arXiv subjects

Jiaqi Liu

Publications and source records attributed to Jiaqi Liu.

At least 145 records · Page 8Linked to original sources

arXiv2025

Towards Human-Centric Autonomous Driving: A Fast-Slow Architecture Integrating Large Language Model Guidance with Reinforcement Learning

Autonomous driving has made significant strides through data-driven techniques, achieving robust performance in standardized tasks. However, existing methods frequently overlook user-specific preferences, offering limited scope for interaction and adaptation with users. To address these challenges, we propose a "fast-slow" decision-making framework that integrates a Large Language Model (LLM) for high-level instruction parsing with a Reinforcement Learning (RL) agent for low-level real-time decision. In this dual system, the LLM operates as the "slow" module, translating user directives into structured guidance, while the RL agent functions as the "fast" module, making time-critical maneuvers under stringent latency constraints. By decoupling high-level decision making from rapid control, our framework enables personalized user-centric operation while maintaining robust safety margins. Experimental evaluations across various driving scenarios demonstrate the effectiveness of our method. Compared to baseline algorithms, the proposed architecture not only reduces collision rates but also aligns driving behaviors more closely with user preferences, thereby achieving a human-centric mode. By integrating user guidance at the decision level and refining it with real-time control, our framework bridges the gap between individual passenger needs and the rigor required for safe, reliable driving in complex traffic environments.

cs.RO↗

arXiv2025

Exact multiple complex mobility edges and quantum state engineering in coupled 1D quasicystals

The key concept of mobility edge, which marks the critical transition between extended and localized states in energy domain, has attracted significant interest in the cutting-edge frontiers of modern physics due to its profound implications for understanding localization and transport properties in disordered systems. However, a generic way to construct multiple mobility edges (MME) is still ambiguous and lacking. In this work, we propose a brief scheme to engineer both real and complex exact multiple mobility edges exploiting a few coupled one-dimensional quasiperiodic chains. We study the extended-localized transitions of coupled one-dimensional quasiperiodic chains along the chain direction. The model combines both the well-established quasiperiodicity and a kind of freshly introduced staggered non-reciprocity, which are aligned in two mutually perpendicular directions, within a unified framework. Based on analytical analysis, we predict that when the couplings between quasiperiodic chains are weak, the system will be in a mixed phase in which the localized states and extended states coexist and intertwine, thus lacking explicit energy separations. However, as the inter-chain couplings increase to certain strength, exact multiple mobility edges emerge. This prediction is clearly verified by concrete numerical calculations of the Fractal Dimension and the scaling index $β$. Moreover, we show that the combination of quasiperiodicity and the staggered non-reciprocity can be utilized to design and realize quantum states of various configurations. Our results reveal a brief and general scheme to implement exact multiple mobility edges for synthetic materials engineering.

cond-mat.dis-nn↗

arXiv2025

Long time asymptotics of a perturbed modified KdV equation

We derive full asymptotics of the modified KdV equation (mKdV) with a higher-order perturbative term. We make use of the perturbative theory of infinite-dimensional integrable systems developed by P. Deift and X. Zhou \cite{DZ-2}, and some new and simpler proofs of certain $L^\infty$ bounds and $L^p$ a priori estimates developed recently in \cite{CLT}. We show that the perturbed equation exhibits the same long-time behavior as the completely integrable mKdV.

math.AP↗

arXiv2025

UQABench: Evaluating User Embedding for Prompting LLMs in Personalized Question Answering

Large language models (LLMs) achieve remarkable success in natural language processing (NLP). In practical scenarios like recommendations, as users increasingly seek personalized experiences, it becomes crucial to incorporate user interaction history into the context of LLMs to enhance personalization. However, from a practical utility perspective, user interactions' extensive length and noise present challenges when used directly as text prompts. A promising solution is to compress and distill interactions into compact embeddings, serving as soft prompts to assist LLMs in generating personalized responses. Although this approach brings efficiency, a critical concern emerges: Can user embeddings adequately capture valuable information and prompt LLMs? To address this concern, we propose \name, a benchmark designed to evaluate the effectiveness of user embeddings in prompting LLMs for personalization. We establish a fair and standardized evaluation process, encompassing pre-training, fine-tuning, and evaluation stages. To thoroughly evaluate user embeddings, we design three dimensions of tasks: sequence understanding, action prediction, and interest perception. These evaluation tasks cover the industry's demands in traditional recommendation tasks, such as improving prediction accuracy, and its aspirations for LLM-based methods, such as accurately understanding user interests and enhancing the user experience. We conduct extensive experiments on various state-of-the-art methods for modeling user embeddings. Additionally, we reveal the scaling laws of leveraging user embeddings to prompt LLMs. The benchmark is available online.

cs.IR↗

arXiv2025

Multi-focal Conditioned Latent Diffusion for Person Image Synthesis

The Latent Diffusion Model (LDM) has demonstrated strong capabilities in high-resolution image generation and has been widely employed for Pose-Guided Person Image Synthesis (PGPIS), yielding promising results. However, the compression process of LDM often results in the deterioration of details, particularly in sensitive areas such as facial features and clothing textures. In this paper, we propose a Multi-focal Conditioned Latent Diffusion (MCLD) method to address these limitations by conditioning the model on disentangled, pose-invariant features from these sensitive regions. Our approach utilizes a multi-focal condition aggregation module, which effectively integrates facial identity and texture-specific information, enhancing the model's ability to produce appearance realistic and identity-consistent images. Our method demonstrates consistent identity and appearance generation on the DeepFashion dataset and enables flexible person image editing due to its generation consistency. The code is available at https://github.com/jqliu09/mcld.

cs.CV↗

arXiv2025

Geometrically Templated Dynamic Wrinkling from Suspended Poly(vinyl alcohol) Soap Films

Wrinkling is commonly observed as mechanical instability when a stiff thin film bound on a compliant thick substrate undergoes in-plane compression exceeding a threshold. Despite significant efforts to create a broad range of surface patterns via wrinkling, little has been studied about a dynamic and transient wrinkling process, where a suspended polymer thin film undergoes liquid-to-solid phase transitions. Here, a spontaneous wrinkling process is reported, when drying poly(vinyl alcohol) (PVA) soap films suspended on 3D printed wireframes with near zero or negative Gaussian curvatures. As water evaporates, a thickness gradient across the sample is developed, leading to non-uniform drying rates, and a concentration gradient between the inner and outer sides (exposed to air) of the suspended PVA soap film induces a differential osmotic pressure. Together, these effects contribute to an in-plane compressive stress, leading to the formation of surface wrinkles, whose growth is guided by the geometry of the frame. Importantly, the wrinkles evolve dynamically: the wavelength and number of the wrinkles can be tuned by altering the concentration of the PVA aqueous solutions, the initial mass, the relative humidity of the drying environment; the patterns of the resulting wrinkles can be programmed by the geometry of the wireframe.

cond-mat.soft↗

arXiv2025

Interact, Instruct to Improve: A LLM-Driven Parallel Actor-Reasoner Framework for Enhancing Autonomous Vehicle Interactions

Autonomous Vehicles (AVs) have entered the commercialization stage, but their limited ability to interact and express intentions still poses challenges in interactions with Human-driven Vehicles (HVs). Recent advances in large language models (LLMs) enable bidirectional human-machine communication, but the conflict between slow inference speed and the need for real-time decision-making challenges practical deployment. To address these issues, this paper introduces a parallel Actor-Reasoner framework designed to enable explicit bidirectional AV-HV interactions across multiple scenarios. First, by facilitating interactions between the LLM-driven Reasoner and heterogeneous simulated HVs during training, an interaction memory database, referred to as the Actor, is established. Then, by introducing the memory partition module and the two-layer memory retrieval module, the Actor's ability to handle heterogeneous HVs is significantly enhanced. Ablation studies and comparisons with other decision-making methods demonstrate that the proposed Actor-Reasoner framework significantly improves safety and efficiency. Finally, with the combination of the external Human-Machine Interface (eHMI) information derived from Reasoner's reasoning and the feasible action solutions retrieved from the Actor, the effectiveness of the proposed Actor-Reasoner is confirmed in multi-scenario field interactions. Our code is available at https://github.com/FanGShiYuu/Actor-Reasoner.

cs.RO↗

arXiv2025

TeLL-Drive: Enhancing Autonomous Driving with Teacher LLM-Guided Deep Reinforcement Learning

Although Deep Reinforcement Learning (DRL) and Large Language Models (LLMs) each show promise in addressing decision-making challenges in autonomous driving, DRL often suffers from high sample complexity, while LLMs have difficulty ensuring real-time decision making. To address these limitations, we propose TeLL-Drive, a hybrid framework that integrates a Teacher LLM to guide an attention-based Student DRL policy. By incorporating risk metrics, historical scenario retrieval, and domain heuristics into context-rich prompts, the LLM produces high-level driving strategies through chain-of-thought reasoning. A self-attention mechanism then fuses these strategies with the DRL agent's exploration, accelerating policy convergence and boosting robustness across diverse driving conditions. The experimental results, evaluated across multiple traffic scenarios, show that TeLL-Drive outperforms existing baseline methods, including other LLM-based approaches, in terms of success rates, average returns, and real-time feasibility. Ablation studies underscore the importance of each model component, especially the synergy between the attention mechanism and LLM-driven guidance. Finally, we build a virtual-real fusion experimental platform to verify the real-time performance, robustness, and reliability of the algorithm running on real vehicles through vehicle-in-loop experiments.

cs.AI↗

arXiv2025

Unexpected large electrostatic gating by pyroelectric charge accumulation

Pyroelectricity refers to the accumulation of charges due to changes in the spontaneous polarization of ferroelectric materials when subjected to temperature variations. Typically, these pyroelectric charges are considered unstable and dissipate quickly through interactions with the external environment. Consequently, the pyroelectric effect has been largely overlooked in ferroelectric field-effect transistors. In this work, we leverage the van der Waals interface of hBN to achieve a substantial and long-term electrostatic gating effect in graphene devices via the pyroelectric properties of a ferroelectric LiNbO3 substrate. Upon cooling, the polarization change in LiNbO3 induces high doping concentrations up to 1013 cm-2 in the adjacent graphene. Through a combination of transport measurements and non-contact techniques, we demonstrate that the pyroelectric charge accumulation, as well as its enhancement in electric fields, are responsible for this unexpectedly high doping level. Our findings introduce a novel mechanism for voltage-free electrostatic gating control with long retention.

cond-mat.mtrl-sci↗

arXiv2025

Broadband photoresponse enhancement by band engineering in Sb-doped MnBi2Te4

Topological materials have attracted considerable attention for their potential in broadband and fast photoresponse, particularly in the infrared regime. However, the high carrier concentration in these systems often leads to rapid recombination of photogenerated carriers, limiting the photoresponsivity. Here, we demonstrate that Sb doping in MnBi2Te4 effectively reduces carrier concentration and suppresses electron-hole recombination, thereby significantly improving the optoelectronic performance across the visible to mid-infrared spectra. The optimally doped Mn(Bi0.82Sb0.18)2Te4 photodetector achieves a responsivity of 3.02 mA W-1 with a response time of 18.5 μs at 1550 nm, and 0.795 mA W-1 with a response time of 9.0 μs at 4 μm. These values represent nearly two orders of magnitude improvement compared to undoped MnBi2Te4. Our results highlight band engineering as an effective strategy to enhance the infrared performance of topological material-based photodetectors, opening new avenues for high-sensitivity infrared detection.

cond-mat.mtrl-sci↗

arXiv2025

Waiting Time Solutions in gas dynamics

In this article, we construct a continuum family of self-similar waiting time solutions for the one-dimensional compressible Euler equations for the adiabatic exponent $\ga\in(1,3)$ in the half-line with the vacuum boundary. The solutions are confined by a stationary vacuum interface for a finite time with at least $C^1$ regularity of the velocity and the sound speed up to the boundary. Subsequently, the solutions undergo the change of the behavior, becoming only Hölder continuous near the singular point, and simultaneously transition to the solutions to the vacuum moving boundary Euler equations satisfying the physical vacuum condition. When the boundary starts moving, a weak discontinuity emanating from the singular point along the sonic curve emerges. The solutions are locally smooth in the interior region away from the vacuum boundary and the sonic curve.

math.AP↗

arXiv2024

Revisiting Multimodal Fusion for 3D Anomaly Detection from an Architectural Perspective

Existing efforts to boost multimodal fusion of 3D anomaly detection (3D-AD) primarily concentrate on devising more effective multimodal fusion strategies. However, little attention was devoted to analyzing the role of multimodal fusion architecture (topology) design in contributing to 3D-AD. In this paper, we aim to bridge this gap and present a systematic study on the impact of multimodal fusion architecture design on 3D-AD. This work considers the multimodal fusion architecture design at the intra-module fusion level, i.e., independent modality-specific modules, involving early, middle or late multimodal features with specific fusion operations, and also at the inter-module fusion level, i.e., the strategies to fuse those modules. In both cases, we first derive insights through theoretically and experimentally exploring how architectural designs influence 3D-AD. Then, we extend SOTA neural architecture search (NAS) paradigm and propose 3D-ADNAS to simultaneously search across multimodal fusion strategies and modality-specific modules for the first time.Extensive experiments show that 3D-ADNAS obtains consistent improvements in 3D-AD across various model capacities in terms of accuracy, frame rate, and memory usage, and it exhibits great potential in dealing with few-shot 3D-AD tasks.

cs.CV↗

arXiv2024

Higher order asymptotics for the nonlinear Schrödinger equation

In this paper we compute the higher order long time asymptotics of the defocussing nonlinear Schrödinger equation using the $\overline{\partial}$-nonlinear steepest descent method. We assume initial condition in weighted Sobolev space with finite order of regularity and decay.

math.AP↗

arXiv2024

Certified Minimax Unlearning with Generalization Rates and Deletion Capacity

We study the problem of $(ε,δ)$-certified machine unlearning for minimax models. Most of the existing works focus on unlearning from standard statistical learning models that have a single variable and their unlearning steps hinge on the direct Hessian-based conventional Newton update. We develop a new $(ε,δ)$-certified machine unlearning algorithm for minimax models. It proposes a minimax unlearning step consisting of a total-Hessian-based complete Newton update and the Gaussian mechanism borrowed from differential privacy. To obtain the unlearning certification, our method injects calibrated Gaussian noises by carefully analyzing the "sensitivity" of the minimax unlearning step (i.e., the closeness between the minimax unlearning variables and the retraining-from-scratch variables). We derive the generalization rates in terms of population strong and weak primal-dual risk for three different cases of loss functions, i.e., (strongly-)convex-(strongly-)concave losses. We also provide the deletion capacity to guarantee that a desired population risk can be maintained as long as the number of deleted samples does not exceed the derived amount. With training samples $n$ and model dimension $d$, it yields the order $\mathcal O(n/d^{1/4})$, which shows a strict gap over the baseline method of differentially private minimax learning that has $\mathcal O(n/d^{1/2})$. In addition, our rates of generalization and deletion capacity match the state-of-the-art rates derived previously for standard statistical learning models.

cs.LG↗

arXiv2024

A Survey on Diffusion Models for Recommender Systems

While traditional recommendation techniques have made significant strides in the past decades, they still suffer from limited generalization performance caused by factors like inadequate collaborative signals, weak latent representations, and noisy data. In response, diffusion models (DMs) have emerged as promising solutions for recommender systems due to their robust generative capabilities, solid theoretical foundations, and improved training stability. To this end, in this paper, we present the first comprehensive survey on diffusion models for recommendation, and draw a bird's-eye view from the perspective of the whole pipeline in real-world recommender systems. We systematically categorize existing research works into three primary domains: (1) diffusion for data engineering & encoding, focusing on data augmentation and representation enhancement; (2) diffusion as recommender models, employing diffusion models to directly estimate user preferences and rank items; and (3) diffusion for content presentation, utilizing diffusion models to generate personalized content such as fashion and advertisement creatives. Our taxonomy highlights the unique strengths of diffusion models in capturing complex data distributions and generating high-quality, diverse samples that closely align with user preferences. We also summarize the core characteristics of the adapting diffusion models for recommendation, and further identify key areas for future exploration, which helps establish a roadmap for researchers and practitioners seeking to advance recommender systems through the innovative application of diffusion models. To further facilitate the research community of recommender systems based on diffusion models, we actively maintain a GitHub repository for papers and other related resources in this rising direction https://github.com/CHIANGEL/Awesome-Diffusion-for-RecSys.

cs.IR↗

arXiv2024

The Yaglom limit for branching Brownian motion with absorption and slightly subcritical drift

Consider branching Brownian motion with absorption in which particles move independently as one-dimensional Brownian motions with drift $-ρ$, each particle splits into two particles at rate one, and particles are killed when they reach the origin. Kesten (1978) showed that this process dies out with probability one if and only if $ρ\geq \sqrt{2}$. We show that in the subcritical case when $ρ> \sqrt{2}$, the law of the process conditioned on survival until time $t$ converges as $t \rightarrow \infty$ to a quasi-stationary distribution, which we call the Yaglom limit. We give a construction of this quasi-stationary distribution. We also study the asymptotic behavior as $ρ\downarrow \sqrt{2}$ of this quasi-stationary distribution. We show that the logarithm of the number of particles and the location of the highest particle are of order $ε^{-1/3}$, and we obtain a limit result for the empirical distribution of the particle locations.

math.PR↗

arXiv2024

Cooperative Decision-Making for CAVs at Unsignalized Intersections: A MARL Approach with Attention and Hierarchical Game Priors

The development of autonomous vehicles has shown great potential to enhance the efficiency and safety of transportation systems. However, the decision-making issue in complex human-machine mixed traffic scenarios, such as unsignalized intersections, remains a challenge for autonomous vehicles. While reinforcement learning (RL) has been used to solve complex decision-making problems, existing RL methods still have limitations in dealing with cooperative decision-making of multiple connected autonomous vehicles (CAVs), ensuring safety during exploration, and simulating realistic human driver behaviors. In this paper, a novel and efficient algorithm, Multi-Agent Game-prior Attention Deep Deterministic Policy Gradient (MA-GA-DDPG), is proposed to address these limitations. Our proposed algorithm formulates the decision-making problem of CAVs at unsignalized intersections as a decentralized multi-agent reinforcement learning problem and incorporates an attention mechanism to capture interaction dependencies between ego CAV and other agents. The attention weights between the ego vehicle and other agents are then used to screen interaction objects and obtain prior hierarchical game relations, based on which a safety inspector module is designed to improve the traffic safety. Furthermore, both simulation and hardware-in-the-loop experiments were conducted, demonstrating that our method outperforms other baseline approaches in terms of driving safety, efficiency, and comfort.

cs.RO↗

arXiv2024

Human-Like Implicit Intention Expression for Autonomous Driving Motion Planning: A Method Based on Learning Human Intention Priors

One of the key factors determining whether autonomous vehicles (AVs) can be seamlessly integrated into existing traffic systems is their ability to interact smoothly and efficiently with human drivers and communicate their intentions. While many studies have focused on enhancing AVs' human-like interaction and communication capabilities at the behavioral decision-making level, a significant gap remains between the actual motion trajectories of AVs and the psychological expectations of human drivers. This discrepancy can seriously affect the safety and efficiency of AV-HV (Autonomous Vehicle-Human Vehicle) interactions. To address these challenges, we propose a motion planning method for AVs that incorporates implicit intention expression. First, we construct a trajectory space constraint based on human implicit intention priors, compressing and pruning the trajectory space to generate candidate motion trajectories that consider intention expression. We then apply maximum entropy inverse reinforcement learning to learn and estimate human trajectory preferences, constructing a reward function that represents the cognitive characteristics of drivers. Finally, using a Boltzmann distribution, we establish a probabilistic distribution of candidate trajectories based on the reward obtained, selecting human-like trajectory actions. We validated our approach on a real trajectory dataset and compared it with several baseline methods. The results demonstrate that our method excels in human-likeness, intention expression capability, and computational efficiency.

cs.RO↗