The evolution from static ranking models to Agentic Recommender Systems (Agentic RecSys) empowers AI agents to maintain long-term user profiles and autonomously plan service tasks. While this paradigm shift enhances personalization, it introduces a vulnerability: reliance on Long-term Memory (LTM). In this paper, we uncover a threat termed "Visual Inception." Unlike traditional adversarial attacks that seek immediate misclassification, Visual Inception injects triggers into user-uploaded images (e.g., lifestyle photos) that act as "sleeper agents" within the system's memory. When retrieved during future planning, these poisoned memories hijack the agent's reasoning chain, steering it toward adversary-defined goals (e.g., promoting high-margin products) without prompt injection. To mitigate this, we propose CognitiveGuard, a dual-process defense framework inspired by human cognition. It consists of a System 1 Perceptual Sanitizer (diffusion-based purification) to cleanse sensory inputs and a System 2 Reasoning Verifier (counterfactual consistency checks) to detect anomalies in memory-driven planning. Extensive experiments on a mock e-commerce agent environment demonstrate that Visual Inception achieves about 85% Goal-Hit Rate (GHR), while CognitiveGuard reduces this risk to around 10% with configurable latency trade-offs (about 1.5s in lite mode to about 6.5s for full sequential verification), without quality degradation under our setup.

The reconfigurable intelligent surface (RIS) has been widely studied in far-field communications (FFC), and further extended to near-field communications (NFC). With the distinct electromagnetic properties in FFC and NFC, it remains debatable whether the RIS deployment strategies established for FFC are still applicable to NFC. To bridge this gap, we examine RIS-aided cooperative relaying in NFC through two representative configurations, single-RIS and multi-RIS relaying, in which a decode-and-forward relay forwards data from the source to the user with the assistance of RIS For the single RIS deployment, we show that the achievable rates with RIS deployed near the base station (BS) or near relay are completely different. It is revealed that, with a single-antenna relay, the multi-RIS is significantly preferable over the single RIS in NFC without the requirement of massive RIS reflection elements, which is completely different from that of FFC. Furthermore, we derive closed-form expressions for the achievable rates under both single-antenna and multi-antenna relay configurations to explicitly determine the distance threshold. The presented analytical results demonstrate that locating RIS near the relay yields substantial performance gains when the transmission distance exceeds a certain threshold. This conclusion is validated by numerical simulations, which systematically illustrates the distinct impact of RIS deployment strategies in near-field versus far-field regions.

Current state-of-the-art text-to-3D generation methods struggle to produce 3D models with fine details and delicate structures due to limitations in differentiable mesh representation techniques. This limitation is particularly pronounced in anime character generation, where intricate features such as fingers, hair, and facial details are crucial for capturing the essence of the characters. In this paper, we introduce a novel, efficient, sparse differentiable mesh representation method, termed SparseCubes, alongside a sparse transformer network designed to generate high-quality 3D models. Our method significantly reduces computational requirements by over 95% and storage memory by 50%, enabling the creation of higher resolution meshes with enhanced details and delicate structures. We validate the effectiveness of our approach through its application to text-to-3D anime character generation, demonstrating its capability to accurately render subtle details and thin structures (e.g. individual fingers) in both meshes and textures.

Generating high-resolution 3D shapes using volumetric representations such as Signed Distance Functions (SDFs) presents substantial computational and memory challenges. We introduce Direct3D-S2, a scalable 3D generation framework based on sparse volumes that achieves superior output quality with dramatically reduced training costs. Our key innovation is the Spatial Sparse Attention (SSA) mechanism, which greatly enhances the efficiency of Diffusion Transformer (DiT) computations on sparse volumetric data. SSA allows the model to effectively process large token sets within sparse volumes, significantly reducing computational overhead and achieving a 3.9$\times$ speedup in the forward pass and a 9.6$\times$ speedup in the backward pass. Our framework also includes a variational autoencoder (VAE) that maintains a consistent sparse volumetric format across input, latent, and output stages. Compared to previous methods with heterogeneous representations in 3D VAE, this unified design significantly improves training efficiency and stability. Our model is trained on public datasets, and experiments demonstrate that Direct3D-S2 not only surpasses state-of-the-art methods in generation quality and efficiency, but also enables training at 1024³ resolution using only 8 GPUs—a task typically requiring at least 32 GPUs for volumetric representations at $256^3$ resolution, thus making gigascale 3D generation both practical and accessible. Project page: https://www.neural4d.com/research-page/direct3d-s2.

In this paper, we investigate the impact of random wobbling on the performance of unmanned aerial vehicle (UAV) communications. We consider two practical implementation forms of UAV-multiple input multiple output (MIMO) system, namely 1) centralized implementation, where a compact multi-antenna array is deployed on a single UAV platform and hence wobbling directly has impact on the entire array; 2) distributed implementation, where a virtual MIMO system is formed via cooperative UAV swarm, and wobbling occurs independently for every single transmit antenna. For both cases, we define and analyze the beamforming gain loss-factor as the performance metric to evaluate the adverse effect caused by wobbling. Based on derived analytical expressions, the centralized and distributed implementations are compared to show their preferable operating scenarios, respectively, considering different system and implementation parameters including the carrier frequency, wobbling variance, user position, antenna number, and inter-antenna spacing, etc. It is revealed that the distributed MIMO implementation could be more robust to wobbling, especially when the number of antennas is large. Simulations verify the accuracy of derived analytical expressions, and confirm the corresponding conclusions.