Domain adaptive point cloud completion (DA PCC) aims to narrow the geometric and semantic discrepancies between the labeled source and unlabeled target domains. Existing methods either suffer from limited receptive fields or quadratic complexity due to using CNNs or vision Transformers. In this paper, we present the first work that studies the adaptability of state space models (SSMs) in DA PCC and find that directly applying SSMs to DA PCC will encounter several challenges: directly serializing 3D point clouds into 1D sequences often disrupts the spatial topology and local geometric features of the target domain. Besides, the overlook of designs in the learning domain-agnostic representations hinders the adaptation performance. To address these issues, we propose a novel framework, DAPointMamba for DA PCC, that exhibits strong adaptability across domains and has the advantages of global receptive fields and efficient linear complexity. It has three novel modules. In particular, Cross-Domain Patch-Level Scanning introduces patch-level geometric correspondences, enabling effective local alignment. Cross-Domain Spatial SSM Alignment further strengthens spatial consistency by modulating patch features based on cross-domain similarity, effectively mitigating fine-grained structural discrepancies. Cross-Domain Channel SSM Alignment actively addresses global semantic gaps by interleaving and aligning feature channels. Extensive experiments on both synthetic and real-world benchmarks demonstrate that our DAPointMamba outperforms state-of-the-art methods with less computational complexity and inference latency.

A primary challenge in network defense is to mine potential attack campaigns from massive, continuously arriving alerts and telemetry data in real time. This paper presents HoSig-Align I, an edge side unsupervised method designed for network streams to discover homologous blocks. Our approach innovatively fuses heterogeneous features like Internet Protocol (IP) and Payload into a unified representation while strictly excluding temporal information from it, only incorporating time via a dual time scale decay model during graph construction to capture temporal proximity. A density robust similarity is computed using an isolation style random partition forest, leading to a sparse k-Nearest Neighbors (k-NN) graph. The stream is then accurately segmented into internally cohesive and mutually isolated homologous blocks through spectral ordering and contrastive change point detection. Each block is encoded into a lightweight HoSig signature, forming the basis for cross organizational collaboration. Experiments on real network streams show that HoSig-Align I identifies coherent attack campaign blocks and improves separation and boundary clarity over baselines, while meeting low-latency and low-overhead requirements for edge processing.

Masked autoencoders (MAEs) have emerged as a powerful self-supervised approach for point cloud analysis. Nevertheless, existing methods often separately focus on global structures or multi-scale features, ignoring their complementary potential. In this paper, we propose a novel dual-Siamese masked autoencoder (DS-MAE) framework that explores integrating global and hierarchical feature learning in a unified architecture for point cloud analysis. In particular, we introduce a consistent dual-branch patch embedding strategy to partition the point cloud into patches using shared group centers, ensuring both global and hierarchical branches process point patches centered at the same spatial locations. Each branch employs dual-branch Siamese encoders to process original and augmented point patches, learning representations that capture both local details and global context. In addition, we have designed cross-attention Siamese decoders to reconstruct masked point patches and align features both within and between branches with cross-attention mechanisms. Comprehensive experiments demonstrate our method consistently achieves superior results to prior methods. Code is available at https://github.com/shaoandy1211/DS-MAE.git.

Whereas contrastive learning eliminates the need for labeled data, existing methods may suffer from inadequate features due to the conventional single shared encoder structure and struggle to fully harness the rich spectrum of 3D augmentations. In this paper, we propose TriCI, a self-supervised method that designs a triple-branch contrastive learning architecture. During contrastive pre-training, we generate three augmented versions of each input point cloud sample and pair each augmented sample with the original one, resulting in three unique positive pairs. We subsequently feed the pairs into three distinct encoders, each of which extracts features from its corresponding input positive pair. We design a novel cross-branch contrastive loss and use it along with the intra-branch contrastive loss to jointly train our network. The proposed cross-branch loss effectively aligns the output features from different perspectives for pre-training and facilitates their integration for downstream tasks, particularly in object-level scenarios. The intra-branch loss helps maximize the feature correspondences within positive pairs. Extensive experiments demonstrate the superiority of our TriCI in self-supervised learning, and show its strong ability in enhancing the performance of downstream object classification and part segmentation tasks. Interestingly, our TriCI achieves a 92.9% accuracy for linear SVM evaluation on ModelNet40, exceeding its closest competitor by 1.7% and even exceeding some supervised methods.

Existing normal estimation methods for point clouds are often less robust to severe noise and complex geometric structures. Also, they usually ignore the contributions of different neighbouring points during normal estimation, which leads to less accurate results. In this paper, we introduce a weighted normal estimation method for 3D point cloud data. We innovate in two key points: 1) we develop a novel weighted normal regression technique that predicts point-wise weights from local point patches and use them for robust, feature-preserving normal regression; 2) we propose to conduct contrastive learning between point patches and the corresponding ground-truth normals of the patches’ central points as a pre-training process to facilitate normal regression. Comprehensive experiments demonstrate that our method can robustly handle noisy and complex point clouds, achieving state-of-the-art performance on both synthetic and real-world datasets.

Abstract Pre-trained language models achieve high performance on machine reading comprehension task, but these models lack robustness and are vulnerable to adversarial samples. Most of the current methods for improving model robustness are based on data enrichment. However, these methods do not solve the problem of poor context representation of the machine reading comprehension model. We find that context representation plays a key role in the robustness of the machine reading comprehension model, dense context representation space results in poor model robustness. To deal with this, we propose a Multi-task machine Reading Comprehension learning framework via Contrastive Learning. Its main idea is to improve the context representation space encoded by the machine reading comprehension models through contrastive learning. This special contrastive learning we proposed called Contrastive Learning in Context Representation Space(CLCRS). CLCRS samples sentences containing context information from the context as positive and negative samples, expanding the distance between the answer sentence and other sentences in the context. Therefore, the context representation space of the machine reading comprehension model has been expanded. The model can better distinguish between sentence containing correct answers and misleading sentence. Thus, the robustness of the model is improved. Experiment results on adversarial datasets show that our method exceeds the comparison models and achieves state-of-the-art performance.

Intracranial aneurysms are common nowadays and how to detect them intelligently is of great significance in digital health. Whereas most existing deep learning research focused on medical images in a supervised way, we introduce an unsupervised method for the detection of intracranial aneurysms based on 3D point cloud data. In particular, our method consists of two stages: unsupervised pre-training and downstream tasks. As for the former, the main idea is to pair each point cloud with its jittering counterpart and maximise their correspondence. Then we design a dual-branch contrastive network with an encoder for each branch and a subsequent common projection head. As for the latter, we design simple networks for supervised classification and segmentation training. Experiments on the public dataset (IntrA) show that our unsupervised method achieves comparable or even better performance than some state-of-the-art supervised techniques, and it is most prominent in the detection of aneurysmal vessels. Experiments on the ModelNet-40 also show that our method achieves the accuracy of 90.79% which outperforms existing state-of-the-art unsupervised models.

Human emotion recognition has been widely used and intensively studied in many areas such as e-Commerce, online education, healthcare, human-computer interaction and recently human-centric software engineering (HCSE). HCSE investigates human factors in the entire software development lifecycle, and human emotion can be used in many scenarios such as requirement gathering and usability testing. However, even though existing studies have already shown the advantages of emotion recognition with multi-source data such as text, audio and video, the current research and practice in HCSE are primarily based on single-source data. In addition, emotion recognition in HCSE faces several challenges such as multiple participants, changing environments and real-time requirement. To tackle these challenges, this paper proposes Edge4Emotion, a novel edge computing-based multisource human emotion recognition platform for HCSE. Edge4Emotion takes the advantage of the edge computing paradigm to efficiently support the collection of multi-source data such as audio, video and physiological data from various IoT devices, and emotion recognition with both single- and multisource models. As an on-going project, this paper focuses on the platform design and the preliminary evaluation of the platform with representative emotion recognition applications. The platform will be further extended to include more multi-source learning models and serve as an open-source platform for the development and evaluation of multi-source emotion recognition models for HCSE.

Abstract Nowadays, unmanned aerial vehicles (UAVs) are widely used in many smart systems such as smart logistics, smart agriculture, and environmental monitoring systems. However, the limited computing capability and restricted battery lifetime of existing UAVs could significantly impact the quality of service (QoS) of UAV‐based smart systems and the quality of experience (QoE) of end users. Recently, Mobile Edge Computing (MEC) which provisions computing resources close to the mobile end devices has become a promising solution. However, since high‐speed UAV often flies through the signal range of the different edge nodes, the interruption of services in the MEC‐based UAV delivery system is a critical issue. A challenging question is when and how to perform dynamic task migration among the edge nodes to ensure service continuity. In this paper, we investigate the task migration issue for multiple UAVs in the MEC‐based UAV delivery system. Specifically, we propose an energy‐aware decision‐making strategy for the dynamic task migration named GAD to optimize the UAV energy consumption. Given the real‐time system status and QoS constraints, and through a dynamic two‐tier decision‐making mechanism, GAD can efficiently make the task migration decision from four candidate decisions, viz. No Migration, Data Migration Only, Cold Migration, and Live Migration. Experimental results based on a real‐world scenario show that our strategy can well outperform other baseline strategies in various metrics including the flying distances and the energy consumption of UAVs.

Recognition of human behaviours including body motions and facial expressions plays a significant role in human-centric software engineering. However, due to the data and computation intensive nature of human behaviour recognition through video analytics, expensive powerful machines are often required, which could hinder the research and application in human-centric software engineering. To address such an issue, this paper proposes a cost-effective human behaviour recognition system named Edge4Real which can be easily deployed in an edge computing environment with commodity machines. Compared with existing centralised solutions, Edge4Real has three major advantages including cost-effectiveness, easy-to-use, and real-time. Specifically, Edge4Real adopts a distributed architecture where components such as motion capturing, human behaviour recognition, data decoding and extraction, and the application of the recognition result, can be deployed on separated end devices and edge nodes in an edge computing environment. Using a virtual reality application which can capture a user's motion and translate into the motion of a 3D avatar in real time, we successfully validate the effectiveness of the system and demonstrate its promising value to the research and application of human-centric software engineering. The demo video can be found at https://youtu.be/tnEshD8j-kA.

Abstract Background Despite the great advance of protein structure prediction, accurate prediction of the structures of mainly β proteins is still highly challenging, but could be assisted by the knowledge of residue-residue pairing in β strands. Previously, we proposed a ridge-detection-based algorithm RDb2C that adopted a multi-stage random forest framework to predict the β-β pairing given the amino acid sequence of a protein. Results In this work, we developed a second version of this algorithm, RDb2C2, by employing the residual neural network to further enhance the prediction accuracy. In the benchmark test, this new algorithm improves the F1-score by > 10 percentage points, reaching impressively high values of ~ 72% and ~ 73% in the BetaSheet916 and BetaSheet1452 sets, respectively. Conclusion Our new method promotes the prediction accuracy of β-β pairing to a new level and the prediction results could better assist the structure modeling of mainly β proteins. We prepared an online server of RDb2C2 at http://structpred.life.tsinghua.edu.cn/rdb2c2.html .