arXiv subjects

Zhen Peng

Publications and source records attributed to Zhen Peng.

At least 37 records · Page 2Linked to original sources

arXiv2025

StarGen: A Spatiotemporal Autoregression Framework with Video Diffusion Model for Scalable and Controllable Scene Generation

Recent advances in large reconstruction and generative models have significantly improved scene reconstruction and novel view generation. However, due to compute limitations, each inference with these large models is confined to a small area, making long-range consistent scene generation challenging. To address this, we propose StarGen, a novel framework that employs a pre-trained video diffusion model in an autoregressive manner for long-range scene generation. The generation of each video clip is conditioned on the 3D warping of spatially adjacent images and the temporally overlapping image from previously generated clips, improving spatiotemporal consistency in long-range scene generation with precise pose control. The spatiotemporal condition is compatible with various input conditions, facilitating diverse tasks, including sparse view interpolation, perpetual view generation, and layout-conditioned city generation. Quantitative and qualitative evaluations demonstrate StarGen's superior scalability, fidelity, and pose accuracy compared to state-of-the-art methods. Project page: https://zju3dv.github.io/StarGen.

cs.CV↗

arXiv2025

Out-of-Distribution Generalization on Graphs via Progressive Inference

The development and evaluation of graph neural networks (GNNs) generally follow the independent and identically distributed (i.i.d.) assumption. Yet this assumption is often untenable in practice due to the uncontrollable data generation mechanism. In particular, when the data distribution shows a significant shift, most GNNs would fail to produce reliable predictions and may even make decisions randomly. One of the most promising solutions to improve the model generalization is to pick out causal invariant parts in the input graph. Nonetheless, we observe a significant distribution gap between the causal parts learned by existing methods and the ground truth, leading to undesirable performance. In response to the above issues, this paper presents GPro, a model that learns graph causal invariance with progressive inference. Specifically, the complicated graph causal invariant learning is decomposed into multiple intermediate inference steps from easy to hard, and the perception of GPro is continuously strengthened through a progressive inference process to extract causal features that are stable to distribution shifts. We also enlarge the training distribution by creating counterfactual samples to enhance the capability of the GPro in capturing the causal invariant parts. Extensive experiments demonstrate that our proposed GPro outperforms the state-of-the-art methods by 4.91% on average. For datasets with more severe distribution shifts, the performance improvement can be up to 6.86%.

cs.LG↗

arXiv2025

Self-supervised Quantized Representation for Seamlessly Integrating Knowledge Graphs with Large Language Models

Due to the presence of the natural gap between Knowledge Graph (KG) structures and the natural language, the effective integration of holistic structural information of KGs with Large Language Models (LLMs) has emerged as a significant question. To this end, we propose a two-stage framework to learn and apply quantized codes for each entity, aiming for the seamless integration of KGs with LLMs. Firstly, a self-supervised quantized representation (SSQR) method is proposed to compress both KG structural and semantic knowledge into discrete codes (\ie, tokens) that align the format of language sentences. We further design KG instruction-following data by viewing these learned codes as features to directly input to LLMs, thereby achieving seamless integration. The experiment results demonstrate that SSQR outperforms existing unsupervised quantized methods, producing more distinguishable codes. Further, the fine-tuned LLaMA2 and LLaMA3.1 also have superior performance on KG link prediction and triple classification tasks, utilizing only 16 tokens per entity instead of thousands in conventional prompting methods.

cs.CL↗

arXiv2024

Has Multimodal Learning Delivered Universal Intelligence in Healthcare? A Comprehensive Survey

The rapid development of artificial intelligence has constantly reshaped the field of intelligent healthcare and medicine. As a vital technology, multimodal learning has increasingly garnered interest due to data complementarity, comprehensive modeling form, and great application potential. Currently, numerous researchers are dedicating their attention to this field, conducting extensive studies and constructing abundant intelligent systems. Naturally, an open question arises that has multimodal learning delivered universal intelligence in healthcare? To answer the question, we adopt three unique viewpoints for a holistic analysis. Firstly, we conduct a comprehensive survey of the current progress of medical multimodal learning from the perspectives of datasets, task-oriented methods, and universal foundation models. Based on them, we further discuss the proposed question from five issues to explore the real impacts of advanced techniques in healthcare, from data and technologies to performance and ethics. The answer is that current technologies have NOT achieved universal intelligence and there remains a significant journey to undertake. Finally, in light of the above reviews and discussions, we point out ten potential directions for exploration towards the goal of universal intelligence in healthcare.

cs.AI↗

arXiv2024

Spatial Structure Supports Diversity in Prebiotic Autocatalytic Chemical Ecosystems

Autocatalysis is thought to have played an important role in the earliest stages of the origin of life. An autocatalytic cycle (AC) is a set of reactions that results in stoichiometric increase in its constituent chemicals. When the reactions of multiple interacting ACs are active in a region of space, they can have interactions analogous to those between species in biological ecosystems. Prior studies of autocatalytic chemical ecosystems (ACEs) have suggested avenues for accumulating complexity, such as ecological succession, as well as obstacles such as competitive exclusion. We extend this ecological framework to investigate the effects of surface adsorption, desorption, and diffusion on ACE ecology. Simulating ACEs as particle-based stochastic reaction-diffusion systems in spatial environments-including open, two-dimensional reaction-diffusion systems and adsorptive mineral surfaces-we demonstrate that spatial structure can enhance ACE diversity by i) permitting otherwise mutually exclusive ACs to coexist and ii) subjecting new AC traits to selection.

q-bio.PE↗

arXiv2024

RayProNet: A Neural Point Field Framework for Radio Propagation Modeling in 3D Environments

The radio wave propagation channel is central to the performance of wireless communication systems. In this paper, we introduce a novel machine learning-empowered methodology for wireless channel modeling. The key ingredients include a point-cloud-based neural network and a Spherical Harmonics encoder with light probes. Our approach offers several significant advantages, including the flexibility to adjust antenna radiation patterns and transmitter/receiver locations, the capability to predict radio power maps, and the scalability of large-scale wireless scenes. As a result, it lays the groundwork for an end-to-end pipeline for network planning and deployment optimization. The proposed work is validated in various outdoor and indoor radio environments.

eess.SP↗

arXiv2023

What it takes to solve the Origin(s) of Life: An integrated review of techniques

Understanding the origin(s) of life (OoL) is a fundamental challenge for science in the 21st century. Research on OoL spans many disciplines, including chemistry, physics, biology, planetary sciences, computer science, mathematics and philosophy. The sheer number of different scientific perspectives relevant to the problem has resulted in the coexistence of diverse tools, techniques, data, and software in OoL studies. This has made communication between the disciplines relevant to the OoL extremely difficult because the interpretation of data, analyses, or standards of evidence can vary dramatically. Here, we hope to bridge this wide field of study by providing common ground via the consolidation of tools and techniques rather than positing a unifying view on how life emerges. We review the common tools and techniques that have been used significantly in OoL studies in recent years. In particular, we aim to identify which information is most relevant for comparing and integrating the results of experimental analyses into mathematical and computational models. This review aims to provide a baseline expectation and understanding of technical aspects of origins research, rather than being a primer on any particular topic. As such, it spans broadly -- from analytical chemistry to mathematical models -- and highlights areas of future work that will benefit from a multidisciplinary approach to tackling the mystery of life's origin. Ultimately, we hope to empower a new generation of OoL scientists by reviewing how they can investigate life's origin, rather than dictating how to think about the problem.

q-bio.PE↗

arXiv2023

Fast gradient method for Low-Rank Matrix Estimation

Projected gradient descent and its Riemannian variant belong to a typical class of methods for low-rank matrix estimation. This paper proposes a new Nesterov's Accelerated Riemannian Gradient algorithm by efficient orthographic retraction and tangent space projection. The subspace relationship between iterative and extrapolated sequences on the low-rank matrix manifold provides a computational convenience. With perturbation analysis of truncated singular value decomposition and two retractions, we systematically analyze the local convergence of gradient algorithms and Nesterov's variants in the Euclidean and Riemannian settings. Theoretically, we estimate the exact rate of local linear convergence under different parameters using the spectral radius in a closed form and give the optimal convergence rate and the corresponding momentum parameter. When the parameter is unknown, the adaptive restart scheme can avoid the oscillation problem caused by high momentum, thus approaching the optimal convergence rate. Extensive numerical experiments confirm the estimations of convergence rate and demonstrate that the proposed algorithm is competitive with first-order methods for matrix completion and matrix sensing.

math.OC↗

arXiv2023

Long-term Visual Localization with Mobile Sensors

Despite the remarkable advances in image matching and pose estimation, image-based localization of a camera in a temporally-varying outdoor environment is still a challenging problem due to huge appearance disparity between query and reference images caused by illumination, seasonal and structural changes. In this work, we propose to leverage additional sensors on a mobile phone, mainly GPS, compass, and gravity sensor, to solve this challenging problem. We show that these mobile sensors provide decent initial poses and effective constraints to reduce the searching space in image matching and final pose estimation. With the initial pose, we are also able to devise a direct 2D-3D matching network to efficiently establish 2D-3D correspondences instead of tedious 2D-2D matching in existing systems. As no public dataset exists for the studied problem, we collect a new dataset that provides a variety of mobile sensor data and significant scene appearance variations, and develop a system to acquire ground-truth poses for query images. We benchmark our method as well as several state-of-the-art baselines and demonstrate the effectiveness of the proposed approach. The code and dataset will be released publicly.

cs.CV↗

arXiv2022

Speed-ANN: Low-Latency and High-Accuracy Nearest Neighbor Search via Intra-Query Parallelism

Nearest Neighbor Search (NNS) has recently drawn a rapid increase of interest due to its core role in managing high-dimensional vector data in data science and AI applications. The interest is fueled by the success of neural embedding, where deep learning models transform unstructured data into semantically correlated feature vectors for data analysis, e.g., recommend popular items. Among several categories of methods for fast NNS, similarity graph is one of the most successful algorithmic trends. Several of the most popular and top-performing similarity graphs, such as NSG and HNSW, at their core employ best-first traversal along the underlying graph indices to search near neighbors. Maximizing the performance of the search is essential for many tasks, especially at the large-scale and high-recall regime. In this work, we provide an in-depth examination of the challenges of the state-of-the-art similarity search algorithms, revealing its challenges in leveraging multi-core processors to speed up the search efficiency. We also exploit whether similarity graph search is robust to deviation from maintaining strict order by allowing multiple walkers to simultaneously advance the search frontier. Based on our insights, we propose Speed-ANN, a parallel similarity search algorithm that exploits hidden intra-query parallelism and memory hierarchy that allows similarity search to take advantage of multiple CPU cores to significantly accelerate search speed while achieving high accuracy. We evaluate Speed-ANN on a wide range of datasets, ranging from million to billion data points, and show its shorter query latency than NSG and HNSW, respectively. Besides, with multicore support, we show that our approach offers faster search latency than highly-optimized GPU implementation and provides good scalability as the increase of the number of hardware resources (e.g., CPU cores) and graph sizes.

cs.DC↗

arXiv2021

Smart Radio Environments

This Roadmap takes the reader on a journey through the research in electromagnetic wave propagation control via reconfigurable intelligent surfaces. Meta-surface modelling and design methods are reviewed along with physical realisation techniques. Several wireless applications are discussed, including beam-forming, focusing, imaging, localisation, and sensing, some rooted in novel architectures for future mobile communications networks towards 6G.

physics.class-ph↗

arXiv2021

Threshold for a discrete-variable sensor of quantum reservoirs

Quantum sensing employs quantum resources of a sensor to attain a smaller estimation error of physical quantities than the limit constrained by classical physics. To measure a quantum reservoir, which is significant in decoherence control, a nonunitary-encoding sensing scheme becomes necessary. However, previous studies showed that the reservoir-induced degradation to quantum resources of the sensor makes the errors divergent with the increase of encoding time. We here propose a scheme to use $N$ two-level systems as the sensor to measure a quantum reservoir. A threshold, above which the shot-noise-limited sensing error saturates or even persistently decreases with the encoding time, is uncovered. Our analysis reveals that it is due to the formation of a bound state of the total sensor-reservoir system. Solving the outstanding error-divergency problem in previous studies, our result supplies an insightful guideline in realizing a sensitive measurement of quantum reservoirs.

quant-ph↗

arXiv2020

Self-Supervised Graph Representation Learning via Global Context Prediction

To take full advantage of fast-growing unlabeled networked data, this paper introduces a novel self-supervised strategy for graph representation learning by exploiting natural supervision provided by the data itself. Inspired by human social behavior, we assume that the global context of each node is composed of all nodes in the graph since two arbitrary entities in a connected network could interact with each other via paths of varying length. Based on this, we investigate whether the global context can be a source of free and effective supervisory signals for learning useful node representations. Specifically, we randomly select pairs of nodes in a graph and train a well-designed neural net to predict the contextual position of one node relative to the other. Our underlying hypothesis is that the representations learned from such within-graph context would capture the global topology of the graph and finely characterize the similarity and differentiation between nodes, which is conducive to various downstream learning tasks. Extensive benchmark experiments including node classification, clustering, and link prediction demonstrate that our approach outperforms many state-of-the-art unsupervised methods and sometimes even exceeds the performance of supervised counterparts.

cs.LG↗

arXiv2020

Graph Representation Learning via Graphical Mutual Information Maximization

The richness in the content of various information networks such as social networks and communication networks provides the unprecedented potential for learning high-quality expressive representations without external supervision. This paper investigates how to preserve and extract the abundant information from graph-structured data into embedding space in an unsupervised manner. To this end, we propose a novel concept, Graphical Mutual Information (GMI), to measure the correlation between input graphs and high-level hidden representations. GMI generalizes the idea of conventional mutual information computations from vector space to the graph domain where measuring mutual information from two aspects of node features and topological structure is indispensable. GMI exhibits several benefits: First, it is invariant to the isomorphic transformation of input graphs---an inevitable constraint in many existing graph representation learning algorithms; Besides, it can be efficiently estimated and maximized by current mutual information estimation methods such as MINE; Finally, our theoretical analysis confirms its correctness and rationality. With the aid of GMI, we develop an unsupervised learning model trained by maximizing GMI between the input and output of a graph neural encoder. Considerable experiments on transductive as well as inductive node classification and link prediction demonstrate that our method outperforms state-of-the-art unsupervised counterparts, and even sometimes exceeds the performance of supervised ones.

cs.LG↗

arXiv2020

An ecological framework for the analysis of prebiotic chemical reaction networks and their dynamical behavior

It is becoming widely accepted that very early in the origin of life, even before the emergence of genetic encoding, reaction networks of diverse small chemicals might have manifested key properties of life, namely self-propagation and adaptive evolution. To explore this possibility, we formalize the dynamics of chemical reaction networks within the framework of chemical ecosystem ecology. To capture the idea that life-like chemical systems are maintained out of equilibrium by fluxes of energy-rich food chemicals, we model chemical ecosystems in well-mixed containers that are subject to constant dilution by a solution with a fixed concentration of food chemicals. Modelling all chemical reactions as fully reversible, we show that seeding an autocatalytic cycle (AC) with tiny amounts of one or more of its member chemicals results in logistic growth of all member chemicals in the cycle. This finding justifies drawing an instructive analogy between an AC and the population of a biological species. We extend this finding to show that pairs of ACs can show competitive, predator-prey, or mutualistic associations just like biological species. Furthermore, when there is stochasticity in the environment, particularly in the seeding of ACs, chemical ecosystems can show complex dynamics that can resemble evolution. The evolutionary character is especially clear when the network architecture results in ecological precedence (survival of the first), which makes the path of succession historically contingent on the order in which cycles are seeded. For all its simplicity, the framework developed here is helpful for visualizing how autocatalysis in prebiotic chemical reaction networks can yield life-like properties. Furthermore, chemical ecosystem ecology could provide a useful foundation for exploring the emergence of adaptive dynamics and the origins of polymer-based genetic systems.

q-bio.PE↗

arXiv2019

Pruned Landmark Labeling Meets Vertex Centric Computation: A Surprisingly Happy Marriage!

In this paper, we study how the Pruned Landmark Labeling (PPL) algorithm can be parallelized in a scalable fashion, producing the same results as the sequential algorithm. More specifically, we parallelize using a Vertex-Centric (VC) computational model on a modern SIMD powered multicore architecture. We design a new VC-PLL algorithm that resolves the apparent mismatch between the inherent sequential dependence of the PLL algorithm and the Vertex- Centric (VC) computing model. Furthermore, we introduce a novel batch execution model for VC computation and the BVC-PLL algorithm to reduce the computational inefficiency in VC-PLL. Quite surprisingly, the theoretical analysis reveals that under a reasonable assumption, BVC-PLL has lower computational and memory access costs than PLL and indicates it may run faster than PLL as a sequential algorithm. We also demonstrate how BVC-PLL algorithm can be extended to handle directed graphs and weighted graphs and how it can utilize the hierarchical parallelism on a modern parallel computing architecture. Extensive experiments on real-world graphs not only show the sequential BVC-PLL can run more than two times faster than the original PLL, but also demonstrates its parallel efficiency and scalability.

cs.DB↗

arXiv2019

Retrieving ideal precision in noisy quantum optical metrology

Quantum metrology employs quantum effects to attain a measurement precision surpassing the limit achievable in classical physics. However, it was previously found that the precision returns the shot-noise limit (SNL) from the ideal Zeno limit (ZL) due to the photon loss in quantum metrology based on Mech-Zehnder interferometer. Here, we find that not only the SNL can be beaten, but also the ZL can be asymptotically recovered in long-encoding-time condition when the photon dissipation is exactly studied in its inherent non-Markovian manner. Our analysis reveals that it is due to the formation of a bound state of the photonic system and its dissipative noise. Highlighting the microscopic mechanism of the dissipative noise on the quantum optical metrology, our result supplies a guideline to realize the ultrasensitive measurement in practice by forming the bound state in the setting of reservoir engineering.

quant-ph↗

arXiv2018

Cohomology rings, differential graded algebras and derived equivalences

In this paper, we will consider derived equivalences for differential graded endomorphism algebras by Keller's approaches. First we construct derived equivalences of differential graded algebras which are endomorphism algebras of the objects from a triangle in the homotopy category of differential graded algebras. We also obtain derived equivalences of differential graded endomorphism algebras from a standard derived equivalence of finite dimensional algebras. Moreover, under some conditions, the cohomology rings of these differential graded endomorphism algebras are also derived equivalent. Then we give an affirmative answer to a problem of Dugas \cite{Dugas2015} in some special case.

math.RT↗