Yuping Lin, Zitao Li, Yue Xing, Pengfei He, Yingqian Cui, Yaliang Li, Bolin Ding, Jingren Zhou, Jiliang Tang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Zidi Xiong, Yuping Lin, Wenya Xie, Pengfei He, Zirui Liu, Jiliang Tang, Himabindu Lakkaraju, Zhen Xiang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Shenglai Zeng, Jiankun Zhang, Bingheng Li, Yuping Lin, Tianqi Zheng, Dante Everaert, Hanqing Lu, Hui Liu, Hui Liu, Yue Xing, Monica Xiao Cheng, Jiliang Tang. 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.

Large language models (LLMs) are susceptible to a type of attack known as jailbreaking, which misleads LLMs to output harmful contents.Although there are diverse jailbreak attack strategies, there is no unified understanding on why some methods succeed and others fail.This paper explores the behavior of harmful and harmless prompts in the LLM's representation space to investigate the intrinsic properties of successful jailbreak attacks.We hypothesize that successful attacks share some similar properties: They are effective in moving the representation of the harmful prompt towards the direction to the harmless prompts.We leverage hidden representations into the objective of existing jailbreak attacks to move the attacks along the acceptance direction, and conduct experiments to validate the above hypothesis using the proposed objective.We hope this study provides new insights into understanding how LLMs understand harmfulness information.1 * These authors contributed equally to this work.1 Our code is available at https://github.com/ yuplin2333/representation-space-jailbreak.

Summary Scene text recognition (STR) is a very popular topic in the field of computer vision, which can extract text from complex natural scenes. In this article, we propose an end‐to‐end trainable and flexible STR method based on a dual attention mechanism. The proposed method consists of four modules: a thin plate spline transformer for normalizing the original image, a Channel‐Att feature extractor for obtaining representative features, a bidirectional long short‐term memory encoder for encoding sequential context features, and a Self‐Att based decoder for predicting text labels. The results on seven different benchmark datasets IIIT, SVT, IC03, IC13, IC15, SVTP, and CUTE, show that the proposed method is comparable to 13 existing methods. Especially, the average text recognition accuracy of the proposed method is about 1.4% higher than the state‐of‐the‐art method.

This two-part paper aims to study lane change behaviors at the tactical level from an on-road perspective. Compared with longitudinal driving tasks, a lane change is more complicated because this task has more interactions with surrounding vehicles; thus, there are more potential risks during this procedure. Based on the results from Part I on an on-road vehicle trajectory collection, this part investigates lane change extraction and scene-based behavior analysis, and it has a particular focus on understanding the interactions between an ego and surrounding vehicles during the procedure. We claim that this paper provides the following novel contributions: 1) an automatic method is proposed for extracting lane change segments from a continuous driving sequence by modeling and recognizing patterns in a steering angle; 2) a lane change database at the trajectory level is generated, which reflects the interactions between an ego and the surrounding vehicles during the procedures; and 3) we present findings from analyzing lane change procedures using real-world data on the axes of both the ego's trajectory and interactions with the scene vehicles. To the authors' knowledge, this is the first lane change behavior study from an on-road perspective that addresses the vehicle interactions in real-world traffic at the trajectory level.

This two-part paper aims to study lane change behaviors at the tactical level from an on-road perspective, with a special focus on analyzing the interactions between an ego and surrounding vehicles during the procedure. Part I addresses vehicle trajectory collection, whereas Part II addresses lane change extraction and scene-based behavioral analysis. Different from the general technique of moving object detection and tracking, trajectory collection for tactical driving behavior study is required to have the properties of consistency, completeness, continuity, and accuracy. This paper proposes a system of on-road vehicle trajectory collection, where an instrumented vehicle is developed with multiple horizontal 2-D lidars that have 360° coverage. The software is developed by fitting the laser points of all lidars on a vehicle model using a coupled estimation of features and reliability along frames to achieve accurate state estimations of occluded data and robust data association in multiviewpoint sensing. The performance is investigated extensively, and a large trajectory set is developed through on-road driving at the Fourth Ring Road in Beijing for a total distance of 64 km, with more than 5700 environmental trajectories with a total length of over 19 h. The performance is demonstrated to be of high quality in terms of the required properties. To the authors' knowledge, this is the first system that is able to automatically collect all-around vehicle trajectories during on-road driving and to demonstrate good performance in providing a high-quality database for driving behavior studies from an on-road perspective that addresses vehicle interactions in real-world traffic at the trajectory level.

Aerial imagery of an urban environment is often characterized by significant occlusions, sharp edges, and textureless regions, leading to poor 3D reconstruction using conventional multi-view stereo methods. In this paper, we propose a novel approach to 3D reconstruction of urban areas from a set of uncalibrated aerial images. A very general structural prior is assumed that urban scenes consist mostly of planar surfaces oriented either in a horizontal or an arbitrary vertical orientation. In addition, most structural edges associated with such surfaces are also horizontal or vertical. These two assumptions provide powerful constraints on the underlying 3D geometry. The main contribution of this paper is to translate the two constraints on 3D structure into intra-image-column and inter-image-column constraints, respectively, and to formulate the dense reconstruction as a 2-pass Dynamic Programming problem, which is solved in complete parallel on a GPU. The result is an accurate cloud of 3D dense points of the underlying urban scene. Our algorithm completes the reconstruction of 1M points with 160 available discrete height levels in under a hundred seconds. Results on multiple datasets show that we are capable of preserving a high level of structural detail and visual quality.

We propose a method to generate a highly accurate 3D face model from a set of wide-baseline images in a weakly calibrated setup. Our approach is purely data driven, and produces faithful 3D models without any pre-defined models, unlike other statistical model-based approaches. Our results do not rely upon a critical initialization step nor parameters for optimization steps. We process 5 images (including profile views), infer the accurate poses of cameras in all views, and then infer a dense 3D face model. The quality of 3D face models depends on the accuracy of estimated head-camera motion. First, we propose to use an iterative bundle adjustment approach to remove outliers in corresponding points. Contours in the profile views are matched to provide reliable correspondences that link two opposite side of views together. For dense reconstruction, we propose to use a face-specific cylindrical representation which allows us to solve a global optimization problem for N-view dense aggregation. Profile contours are used once again to provide constraints in the optimization step. Experimental results using synthetic and real images show that our method provides accurate and stable reconstruction results on wide-baseline images. We compare our method with state of the art methods, and show that it provides significantly better results in terms of both accuracy and efficiency.

We present a 3D face reconstruction system that takes as input either one single view or several different views. Given a facial image, we first classify the facial pose into one of five predefined poses, then detect two anchor points that are then used to detect a set of predefined facial landmarks. Based on these initial steps, for a single view we apply a warping process using a generic 3D face model to build a 3D face. For multiple views, we apply sparse bundle adjustment to reconstruct 3D landmarks which are used to deform the generic 3D face model. Experimental results on the Color FERET and CMU multi-PIE databases confirm our framework is effective in creating realistic 3D face models that can be used in many computer vision applications, such as 3D face recognition at a distance.

Appearance manifolds have been one of the most powerful methods for object recognition. However, they could not be used for grayscale shape classification, particularly in three dimensions, such as classifying medical lesion volumes or galaxy images. The main cause of the difficulty is that the appearance manifolds of shape classes have entangled fibers in their embedded Euclidean space. This paper proposes a novel appearance-based method called the quotient appearance manifold mapping to untangle the fibers of the appearance manifolds. First, the quotient manifold is constructed to untangle the fiber bundles of appearance manifolds. The mapping from each point of the manifold to the quotient submanifold is then proposed to classify grayscale shapes. We show the effectiveness in grayscale 3D shape recognition using medical images.

We propose a 2D registration method for multi-modal image sequences of the retinal fundus, and a 3D metric reconstruction of near planar surface from multiple views. There are two major contributions in our paper. For 2D registration, our method produces high registration rates while accounting for large modality differences. Compared with the state of the art method [5], our approach has higher registration rate (97.2% vs. 82.31%) while the computation time is much less. This is achieved by extracting features from the edge maps of the contrast enhanced images, and performing pairwise registration by matching the features in an iterative manner, maximizing the number of matches and estimating homographies accurately. The pairwise registration result is further globally optimized by an indirect registration process. For 3D registration part, images are registered to the reference frame by transforming points via a reconstructed 3D surface. The challenge is the reconstruction of a near planar surface, in which the shallow depth makes it a quasi-degenerate case for estimating the geometry from images. Our contribution is the proposed 4-pass bundle adjustment method that gives optimal estimation of all camera poses. With accurate camera poses, the 3D surface can be reconstructed using the images associated with the cameras with the largest baseline. Compared with state of the art 3D retinal image registration methods, our approach produces better results in all image sets.

Registering consecutive images from an airborne sensor into a mosaic is an essential tool for image analysts. Strictly local methods tend to accumulate errors, resulting in distortion. We propose here to use a reference image (such as a high resolution map image) to overcome this limitation. In our approach, we register a frame in an image sequence to the map using both frame-to-frame registration and frame-to-map registration iteratively. In frame-to-frame registration, a frame is registered to its previous frame. With its previous frame been registered to the map in the previous iteration, we can derive an estimated transformation from the frame to the map. In frame-to-map registration, we warp the frame to the map by this transformation to compensate for scale and rotation difference and then perform an area based matching using mutual information to find correspondences between this warped frame and the map. These correspondences together with the correspondences in previous frames could be regarded as correspondences between the partial local mosaic and the map. By registering the partial local mosaic to the map, we derive a transformation from the frame to the map. With this two-step registration, the errors between each consecutive frames are not accumulated. We then extend our approach to synchronize multiple image sequences by tracking moving objects in each image sequence, and aligning the frames based on the object's coordinates in the reference image.

Determining the status of a runway prior to landing is essential for any aircraft, whether manned or unmanned. In this paper, we present a method that can detect moving objects on the runway from an onboard infrared camera prior to the landing phase. Since the runway is a planar surface, we first locally stabilize the sequence to automatically selected reference frames using feature points in the neighborhood of the runway. Next, we normalize the stabilized sequence to compensate for the global intensity variation caused by the gain control of the infrared camera. We then create a background model to learn an appearance model of the runway. Finally, we identify moving objects by comparing the image sequence with the background model. We have tested our system with both synthetic and real world data and show that it can detect distant moving objects on the runway. We also provide a quantitative analysis of the performance with respect to variations in size, direction and speed of the target.

Registering consecutive images from an airborne sensor into a mosaic is an essential tool for image analysts. Strictly local methods tend to accumulate errors, resulting in distortion. We propose here to use a reference image (such as a high resolution map image) to overcome this limitation. In our approach, we register a frame in an image sequence to the map using both frame-to-frame registration and frame-to-map registration iteratively. In frame-to-frame registration, a frame is registered to its previous frame. With its previous frame been registered to the map in the previous iteration, we can derive an estimated transformation from the frame to the map. In frame-to-map registration, we warp the frame to the map by this transformation to compensate for scale and rotation difference and then perform an area based matching using mutual information to find correspondences between this warped frame and the map. From these correspondences, we derive a transformation that further registers the warped frame to the map. With this two-step registration, the errors between each consecutive frames are not accumulated. We present results on real image sequences from a hot air balloon