arXiv subjects

Peng Li

Publications and source records attributed to Peng Li.

At least 199 records · Page 11Linked to original sources

arXiv2025

Agent Hospital: A Simulacrum of Hospital with Evolvable Medical Agents

The recent rapid development of large language models (LLMs) has sparked a new wave of technological revolution in medical artificial intelligence (AI). While LLMs are designed to understand and generate text like a human, autonomous agents that utilize LLMs as their "brain" have exhibited capabilities beyond text processing such as planning, reflection, and using tools by enabling their "bodies" to interact with the environment. We introduce a simulacrum of hospital called Agent Hospital that simulates the entire process of treating illness, in which all patients, nurses, and doctors are LLM-powered autonomous agents. Within the simulacrum, doctor agents are able to evolve by treating a large number of patient agents without the need to label training data manually. After treating tens of thousands of patient agents in the simulacrum (human doctors may take several years in the real world), the evolved doctor agents outperform state-of-the-art medical agent methods on the MedQA benchmark comprising US Medical Licensing Examination (USMLE) test questions. Our methods of simulacrum construction and agent evolution have the potential in benefiting a broad range of applications beyond medical AI.

cs.AI↗

arXiv2025

Detection and control of electronic orbital magnetism by spin waves in honeycomb ferromagnets

Exploring and manipulating the orbital degrees of freedom in solids has become a fascinating research topic in modern magnetism. Here, we demonstrate that spin waves can provide a way to control electronic orbital magnetism by the mechanism of scalar spin chirality, allowing for experimental detection using techniques such as the magneto-optical Kerr effect and scanning transmission electron microscopy. By applying linear spin wave theory, we uncover that electronic magnon-driven orbital magnetization is extremely sensitive to the character of the magnonic excitations. Furthermore, we show that both the induced electronic orbital magnetism and the Nernst transport properties of the orbital angular momentum can be regulated by the strength of the Dzyaloshinskii-Moriya interaction, Kitaev interaction, as well as the direction and magnitude of the external magnetic field. We argue that magnon-mediated electronic orbital magnetism presents an emergent variable which has to be taken into account when considering the physics of coupling magnonic excitiations to phonons and light.

cond-mat.str-el↗

arXiv2025

Correlated electronic structures and unconventional superconductivity in bilayer nickelate heterostructures

The recent discovery of ambient-pressure superconductivity in thin-film bilayer nickelates opens new possibilities for investigating electronic structures in this new class of high-transition temperature $T_C$ superconductors. Here, we construct a realistic multi-orbital Hubbard model for the thin-film system, by integrating ab initio calculations with scanning transmission electron microscopy (STEM) measurements, which reveal a higher-symmetry lattice. The interaction parameters are calculated with the constrained random phase approximation (cRPA). Density functional theory (DFT) plus cluster dynamical mean-field theory (CDMFT) calculations, with cRPA calculated on-site Coulomb repulsive $U$ and experimentally measured electron filling $n$, quantitatively reproduces Fermi surfaces from angle-resolved photoemission spectroscopy (ARPES) experiments. The distinct Fermi surface topology from simple DFT+$U$ results features the indispensable role of correlation effects. Based upon the correlated electronic structures, A modified random-phase-approximation (RPA) approach yields a pronounced $s^{\pm}$-wave pairing instability, due to the strong spin fluctuations originated from Fermi surface nesting between bands with predominantly $d_{z^{2}}$ characters. Our findings highlight the quantitative effectiveness of the DFT+cRPA+CDMFT approach that precisely determines correlated electronic structure parameters without fine-tuning. The revealed intermediate correlation effect may explain the same order-of-magnitude onset $T_C$ observed both in pressured bulk and strained thin film bilayer nickelates.

cond-mat.str-el↗

arXiv2025

Hierarchical Superpixel Segmentation via Structural Information Theory

Superpixel segmentation is a foundation for many higher-level computer vision tasks, such as image segmentation, object recognition, and scene understanding. Existing graph-based superpixel segmentation methods typically concentrate on the relationships between a given pixel and its directly adjacent pixels while overlooking the influence of non-adjacent pixels. These approaches do not fully leverage the global information in the graph, leading to suboptimal segmentation quality. To address this limitation, we present SIT-HSS, a hierarchical superpixel segmentation method based on structural information theory. Specifically, we first design a novel graph construction strategy that incrementally explores the pixel neighborhood to add edges based on 1-dimensional structural entropy (1D SE). This strategy maximizes the retention of graph information while avoiding an overly complex graph structure. Then, we design a new 2D SE-guided hierarchical graph partitioning method, which iteratively merges pixel clusters layer by layer to reduce the graph's 2D SE until a predefined segmentation scale is achieved. Experimental results on three benchmark datasets demonstrate that the SIT-HSS performs better than state-of-the-art unsupervised superpixel segmentation algorithms. The source code is available at \url{https://github.com/SELGroup/SIT-HSS}.

cs.CV↗

arXiv2025

T2LR-Net: An unrolling network learning transformed tensor low-rank prior for dynamic MR image reconstruction

The tensor low-rank prior has attracted considerable attention in dynamic MR reconstruction. Tensor low-rank methods preserve the inherent high-dimensional structure of data, allowing for improved extraction and utilization of intrinsic low-rank characteristics. However, most current methods are still confined to utilizing low-rank structures either in the image domain or predefined transformed domains. Designing an optimal transformation adaptable to dynamic MRI reconstruction through manual efforts is inherently challenging. In this paper, we propose a deep unrolling network that utilizes the convolutional neural network (CNN) to adaptively learn the transformed domain for leveraging tensor low-rank priors. Under the supervised mechanism, the learning of the tensor low-rank domain is directly guided by the reconstruction accuracy. Specifically, we generalize the traditional t-SVD to a transformed version based on arbitrary high-dimensional unitary transformations and introduce a novel unitary transformed tensor nuclear norm (UTNN). Subsequently, we present a dynamic MRI reconstruction model based on UTNN and devise an efficient iterative optimization algorithm using ADMM, which is finally unfolded into the proposed T2LR-Net. Experiments on two dynamic cardiac MRI datasets demonstrate that T2LR-Net outperforms the state-of-the-art optimization-based and unrolling network-based methods.

eess.IV↗

arXiv2025

Diffusion as Shader: 3D-aware Video Diffusion for Versatile Video Generation Control

Diffusion models have demonstrated impressive performance in generating high-quality videos from text prompts or images. However, precise control over the video generation process, such as camera manipulation or content editing, remains a significant challenge. Existing methods for controlled video generation are typically limited to a single control type, lacking the flexibility to handle diverse control demands. In this paper, we introduce Diffusion as Shader (DaS), a novel approach that supports multiple video control tasks within a unified architecture. Our key insight is that achieving versatile video control necessitates leveraging 3D control signals, as videos are fundamentally 2D renderings of dynamic 3D content. Unlike prior methods limited to 2D control signals, DaS leverages 3D tracking videos as control inputs, making the video diffusion process inherently 3D-aware. This innovation allows DaS to achieve a wide range of video controls by simply manipulating the 3D tracking videos. A further advantage of using 3D tracking videos is their ability to effectively link frames, significantly enhancing the temporal consistency of the generated videos. With just 3 days of fine-tuning on 8 H800 GPUs using less than 10k videos, DaS demonstrates strong control capabilities across diverse tasks, including mesh-to-video generation, camera control, motion transfer, and object manipulation.

cs.CV↗

arXiv2024

Trimming Down Large Spiking Vision Transformers via Heterogeneous Quantization Search

Spiking Neural Networks (SNNs) are amenable to deployment on edge devices and neuromorphic hardware due to their lower dissipation. Recently, SNN-based transformers have garnered significant interest, incorporating attention mechanisms akin to their counterparts in Artificial Neural Networks (ANNs) while demonstrating excellent performance. However, deploying large spiking transformer models on resource-constrained edge devices such as mobile phones, still poses significant challenges resulted from the high computational demands of large uncompressed high-precision models. In this work, we introduce a novel heterogeneous quantization method for compressing spiking transformers through layer-wise quantization. Our approach optimizes the quantization of each layer using one of two distinct quantization schemes, i.e., uniform or power-of-two quantification, with mixed bit resolutions. Our heterogeneous quantization demonstrates the feasibility of maintaining high performance for spiking transformers while utilizing an average effective resolution of 3.14-3.67 bits with less than a 1% accuracy drop on DVS Gesture and CIFAR10-DVS datasets. It attains a model compression rate of 8.71x-10.19x for standard floating-point spiking transformers. Moreover, the proposed approach achieves a significant energy reduction of 5.69x, 8.72x, and 10.2x while maintaining high accuracy levels of 85.3%, 97.57%, and 80.4% on N-Caltech101, DVS-Gesture, and CIFAR10-DVS datasets, respectively.

cs.NE↗

arXiv2024

Towards 3D Acceleration for low-power Mixture-of-Experts and Multi-Head Attention Spiking Transformers

Spiking Neural Networks(SNNs) provide a brain-inspired and event-driven mechanism that is believed to be critical to unlock energy-efficient deep learning. The mixture-of-experts approach mirrors the parallel distributed processing of nervous systems, introducing conditional computation policies and expanding model capacity without scaling up the number of computational operations. Additionally, spiking mixture-of-experts self-attention mechanisms enhance representation capacity, effectively capturing diverse patterns of entities and dependencies between visual or linguistic tokens. However, there is currently a lack of hardware support for highly parallel distributed processing needed by spiking transformers, which embody a brain-inspired computation. This paper introduces the first 3D hardware architecture and design methodology for Mixture-of-Experts and Multi-Head Attention spiking transformers. By leveraging 3D integration with memory-on-logic and logic-on-logic stacking, we explore such brain-inspired accelerators with spatially stackable circuitry, demonstrating significant optimization of energy efficiency and latency compared to conventional 2D CMOS integration.

cs.NE↗

arXiv2024

Rate-Distortion Optimized Skip Coding of Region Adaptive Hierarchical Transform Coefficients for MPEG G-PCC

Three-dimensional (3D) point clouds are becoming more and more popular for representing 3D objects and scenes. Due to limited network bandwidth, efficient compression of 3D point clouds is crucial. To tackle this challenge, the Moving Picture Experts Group (MPEG) is actively developing the Geometry-based Point Cloud Compression (G-PCC) standard, incorporating innovative methods to optimize compression, such as the Region-Adaptive Hierarchical Transform (RAHT) nestled within a layer-by-layer octree-tree structure. Nevertheless, a notable problem still exists in RAHT, i.e., the proportion of zero residuals in the last few RAHT layers leads to unnecessary bitrate consumption. To address this problem, we propose an adaptive skip coding method for RAHT, which adaptively determines whether to encode the residuals of the last several layers or not, thereby improving the coding efficiency. In addition, we propose a rate-distortion cost calculation method associated with an adaptive Lagrange multiplier. Experimental results demonstrate that the proposed method achieves average Bjøntegaard rate improvements of -3.50%, -5.56%, and -4.18% for the Luma, Cb, and Cr components, respectively, on dynamic point clouds, when compared with the state-of-the-art G-PCC reference software under the common test conditions recommended by MPEG.

cs.CV↗

arXiv2024

Align3R: Aligned Monocular Depth Estimation for Dynamic Videos

Recent developments in monocular depth estimation methods enable high-quality depth estimation of single-view images but fail to estimate consistent video depth across different frames. Recent works address this problem by applying a video diffusion model to generate video depth conditioned on the input video, which is training-expensive and can only produce scale-invariant depth values without camera poses. In this paper, we propose a novel video-depth estimation method called Align3R to estimate temporal consistent depth maps for a dynamic video. Our key idea is to utilize the recent DUSt3R model to align estimated monocular depth maps of different timesteps. First, we fine-tune the DUSt3R model with additional estimated monocular depth as inputs for the dynamic scenes. Then, we apply optimization to reconstruct both depth maps and camera poses. Extensive experiments demonstrate that Align3R estimates consistent video depth and camera poses for a monocular video with superior performance than baseline methods.

cs.CV↗

arXiv2024

Multi-View Large Reconstruction Model via Geometry-Aware Positional Encoding and Attention

Despite recent advancements in the Large Reconstruction Model (LRM) demonstrating impressive results, when extending its input from single image to multiple images, it exhibits inefficiencies, subpar geometric and texture quality, as well as slower convergence speed than expected. It is attributed to that, LRM formulates 3D reconstruction as a naive images-to-3D translation problem, ignoring the strong 3D coherence among the input images. In this paper, we propose a Multi-view Large Reconstruction Model (M-LRM) designed to reconstruct high-quality 3D shapes from multi-views in a 3D-aware manner. Specifically, we introduce a multi-view consistent cross-attention scheme to enable M-LRM to accurately query information from the input images. Moreover, we employ the 3D priors of the input multi-view images to initialize the triplane tokens. Compared to previous methods, the proposed M-LRM can generate 3D shapes of high fidelity. Experimental studies demonstrate that our model achieves a significant performance gain and faster training convergence. Project page: \url{https://murphylmf.github.io/M-LRM/}.

cs.CV↗

arXiv2024

Well-posedness for the semilinear wave equations with nonlinear damping on manifolds with conical degeneration

This paper deals with a class of semilinear wave equation with nonlinear damping term $|u_{t}|^{m-2}u_t $ and nonlinear source term $g(x)|u|^{p-2}u$ on the manifolds with conical singularities. Firstly, we prove the local existence and uniqueness of the solution by the semigroup method. Secondly, we establish the global existence, the energy decay estimate and the blow-up in finite time of the solution with subcritial ($E(0)<d$) and critial ($E(0)=d$) initial energy level by constructing potential wells. We also show that the solution is global provided the damping dominates the source (that is $m\geq p$). Moreover, we prove the blow-up in finite time of the solution with arbitrary positive initial energy and give some necessary and sufficient condition for existing finite time blow-up solutions.

math.AP↗

arXiv2024

Era3D: High-Resolution Multiview Diffusion using Efficient Row-wise Attention

In this paper, we introduce Era3D, a novel multiview diffusion method that generates high-resolution multiview images from a single-view image. Despite significant advancements in multiview generation, existing methods still suffer from camera prior mismatch, inefficacy, and low resolution, resulting in poor-quality multiview images. Specifically, these methods assume that the input images should comply with a predefined camera type, e.g. a perspective camera with a fixed focal length, leading to distorted shapes when the assumption fails. Moreover, the full-image or dense multiview attention they employ leads to an exponential explosion of computational complexity as image resolution increases, resulting in prohibitively expensive training costs. To bridge the gap between assumption and reality, Era3D first proposes a diffusion-based camera prediction module to estimate the focal length and elevation of the input image, which allows our method to generate images without shape distortions. Furthermore, a simple but efficient attention layer, named row-wise attention, is used to enforce epipolar priors in the multiview diffusion, facilitating efficient cross-view information fusion. Consequently, compared with state-of-the-art methods, Era3D generates high-quality multiview images with up to a 512*512 resolution while reducing computation complexity by 12x times. Comprehensive experiments demonstrate that Era3D can reconstruct high-quality and detailed 3D meshes from diverse single-view input images, significantly outperforming baseline multiview diffusion methods. Project page: https://penghtyx.github.io/Era3D/.

cs.CV↗

arXiv2024

AIGS: Generating Science from AI-Powered Automated Falsification

Rapid development of artificial intelligence has drastically accelerated the development of scientific discovery. Trained with large-scale observation data, deep neural networks extract the underlying patterns in an end-to-end manner and assist human researchers with highly-precised predictions in unseen scenarios. The recent rise of Large Language Models (LLMs) and the empowered autonomous agents enable scientists to gain help through interaction in different stages of their research, including but not limited to literature review, research ideation, idea implementation, and academic writing. However, AI researchers instantiated by foundation model empowered agents with full-process autonomy are still in their infancy. In this paper, we study $\textbf{AI-Generated Science}$ (AIGS), where agents independently and autonomously complete the entire research process and discover scientific laws. By revisiting the definition of scientific research, we argue that $\textit{falsification}$ is the essence of both human research process and the design of an AIGS system. Through the lens of falsification, prior systems attempting towards AI-Generated Science either lack the part in their design, or rely heavily on existing verification engines that narrow the use in specialized domains. In this work, we propose Baby-AIGS as a baby-step demonstration of a full-process AIGS system, which is a multi-agent system with agents in roles representing key research process. By introducing FalsificationAgent, which identify and then verify possible scientific discoveries, we empower the system with explicit falsification. Experiments on three tasks preliminarily show that Baby-AIGS could produce meaningful scientific discoveries, though not on par with experienced human researchers. Finally, we discuss on the limitations of current Baby-AIGS, actionable insights, and related ethical issues in detail.

cs.LG↗

arXiv2024

Algorithmics and Complexity of Cost-Driven Task Offloading with Submodular Optimization in Edge-Cloud Environments

Emerging applications such as autonomous driving pose the challenge of efficient cost-driven offloading in edge-cloud environments. This involves assigning tasks to edge and cloud servers for separate execution, with the goal of minimizing the total service cost including communication and computation costs. In this paper, observing that the intra-cloud communication costs are relatively low and can often be neglected in many real-world applications, we consequently introduce the so-called communication assumption which posits that the intra-cloud communication costs are not higher than the inter-partition communication cost between cloud and edge servers, nor the cost among edge servers. As a preliminary analysis, we first prove that the offloading problem without the communication assumption is NP-hard, using a reduction from MAX-CUT. Then, we show that the offloading problem can be modeled as a submodular minimization problem, making it polynomially solvable. Moreover, this polynomial solvability remains even when additional constraints are imposed, such as when certain tasks must be executed on edge servers due to latency constraints. By combining both algorithmics and computational complexity results, we demonstrate that the difficulty of the offloading problem largely depends on whether the communication assumption is satisfied. Lastly, extensive experiments are conducted to evaluate the practical performance of the proposed algorithm, demonstrating its significant advantages over the state-of-the-art methods in terms of efficiency and cost-effectiveness.

cs.DM↗

arXiv2024

Communication-Efficient Sparsely-Activated Model Training via Sequence Migration and Token Condensation

Mixture-of-Experts (MoE) is an emerging technique for scaling large models with sparse activation. MoE models are typically trained in a distributed manner with an expert parallelism scheme, where experts in each MoE layer are distributed across multiple GPUs. However, the default expert parallelism suffers from the heavy network burden due to the all-to-all intermediate data exchange among GPUs before and after the expert run. Some existing works have proposed to reduce intermediate data exchanges by transferring experts to reduce the network loads, however, which would decrease parallelism level of expert execution and make computation inefficient. The weaknesses of existing works motivate us to explore whether it is possible to reduce inter-GPU traffic while maintaining a high degree of expert parallelism. This paper gives a positive response by presenting Luffy, a communication-efficient distributed MoE training system with two new techniques. First, Luffy migrates sequences among GPUs to hide heavy token pulling paths within GPUs and avoid copying experts over GPUs. Second, we propose token condensation that identifies similar tokens and then eliminates redundant transmissions. We implement Luffy based on PyTorch and evaluate its performance on a testbed of 16 V100 GPUs. Luffy system can achieve a speedup of up to 2.73x compared to state-of-the-art MoE training systems.

cs.DC↗

arXiv2024

Interactive Visual Assessment for Text-to-Image Generation Models

Visual generation models have achieved remarkable progress in computer graphics applications but still face significant challenges in real-world deployment. Current assessment approaches for visual generation tasks typically follow an isolated three-phase framework: test input collection, model output generation, and user assessment. These fashions suffer from fixed coverage, evolving difficulty, and data leakage risks, limiting their effectiveness in comprehensively evaluating increasingly complex generation models. To address these limitations, we propose DyEval, an LLM-powered dynamic interactive visual assessment framework that facilitates collaborative evaluation between humans and generative models for text-to-image systems. DyEval features an intuitive visual interface that enables users to interactively explore and analyze model behaviors, while adaptively generating hierarchical, fine-grained, and diverse textual inputs to continuously probe the capability boundaries of the models based on their feedback. Additionally, to provide interpretable analysis for users to further improve tested models, we develop a contextual reflection module that mines failure triggers of test inputs and reflects model potential failure patterns supporting in-depth analysis using the logical reasoning ability of LLM. Qualitative and quantitative experiments demonstrate that DyEval can effectively help users identify max up to 2.56 times generation failures than conventional methods, and uncover complex and rare failure patterns, such as issues with pronoun generation and specific cultural context generation. Our framework provides valuable insights for improving generative models and has broad implications for advancing the reliability and capabilities of visual generation systems across various domains.

cs.CV↗

arXiv2024

A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration

Recent studies show that collaborating multiple large language model (LLM) powered agents is a promising way for task solving. However, current approaches are constrained by using a fixed number of agents and static communication structures. In this work, we propose automatically selecting a team of agents from candidates to collaborate in a dynamic communication structure toward different tasks and domains. Specifically, we build a framework named Dynamic LLM-Powered Agent Network ($\textbf{DyLAN}$) for LLM-powered agent collaboration, operating a two-stage paradigm: (1) Team Optimization and (2) Task Solving. During the first stage, we utilize an $\textit{agent selection}$ algorithm, based on an unsupervised metric called $\textit{Agent Importance Score}$, enabling the selection of best agents according to their contributions in a preliminary trial, oriented to the given task. Then, in the second stage, the selected agents collaborate dynamically according to the query. Empirically, we demonstrate that DyLAN outperforms strong baselines in code generation, decision-making, general reasoning, and arithmetic reasoning tasks with moderate computational cost. On specific subjects in MMLU, selecting a team of agents in the team optimization stage improves accuracy by up to 25.0% in DyLAN.

cs.CL↗