Recent advancements in Latent Diffusion Models (LDMs) have revolutionized image synthesis and manipulation, raising significant concerns about data misappropriation and intellectual property infringement. While adversarial attacks have been extensively explored as a protective measure against such misuse of generative AI, current approaches are severely limited by their heavy reliance on model-specific knowledge and substantial computational costs. Drawing inspiration from the posterior collapse phenomenon observed in VAE training, we propose the Posterior Collapse Attack (PCA), a novel framework for protecting images from unauthorized manipulation. Through comprehensive theoretical analysis and empirical validation, we identify two distinct collapse phenomena during VAE inference: diffusion collapse and concentration collapse. Based on this discovery, we design a unified loss function that can flexibly achieve both types of collapse through parameter adjustment, each corresponding to different protection objectives in preventing image manipulation. Our method significantly reduces dependence on model-specific knowledge by requiring access to only the VAE encoder, which constitutes less than 4% of LDM parameters. Notably, PCA achieves prompt-invariant protection by operating on the VAE encoder before text conditioning occurs, eliminating the need for empty prompt optimization required by existing methods. This minimal requirement enables PCA to maintain adequate transferability across various VAE-based LDM architectures while effectively preventing unauthorized image editing. Extensive experiments show PCA outperforms existing techniques in protection effectiveness, computational efficiency (runtime and VRAM), and generalization across VAE-based LDM variants. Our code is available at https://github.com/ZhongliangGuo/PosteriorCollapseAttack.

In this paper, we tackle the challenge of white-box false positive adversarial attacks on contrastive loss based offline handwritten signature verification models. We propose a novel attack method that treats the attack as a style transfer between closely related but distinct writing styles. To guide the generation of deceptive images, we introduce two new loss functions that enhance the attack success rate by perturbing the Euclidean distance between the embedding vectors of the original and synthesized samples, while ensuring minimal perturbations by reducing the difference between the generated image and the original image. Our method demonstrates state-of-the-art performance in white-box attacks on contrastive loss based offline handwritten signature verification models, as evidenced by our experiments. The key contributions of this paper include a novel false positive attack method, two new loss functions, effective style transfer in handwriting styles, and superior performance in white-box false positive attacks compared to other white-box attack methods.

This paper introduces significant enhancements to RepoSim4Py and RepoSnipy, advanced semantic tools for deep analysis of software repositories. RepoSim4Py commandline toolbox now supports multi-level embedding, encompassing code, documentation, requirements, README, and comprehensive repository analysis, which enable the understanding of repository dynamics. Concurrently, RepoSnipy webbased search engine facilitates sophisticated repository similarity searches and introduces clustering based on both repository tags (topic_cluster) and code embeddings (code_cluster). We also introduce SimilarityCal, a novel binary classification model trained on these clusters, to predict and quantify repository similarities with high accuracy. These developments provide researchers and developers with powerful tools to navigate the complex landscape of software repositories, improving efficiency in software development and fostering innovation through better reuse of existing resources.

Neural style transfer (NST) generates new images by combining the style of one image with the content of another. However, unauthorized NST can exploit artwork, raising concerns about artists’ rights and motivating the development of proactive protection methods. We propose Locally Adaptive Adversarial Color Attack (LAACA), empowering artists to protect their artwork from unauthorized style transfer by processing before public release. By delving into the intricacies of human visual perception and the role of different frequency components, our method strategically introduces frequency-adaptive perturbations in the image. These perturbations significantly degrade the generation quality of NST while maintaining an acceptable level of visual change in the original image, ensuring that potential infringers are discouraged from using the protected artworks, because of its bad NST generation quality. Additionally, existing metrics often overlook the importance of color fidelity in evaluating color-mattered tasks, such as the quality of NST-generated images, which is crucial in the context of artistic works. To comprehensively assess the color-mattered tasks, we propose the Aesthetic Color Distance Metric (ACDM), designed to quantify the color difference of images pre- and post-manipulations. Experimental results confirm that attacking NST using LAACA results in visually inferior style transfer, and the ACDM can efficiently measure color-mattered tasks. By providing artists with a tool to safeguard their intellectual property, our work relieves the socio-technical challenges posed by the misuse of NST in the art community.

The rapid development of heterogeneous networks has proposed new challenges to the long-standing link prediction problem. Existing models trained on the verified edge samples from different types usually learn type-specific knowledge, and their type-specific predictions may be contradictory for unverified edge samples with uncertain types. This challenge is termed edge-type disturbance in link prediction in heterogeneous networks. To address this challenge, we develop a disturbance-resilient prediction method ( DRPM ) comprising a structural characterizer, a type differentiator, and a resilient predictor. The structural characterizer is responsible for learning edge representations for link prediction. Concurrently, the type differentiator distinguishes type-specific edge representations to generate diverse type experts while maximizing their link prediction performances on specific types. Furthermore, the resilient predictor evaluates the reliability weights of different type experts to develop a resilient prediction mechanism to aggregate discriminable predictions. Extensive experiments conducted on various real-world datasets demonstrate the importance of the explainable introduction of the edge-type disturbance and the superiority of DRPM over state-of-the-art methods.

Sensor-based human activity recognition (HAR), with the ability to recognise human activities from wearable or embedded sensors, has been playing an important role in many applications including personal health monitoring, smart home, and manufacturing. The real-world, long-term deployment of these HAR systems drives a critical research question: how to evolve the HAR model automatically over time to accommodate changes in an environment or activity patterns. This paper presents an online continual learning (OCL) scenario for HAR, where sensor data arrives in a streaming manner which contains unlabelled samples from already learnt activities or new activities. We propose a technique, OCL-HAR, making a real-time prediction on the streaming sensor data while at the same time discovering and learning new activities. We have empirically evaluated OCL-HAR on four third-party, publicly available HAR datasets. Our results have shown that this OCL scenario is challenging to state-of-the-art continual learning techniques that have significantly underperformed. Our technique OCL-HAR has consistently outperformed them in all experiment setups, leading up to 0.17 and 0.23 improvements in micro and macro F1 scores.

Heterogeneous social networks, which are characterized by diverse interaction types, have resulted in new challenges for missing link prediction. Most deep learning models tend to capture type-specific features to maximize the prediction performances on specific link types. However, the types of missing links are uncertain in heterogeneous social networks; this restricts the prediction performances of existing deep learning models. To address this issue, we propose a multi-type transferable method (MTTM) for missing link prediction in heterogeneous social networks, which exploits adversarial neural networks to remain robust against type differences. It comprises a generative predictor and a discriminative classifier. The generative predictor can extract link representations and predict whether the unobserved link is a missing link. To generalize well for different link types to improve the prediction performance, it attempts to deceive the discriminative classifier by learning transferable feature representations among link types. In order not to be deceived, the discriminative classifier attempts to accurately distinguish link types, which indirectly helps the generative predictor judge whether the learned feature representations are transferable among link types. Finally, the integratedMTTMis constructed on this minimax two-player game between the generative predictor and discriminative classifier to predict missing links based on transferable feature representations among link types. Extensive experiments show that the proposedMTTMcan outperform state-of-the-art baselines for missing link prediction in heterogeneous social networks.

Recently, dynamic social network research has attracted a great amount of attention, especially in the area of anomaly analysis that analyzes the anomalous change in the evolution of dynamic social networks. However, most of the current research focused on anomaly analysis of the macro representation of dynamic social networks and failed to analyze the nodes that have anomalous structural changes at a micro level. To identify and evaluate anomalous structural change-based nodes in generalized dynamic social networks that only have limited structural information, this research considers undirected and unweighted graphs and develops a multiple-neighbor superposition similarity method ( ), which mainly consists of a multiple-neighbor range algorithm ( ) and a superposition similarity fluctuation algorithm ( ). introduces observation nodes, characterizes the structural similarities of nodes within multiple-neighbor ranges, and proposes a new multiple-neighbor similarity index on the basis of extensional similarity indices. Subsequently, maximally reflects the structural change of each node, using a new superposition similarity fluctuation index from the perspective of diverse multiple-neighbor similarities. As a result, based on and , not only identifies anomalous structural change-based nodes by detecting the anomalous structural changes of nodes but also evaluates their anomalous degrees by quantifying these changes. Results obtained by comparing with state-of-the-art methods via extensive experiments show that can accurately identify anomalous structural change-based nodes and evaluate their anomalous degrees well.

Human activity recognition plays a significant role in enabling pervasive applications as it abstracts low-level noisy sensor data into high-level human activities, which applications can respond to. With more and more activity-aware applications deployed in real-world environments, a research challenge emerges-discovering and learning new activities that have not been pre-defined or observed in the training phase. This paper tackles this challenge by proposing a hierarchical mixture of directional statistical models. The model supports incrementally, continuously updating the activity model over time with the reduced annotation effort and without the need for storing historical sensor data. We have validated this solution on four publicly available, third-party smart home datasets, and have demonstrated up to 91.5 percent accuracies of detecting and recognising new activities.

We have witnessed an increasing number of activity-aware applications being deployed in real-world environments, including smart home and mobile healthcare. The key enabler to these applications is sensor-based human activity recognition; that is, recognising and analysing human daily activities from wearable and ambient sensors. With the power of machine learning we can recognise complex correlations between various types of sensor data and the activities being observed. However the challenges still remain: (1) they often rely on a large amount of labelled training data to build the model, and (2) they cannot dynamically adapt the model with emerging or changing activity patterns over time. To directly address these challenges, we propose a Bayesian nonparametric model, i.e. Dirichlet process mixture of conditionally independent von Mises Fisher models, to enable both unsupervised and semi-supervised dynamic learning of human activities. The Bayesian nonparametric model can dynamically adapt itself to the evolving activity patterns without human intervention and the learning results can be used to alleviate the annotation effort. We evaluate our approach against real-world, third-party smart home datasets, and demonstrate significant improvements over the state-of-the-art techniques in both unsupervised and supervised settings.