural Networks and Support Vector Machine. Training the SVMCNN model with labeled datasets, and using the collected Twitter data for classification test. The results show that the SVMCNN, especially pre-trained SVMCNN has good performance in short text classification, which gets the high Precision rate, Recall rate and F1-measure.

Vacation rental websites such as Airbnb have become increasingly popular where rentals are typically short-term and travels or vacations related. Reasonable rental prices play a crucial role in improving user experiences and engagements in these websites. However, the unique properties of their rentals challenge traditional house rentals that are often long-term and study or work related. Therefore, in this paper we investigate the novel problem of price recommendation in vacation rental websites. We identify some important factors that affect the rental prices and propose a framework that consists of Multi-Scale Affinity Propagation (MSAP) to cluster houses, Nash Equilibrium filter to remove unreasonable price and Linear Regression model with Normal Noise (LRNN) to predict the reasonable prices. Experimental results demonstrate the effectiveness of the proposed framework. We conduct further experiments to understand the important factors in rental price recommendation.