We present SLAG, a Sensitive Layer ActivationGuided jailbreak attack for Large Vision-Language Models (LVLMs). SLAG identifies safety-sensitive layers whose activations differ between harmful and benign inputs, and perturbs images using a dual-objective loss to enhance harmful generation while suppressing refusals. On MiniGPT-4, SLAG achieves a 95.0% Attack Success Rate with only 2000 steps, outperforming existing image-only attacks and approaching multimodal methods. Layer analysis shows that a few mid-to-late layers suffice, revealing multiple activation pathways linked to safety failures.

Navigating an arbitrary-shaped ground robot safely in cluttered environments remains a challenging problem. The existing trajectory planners that account for the robot’s physical geometry severely suffer from the intractable runtime. To achieve both computational efficiency and Continuous Collision Avoidance (CCA) of arbitrary-shaped ground robot planning, we proposed a novel coarse-to-fine navigation framework that significantly accelerates planning. In the first stage, a sampling-based method selectively generates distinct topological paths that guarantee a minimum inflated margin. In the second stage, a geometry-aware front-end strategy is designed to discretize these topologies into full-state robot motion sequences while concurrently partitioning the paths into SE(2) sub-problems and simpler ℝ2sub-problems for back-end optimization. In the final stage, an SVSDF-based optimizer generates trajectories tailored to these sub-problems and seamlessly splices them into a continuous final motion plan. Extensive benchmark comparisons show that the proposed method is one to several orders of magnitude faster than the cutting-edge methods in runtime while maintaining a high planning success rate and ensuring CCA.

Hybrid motor imagery brain-computer interfaces (MI-BCIs), which integrate both electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) signals, outperform those based solely on EEG. However, simultaneously recording EEG and fNIRS signals is highly challenging due to the difficulty of colocating both types of sensors on the same scalp surface. This physical constraint complicates the acquisition of high-quality hybrid signals, thereby limiting the widespread application of hybrid MI-BCIs. To address this issue, this study proposes the spatio-temporal controlled diffusion model (SCDM) as a framework for cross-modal generation from EEG to fNIRS. The model utilizes two core modules, the spatial cross-modal generation (SCG) module and the multi-scale temporal representation (MTR) module, which adaptively learn the respective latent temporal and spatial representations of both signals in a unified representation space. The SCG module further maps EEG representations to fNIRS representations by leveraging their spatial relationships. Experimental results show high similarity between synthetic and real fNIRS signals. The joint classification performance of EEG and synthetic fNIRS signals is comparable to or even better than that of EEG with real fNIRS signals. Furthermore, the synthetic signals exhibit similar spatio-temporal features to real signals while preserving spatial relationships with EEG signals. To our knowledge, it is the first work that an end-to-end framework is proposed to achieve cross-modal generation from EEG to fNIRS. Experimental results suggest that the SCDM may represent a promising paradigm for the acquisition of hybrid EEG-fNIRS signals in MI-BCI systems.

Existing forgery detection models perform well on in-distribution images, but struggle with unseen images. In this paper, we explore the performance of deepfake detection models trained on the large dataset and fine-tuned on small datasets. The experiments show that the models exhibit limited generalization capabilities, and the process of fine-tuning further decreases their ability to generalize. We believe that incorrect feature mapping is key to affecting model performance. Therefore, we tested the linear probing method on pre-trained models, which reduced the loss of generalization performance and confirmed our hypothesis. Furthermore, we proposed a fine-tuning strategy that adapts the model to different datasets on the classifier, improving the performance on both in-distribution and out-of-distribution samples.