arXiv subjects

Jiaqi Feng

Publications and source records attributed to Jiaqi Feng.

13 recordsLinked to original sources

arXiv2026

Stream4D: 4D-Consistency for Streaming Autoregressive Diffusion Video Models

Streaming autoregressive diffusion models enable real-time, long-horizon video generation, but their training objectives optimize local frame prediction rather than the geometry and dynamics of a coherent world: long rollouts accumulate geometric drift and degrade into static or unnatural motion. Recent bidirectional approaches address this problem using rewards signals built upon 3D Gaussian-Splatting reconstruction. However, a single rigid 3d reconstruction cannot model a dynamic scene, so this critic penalizes genuine object motion as reconstruction error and is maximized by freezing the video. This shortcut is especially detrimental in the AR setting, where each chunk can propagate an already-static configuration. In this work, we propose Stream4D, which replaces the static critic with a feed-forward 4D reconstruction reward that explicitly models scene dynamics, allowing coherent motion to receive high consistency rewards. To further guide motion magnitude and quality, we add a motion prior that rewards natural scene-flow magnitude while penalizing jitter and non-rigid artifacts. Our final recipe combines these two terms with a lightweight perceptual anchor. Across various autoregressive video backbones and various generation horizons, Stream4D improves 4D reconstruction quality, preserves motion more effectively, and achieves higher human-aligned preference. Project page: https://banyuanhao.github.io/Stream4D/

cs.CV↗

arXiv2026

RecoReward: Recommender-Guided Multimodal Description Generation for Recommendation

Multimodal large language models (MLLMs) can convert multimodal item content into structured descriptions used as semantic features for recommendation. Conventional content-only generation, however, cannot use downstream user signals to determine which semantics should be emphasized. Recent user-conditioned methods incorporate these signals through user histories or profiles, but they require user information at inference and make generation user-dependent. In this paper, we introduce RecoReward, which instead uses behavior-derived rewards during training and preserves content-only inference. To instantiate this idea in live-stream recommendation, we treat historically engaged users as a proxy for future target users and use observational non-target users to estimate affinity shared broadly across users. The Recommender Affinity Score (RAS) contrasts these signals to provide user-selective feedback for reinforcement learning, allowing the learned policy to generate a single shared description without user inputs. In our offline benchmark, RecoReward-9B outperforms its Qwen3.5-9B baseline and all other evaluated models across seven recall metrics. Online A/B testing also shows performance gains. These results show that RecoReward trains the MLLM to produce item features that benefit downstream recommendation while retaining content-only serving.

cs.IR↗

arXiv2026

One-Forcing: Towards Stable One-Step Autoregressive Video Generation

Recent advances have substantially improved real-time interactive video generation in the autoregressive regime. However, most existing few-step autoregressive video generation methods, often distilled from a corresponding many-step teacher, default to a 4-step sampling configuration, which still incurs considerable latency during deployment and suffers from severe quality degradation when the number of sampling steps is further reduced, particularly in the one-step setting. Trajectory-style consistency distillation methods often produce videos with weak dynamics, while DMD-based approaches, such as Self-Forcing, tend to yield blurry frames. To address this challenge, we propose One-Forcing, a simple yet effective approach which augments the DMD objective with an auxiliary GAN loss for high-quality and efficient one-step video generation. Experiments on VBench show that One-Forcing achieves a total score of 83.76, establishing state-of-the-art performance among one-step causal video generation methods and remaining competitive with strong many-step approaches. We further demonstrate that one-step framewise autoregressive generation can be achieved stably with merely one-third of the training cost of the chunkwise model, a setting that prior methods have failed to achieve successfully.

cs.CV↗

arXiv2026

All Vehicles Can Lie: Efficient Adversarial Defense in Fully Untrusted-Vehicle Collaborative Perception via Pseudo-Random Bayesian Inference

Collaborative perception (CP) enables multiple vehicles to augment their individual perception capacities through the exchange of feature-level sensory data. However, this fusion mechanism is inherently vulnerable to adversarial attacks, especially in fully untrusted-vehicle environments. Existing defense approaches often assume a trusted ego vehicle as a reference or incorporate additional binary classifiers. These assumptions limit their practicality in real-world deployments due to the questionable trustworthiness of ego vehicles, the requirement for real-time detection, and the need for generalizability across diverse scenarios. To address these challenges, we propose a novel Pseudo-Random Bayesian Inference (PRBI) framework, a first efficient defense method tailored for fully untrusted-vehicle CP. PRBI detects adversarial behavior by leveraging temporal perceptual discrepancies, using the reliable perception from the preceding frame as a dynamic reference. Additionally, it employs a pseudo-random grouping strategy that requires only two verifications per frame, while applying Bayesian inference to estimate both the number and identities of malicious vehicles. Theoretical analysis has proven the convergence and stability of the proposed PRBI framework. Extensive experiments show that PRBI requires only 2.5 verifications per frame on average, outperforming existing methods significantly, and restores detection precision to between 79.4% and 86.9% of pre-attack levels.

cs.CV↗

arXiv2026

Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents

Large language model (LLM) agents face fundamental limitations in long-horizon reasoning due to finite context windows, making effective memory management critical. Existing methods typically handle long-term memory (LTM) and short-term memory (STM) as separate components, relying on heuristics or auxiliary controllers, which limits adaptability and end-to-end optimization. In this paper, we propose Agentic Memory (AgeMem), a unified framework that integrates LTM and STM management directly into the agent's policy. AgeMem exposes memory operations as tool-based actions, enabling the LLM agent to autonomously decide what and when to store, retrieve, update, summarize, or discard information. To train such unified behaviors, we propose a three-stage progressive reinforcement learning strategy and design a step-wise GRPO to address sparse and discontinuous rewards induced by memory operations. Experiments on five long-horizon benchmarks demonstrate that AgeMem consistently outperforms strong memory-augmented baselines across multiple LLM backbones, achieving improved task performance, higher-quality long-term memory, and more efficient context usage.

cs.CL↗

arXiv2025

Astra: General Interactive World Model with Autoregressive Denoising

Recent advances in diffusion transformers have empowered video generation models to generate high-quality video clips from texts or images. However, world models with the ability to predict long-horizon futures from past observations and actions remain underexplored, especially for general-purpose scenarios and various forms of actions. To bridge this gap, we introduce Astra, an interactive general world model that generates real-world futures for diverse scenarios (e.g., autonomous driving, robot grasping) with precise action interactions (e.g., camera motion, robot action). We propose an autoregressive denoising architecture and use temporal causal attention to aggregate past observations and support streaming outputs. We use a noise-augmented history memory to avoid over-reliance on past frames to balance responsiveness with temporal coherence. For precise action control, we introduce an action-aware adapter that directly injects action signals into the denoising process. We further develop a mixture of action experts that dynamically route heterogeneous action modalities, enhancing versatility across diverse real-world tasks such as exploration, manipulation, and camera control. Astra achieves interactive, consistent, and general long-term video prediction and supports various forms of interactions. Experiments across multiple datasets demonstrate the improvements of Astra in fidelity, long-range prediction, and action alignment over existing state-of-the-art world models.

cs.CV↗

arXiv2025

Ferroelectric Antiferromagnetic Lifting of Spin-Valley Degeneracy

The generation and control of spin- and valley-polarization in antiferromagnets (AFMs) have garnered increasing attention due to their potential for enabling faster and more stable multifunctional spintronic and valleytronic memory and logic devices. However, the two primary categories of AFMs, altermagnets and TP-symmetric AFMs, either lack intrinsic valley-polarization or net spin-polarization. Here, we propose an effective approach for achieving spontaneous spin-valley polarization in TP-broken layered ferroelectric antiferromagnets (FE-AFMs). The FE-AFMs exhibit lifted spin degeneracy across the entire Brillouin zone, along with uncompensated spin density of states. They combine the benefits of spin-polarization in altermagnets with valley-polarization in TP-symmetric AFMs. Furthermore, the FE-AFMs feature layer-dependent spin-polarization, rooted in their intrinsic ferroelectric property, allowing for the flexible control over spin-valley polarization by interlayer sliding. This tunability facilitates sign-reversible and size-tunable valley Hall and Nernst effects, along with other spin-valley-dependent transport properties. Our findings are demonstrated in a broad class of TP-broken bilayer antiferromagnets such as Nb3X8 (X = Cl, Br, I), VX2 (X = S, Se), and VSi2X4 (X = N, P), underscoring the potential of FE-AFMs for advancing next-generation spin- and valley-based information technologies.

cond-mat.mes-hall↗

arXiv2025

EagleVision: Object-level Attribute Multimodal LLM for Remote Sensing

Recent advances in multimodal large language models (MLLMs) have demonstrated impressive results in various visual tasks. However, in remote sensing (RS), high resolution and small proportion of objects pose challenges to existing MLLMs, which struggle with object-centric tasks, particularly in precise localization and fine-grained attribute description for each object. These RS MLLMs have not yet surpassed classical visual perception models, as they only provide coarse image understanding, leading to limited gains in real-world scenarios. To address this gap, we establish EagleVision, an MLLM tailored for remote sensing that excels in object detection and attribute comprehension. Equipped with the Attribute Disentangle module, EagleVision learns disentanglement vision tokens to express distinct attributes. To support object-level visual-language alignment, we construct EVAttrs-95K, the first large-scale object attribute understanding dataset in RS for instruction tuning, along with a novel evaluation benchmark, EVBench. EagleVision achieves state-of-the-art performance on both fine-grained object detection and object attribute understanding tasks, highlighting the mutual promotion between detection and understanding capabilities in MLLMs. The code, model, data, and demo will be available at https://github.com/XiangTodayEatsWhat/EagleVision.

cs.CV↗

arXiv2024

Topology-Aware 3D Gaussian Splatting: Leveraging Persistent Homology for Optimized Structural Integrity

Gaussian Splatting (GS) has emerged as a crucial technique for representing discrete volumetric radiance fields. It leverages unique parametrization to mitigate computational demands in scene optimization. This work introduces Topology-Aware 3D Gaussian Splatting (Topology-GS), which addresses two key limitations in current approaches: compromised pixel-level structural integrity due to incomplete initial geometric coverage, and inadequate feature-level integrity from insufficient topological constraints during optimization. To overcome these limitations, Topology-GS incorporates a novel interpolation strategy, Local Persistent Voronoi Interpolation (LPVI), and a topology-focused regularization term based on persistent barcodes, named PersLoss. LPVI utilizes persistent homology to guide adaptive interpolation, enhancing point coverage in low-curvature areas while preserving topological structure. PersLoss aligns the visual perceptual similarity of rendered images with ground truth by constraining distances between their topological features. Comprehensive experiments on three novel-view synthesis benchmarks demonstrate that Topology-GS outperforms existing methods in terms of PSNR, SSIM, and LPIPS metrics, while maintaining efficient memory usage. This study pioneers the integration of topology with 3D-GS, laying the groundwork for future research in this area.

cs.CV↗

arXiv2024

Layer Control of Magneto-Optical Effects and Their Quantization in Spin-Valley Splitting Antiferromagnets

Magneto-optical effects (MOE), interfacing the fundamental interplay between magnetism and light, have served as a powerful probe for magnetic order, band topology, and valley index. Here, based on multiferroic and topological bilayer antiferromagnets (AFMs), we propose a layer control of MOE (L-MOE), which is created and annihilated by layer-stacking or an electric field effect. The key character of L-MOE is the sign-reversible response controlled by ferroelectric polarization, the Neel vector, or the electric field direction. Moreover, the sign-reversible L-MOE can be quantized in topologically insulating AFMs. We reveal that the switchable L-MOE originates from the combined contributions of spin-conserving and spin-flip interband transitions in spin-valley splitting AFMs, a phenomenon not observed in conventional AFMs. Our findings bridge the ancient MOE to the emergent realms of layertronics, valleytronics, and multiferroics and may hold immense potential in these fields.

cond-mat.mes-hall↗

arXiv2024

Graph Neural Networks in Intelligent Transportation Systems: Advances, Applications and Trends

Intelligent Transportation System (ITS) is crucial for improving traffic congestion, reducing accidents, optimizing urban planning, and more. However, the complexity of traffic networks has rendered traditional machine learning and statistical methods less effective. With the advent of artificial intelligence, deep learning frameworks have achieved remarkable progress across various fields and are now considered highly effective in many areas. Since 2019, Graph Neural Networks (GNNs) have emerged as a particularly promising deep learning approach within the ITS domain, owing to their robust ability to model graph-structured data and address complex problems. Consequently, there has been increasing scholarly attention to the applications of GNNs in transportation, which have demonstrated excellent performance. Nevertheless, current research predominantly focuses on traffic forecasting, with other ITS domains, such as autonomous vehicles and demand prediction, receiving less attention. This paper aims to review the applications of GNNs across six representative and emerging ITS research areas: traffic forecasting, vehicle control system, traffic signal control, transportation safety, demand prediction, and parking management. We have examined a wide range of graph-related studies from 2018 to 2023, summarizing their methodologies, features, and contributions in detailed tables and lists. Additionally, we identify the challenges of applying GNNs in ITS and propose potential future research directions.

cs.LG↗

arXiv2022

Sampling Is All You Need on Modeling Long-Term User Behaviors for CTR Prediction

Rich user behavior data has been proven to be of great value for Click-Through Rate (CTR) prediction applications, especially in industrial recommender, search, or advertising systems. However, it's non-trivial for real-world systems to make full use of long-term user behaviors due to the strict requirements of online serving time. Most previous works adopt the retrieval-based strategy, where a small number of user behaviors are retrieved first for subsequent attention. However, the retrieval-based methods are sub-optimal and would cause more or less information losses, and it's difficult to balance the effectiveness and efficiency of the retrieval algorithm. In this paper, we propose SDIM (Sampling-based Deep Interest Modeling), a simple yet effective sampling-based end-to-end approach for modeling long-term user behaviors. We sample from multiple hash functions to generate hash signatures of the candidate item and each item in the user behavior sequence, and obtain the user interest by directly gathering behavior items associated with the candidate item with the same hash signature. We show theoretically and experimentally that the proposed method performs on par with standard attention-based models on modeling long-term user behaviors, while being sizable times faster. We also introduce the deployment of SDIM in our system. Specifically, we decouple the behavior sequence hashing, which is the most time-consuming part, from the CTR model by designing a separate module named BSE (behavior Sequence Encoding). BSE is latency-free for the CTR server, enabling us to model extremely long user behaviors. Both offline and online experiments are conducted to demonstrate the effectiveness of SDIM. SDIM now has been deployed online in the search system of Meituan APP.

cs.IR↗

arXiv2021

Efficient electrochemical reduction of CO2 to CO by soft functional materials

Electrochemical reduction of CO2 to CO is a promising strategy. However, achieving high Faradaic efficiency with high current density using ILs electrolyte remains a challenge. In this study, the IL N octyltrimethyl 1,2,4 triazole ammonium shows outstanding performance for electrochemical reduction of CO2 to CO on the commercial Ag electrode, and the current density can be up to 50.8 mA cm-2 with a Faradaic efficiency of 90.6%. The current density of CO is much higher than those reported in the ILs electrolyte. In addition, the density functional theory calculation further proved that IL interacts with CO2 to form IL CO2 complex which played a key role in reducing the activation energy of CO2. According to the molecular orbital theory, the electrons obtained from ILs was filled in the anti bonding orbit of the CO2, resulting in reducing the C=O bond energy. This work provides a new strategy to design novel ILs for high efficiency electrochemical reduction of CO2 to CO.

cond-mat.mtrl-sci↗