What Your Features Reveal: Data-Efficient Black-Box Feature Inversion Attack for Split DNNs

Split DNNs enable edge devices by offloading intensive computation to a cloud server, but this paradigm exposes privacy vulnerabilities, as the intermediate features can be exploited to reconstruct the private inputs via Feature Inversion Attack (FIA). Existing FIA methods often produce limited reconstruction quality, making it difficult to assess the true extent of privacy leakage. To reveal the privacy risk of the leaked features, we introduce FIA-Flow, a black-box FIA framework that achieves high-fidelity image reconstruction from intermediate features. To exploit the semantic information within intermediate features, we design a Latent Feature Space Alignment Module (LFSAM) to bridge the semantic gap between the intermediate feature space and the latent space. Furthermore, to rectify distributional mismatch, we develop Deterministic Inversion Flow Matching (DIFM), which projects off-manifold features onto the target manifold with one-step inference. This decoupled design simplifies learning and enables effective training with few image-feature pairs. To quantify privacy leakage from a human perspective, we also propose two metrics based on a large vision-language model. Experiments show that FIA-Flow achieves more faithful and semantically aligned feature inversion across various models (AlexNet, ResNet, Swin Transformer, DINO, and YOLO11) and layers, revealing a more severe privacy threat in Split DNNs than previously recognized.

Analysis on Energy Efficiency of RIS-Assisted Multiuser Downlink Near-Field Communications

In this paper, we focus on the energy efficiency (EE) optimization and analysis of reconfigurable intelligent surface (RIS)-assisted multiuser downlink near-field communications. Specifically, we conduct a comprehensive study on several key factors affecting EE performance, including the number of RIS elements, the types of reconfigurable elements, reconfiguration resolutions, and the maximum transmit power. To accurately capture the power characteristics of RISs, we adopt more practical power consumption models for three commonly used reconfigurable elements in RISs: PIN diodes, varactor diodes, and radio frequency (RF) switches. These different elements may result in RIS systems exhibiting significantly different energy efficiencies (EEs), even when their spectral efficiencies (SEs) are similar. Considering discrete phases implemented at most RISs in practice, which makes their optimization NP-hard, we develop a nested alternating optimization framework to maximize EE, consisting of an outer integer-based optimization for discrete RIS phase reconfigurations and a nested non-convex optimization for continuous transmit power allocation within each iteration. Extensive comparisons with multiple benchmark schemes validate the effectiveness and efficiency of the proposed framework. Furthermore, based on the proposed optimization method, we analyze the EE performance of RISs across different key factors and identify the optimal RIS architecture yielding the highest EE.

Dependency Structure Augmented Contextual Scoping Framework for Multimodal Aspect-Based Sentiment Analysis

Multimodal Aspect-Based Sentiment Analysis (MABSA) seeks to extract fine-grained information from image-text pairs to identify aspect terms and determine their sentiment polarity. However, existing approaches often fall short in simultaneously addressing three core challenges: Sentiment Cue Perception (SCP), Multimodal Information Misalignment (MIM), and Semantic Noise Elimination (SNE). To overcome these limitations, we propose DASCO (\textbf{D}ependency Structure \textbf{A}ugmented \textbf{Sco}ping Framework), a fine-grained scope-oriented framework that enhances aspect-level sentiment reasoning by leveraging dependency parsing trees. First, we designed a multi-task pretraining strategy for MABSA on our base model, combining aspect-oriented enhancement, image-text matching, and aspect-level sentiment-sensitive cognition. This improved the model's perception of aspect terms and sentiment cues while achieving effective image-text alignment, addressing key challenges like SCP and MIM. Furthermore, we incorporate dependency trees as syntactic branch combining with semantic branch, guiding the model to selectively attend to critical contextual elements within a target-specific scope while effectively filtering out irrelevant noise for addressing SNE problem. Extensive experiments on two benchmark datasets across three subtasks demonstrate that DASCO achieves state-of-the-art performance in MABSA, with notable gains in JMASA (+3.1\% F1 and +5.4\% precision on Twitter2015).