arXiv subjects

Yu Chen

Publications and source records attributed to Yu Chen.

At least 289 records · Page 16Linked to original sources

arXiv2024

Minimum-Variance Recursive State Estimation for 2-D Systems: When Asynchronous Multi-Channel Delays meet Energy Harvesting Constraints

This paper is concerned with the state estimation problem for two-dimensional systems with asynchronous multichannel delays and energy harvesting constraints. In the system, each smart sensor has a certain probability of harvesting energy from the external environment, the authorized transmission between the sensor and the remote filter is contingent upon the current energy level of the sensor, which results in intermittent transmission of observation information. Addressing the issue of incomplete observation information due to asynchronous multi-channel delays, a novel approach for observation partition reconstruction is proposed to convert the delayed activated observation sequences into equivalent delay-free activated observation sequences. Through generating spatial equivalency validation, it is found that the reconstructed delay-free activated observation sequences contain the same information as the original delayed activated observation sequences. Based on the reconstructed activated observation sequence and activated probability, a novel unbiased h+1-step recursive estimator is constructed. Then, the evolution of the probability distribution of the energy level is discussed. The estimation gains are obtained by minimizing the filtering error covariance. Subsequently, through parameter assumptions, a uniform lower bound and a recursive upper bound for the filtering error covariance are presented. And the monotonicity analysis of activated probability on estimation performance is given. Finally, the effectiveness of the proposed estimation scheme is verified through a numerical simulation example.

eess.SY↗

arXiv2024

HAGAN: Hybrid Augmented Generative Adversarial Network for Medical Image Synthesis

Medical Image Synthesis (MIS) plays an important role in the intelligent medical field, which greatly saves the economic and time costs of medical diagnosis. However, due to the complexity of medical images and similar characteristics of different tissue cells, existing methods face great challenges in meeting their biological consistency. To this end, we propose the Hybrid Augmented Generative Adversarial Network (HAGAN) to maintain the authenticity of structural texture and tissue cells. HAGAN contains Attention Mixed (AttnMix) Generator, Hierarchical Discriminator and Reverse Skip Connection between Discriminator and Generator. The AttnMix consistency differentiable regularization encourages the perception in structural and textural variations between real and fake images, which improves the pathological integrity of synthetic images and the accuracy of features in local areas. The Hierarchical Discriminator introduces pixel-by-pixel discriminant feedback to generator for enhancing the saliency and discriminance of global and local details simultaneously. The Reverse Skip Connection further improves the accuracy for fine details by fusing real and synthetic distribution features. Our experimental evaluations on three datasets of different scales, i.e., COVID-CT, ACDC and BraTS2018, demonstrate that HAGAN outperforms the existing methods and achieves state-of-the-art performance in both high-resolution and low-resolution.

eess.IV↗

arXiv2024

Quantitative Evaluation of driver's situation awareness in virtual driving through Eye tracking analysis

In driving tasks, the driver's situation awareness of the surrounding scenario is crucial for safety driving. However, current methods of measuring situation awareness mostly rely on subjective questionnaires, which interrupt tasks and lack non-intrusive quantification. To address this issue, our study utilizes objective gaze motion data to provide an interference-free quantification method for situation awareness. Three quantitative scores are proposed to represent three different levels of awareness: perception, comprehension, and projection, and an overall score of situation awareness is also proposed based on above three scores. To validate our findings, we conducted experiments where subjects performed driving tasks in a virtual reality simulated environment. All the four proposed situation awareness scores have clearly shown a significant correlation with driving performance. The proposed not only illuminates a new path for understanding and evaluating the situation awareness but also offers a satisfying proxy for driving performance.

cs.HC↗

arXiv2024

Event-Triggered Resilient Filtering for 2-D Systems with Asynchronous-Delay: Handling Binary Encoding Decoding with Probabilistic Bit Flips

In this paper, the event-triggered resilient filtering problem is investigated for a class of two-dimensional systems with asynchronous-delay under binary encoding-decoding schemes with probabilistic bit flips. To reduce unnecessary communications and computations in complex network systems, alleviate network energy consumption, and optimize the use of network resources, a new event-triggered mechanism is proposed, which focuses on broadcasting necessary measurement information to update innovation only when the event generator function is satisfied. A binary encoding-decoding scheme is used in the communication process to quantify the measurement information into a bit stream, transmit it through a memoryless binary symmetric channel with a certain probability of bit flipping, and restore it at the receiver. In order to utilize the delayed decoded measurement information that a measurement reconstruction approach is proposed. Through generating space equivalence verification, it is found that the reconstructed delay-free decoded measurement sequence contains the same information as the original delayed decoded measurement sequence. In addition, resilient filter is utilized to accommodate possible estimation gain perturbations. Then, a recursive estimator framework is presented based on the reconstructed decoded measurement sequence. By means of the mathematical induction technique, the unbiased property of the proposed estimator is proved. The estimation gain is obtained by minimizing an upper bound on the filtering error covariance. Subsequently, through rigorous mathematical analysis, the monotonicity of filtering performance with respect to triggering parameters is discussed.

cs.IT↗

arXiv2024

Adapting Large Language Models by Integrating Collaborative Semantics for Recommendation

Recently, large language models (LLMs) have shown great potential in recommender systems, either improving existing recommendation models or serving as the backbone. However, there exists a large semantic gap between LLMs and recommender systems, since items to be recommended are often indexed by discrete identifiers (item ID) out of the LLM's vocabulary. In essence, LLMs capture language semantics while recommender systems imply collaborative semantics, making it difficult to sufficiently leverage the model capacity of LLMs for recommendation. To address this challenge, in this paper, we propose a new LLM-based recommendation model called LC-Rec, which can better integrate language and collaborative semantics for recommender systems. Our approach can directly generate items from the entire item set for recommendation, without relying on candidate items. Specifically, we make two major contributions in our approach. For item indexing, we design a learning-based vector quantization method with uniform semantic mapping, which can assign meaningful and non-conflicting IDs (called item indices) for items. For alignment tuning, we propose a series of specially designed tuning tasks to enhance the integration of collaborative semantics in LLMs. Our fine-tuning tasks enforce LLMs to deeply integrate language and collaborative semantics (characterized by the learned item indices), so as to achieve an effective adaptation to recommender systems. Extensive experiments demonstrate the effectiveness of our method, showing that our approach can outperform a number of competitive baselines including traditional recommenders and existing LLM-based recommenders. Our code is available at https://github.com/RUCAIBox/LC-Rec/.

cs.IR↗

arXiv2024

DualFluidNet: an Attention-based Dual-pipeline Network for FLuid Simulation

Fluid motion can be considered as a point cloud transformation when using the SPH method. Compared to traditional numerical analysis methods, using machine learning techniques to learn physics simulations can achieve near-accurate results, while significantly increasing efficiency. In this paper, we propose an innovative approach for 3D fluid simulations utilizing an Attention-based Dual-pipeline Network, which employs a dual-pipeline architecture, seamlessly integrated with an Attention-based Feature Fusion Module. Unlike previous methods, which often make difficult trade-offs between global fluid control and physical law constraints, we find a way to achieve a better balance between these two crucial aspects with a well-designed dual-pipeline approach. Additionally, we design a Type-aware Input Module to adaptively recognize particles of different types and perform feature fusion afterward, such that fluid-solid coupling issues can be better dealt with. Furthermore, we propose a new dataset, Tank3D, to further explore the network's ability to handle more complicated scenes. The experiments demonstrate that our approach not only attains a quantitative enhancement in various metrics, surpassing the state-of-the-art methods but also signifies a qualitative leap in neural network-based simulation by faithfully adhering to the physical laws. Code and video demonstrations are available at https://github.com/chenyu-xjtu/DualFluidNet.

cs.CV↗

arXiv2024

Towards the Characterization of Terminal Cut Functions: a Condition for Laminar Families

We study the following characterization problem. Given a set $T$ of terminals and a $(2^{|T|}-2)$-dimensional vector $π$ whose coordinates are indexed by proper subsets of $T$, is there a graph $G$ that contains $T$, such that for all subsets $\emptyset\subsetneq S\subsetneq T$, $π_S$ equals the value of the min-cut in $G$ separating $S$ from $T\setminus S$? The only known necessary conditions are submodularity and a special class of linear inequalities given by Chaudhuri, Subrahmanyam, Wagner and Zaroliagis. Our main result is a new class of linear inequalities concerning laminar families, that generalize all previous ones. Using our new class of inequalities, we can generalize Karger's approximate min-cut counting result to graphs with terminals.

cs.DS↗

arXiv2024

Ultralarge polarization in ferroelectric hafnia-based thin films

Hafnia-based ferroelectrics have become a valuable class of electronic functional materials at the nanoscale, showing great potential for next-generation memory and logic devices. However, more robust ferroelectric properties and better understanding of the polarization mechanisms are currently needed both in technology and science. Herein, we report the properties of oxygen-deficient Hf0.5Zr0.5O2 films with ultralarge remanent polarization (Pr) of 387 uC cm-2 at room temperature (1 kHz). Structure characterizations identify a new ferroelectric monoclinic Pc phase in these Hf0.5Zr0.5O2 films. The in-situ STEM measurements evidence polar displacements of the oxygen atoms, which move up and down in the Pc structure under applied DC bias fields, showing a huge displacement (1.6 A). DFT calculations optimized the Pc structure and also predicted a large polarization. The coexistence of the ferroelectric monoclinic (Pc) phases and orthorhombic (Pca21) is responsible for this superior ferroelectric properties. These findings are promising for hafnia-based ferroelectric applications in integrated ferroelectric devices, energy harvesting and actuators, etc.

cond-mat.mtrl-sci↗

arXiv2024

Model Independent Approach of the JUNO $^8$B Solar Neutrino Program

The physics potential of detecting $^8$B solar neutrinos will be exploited at the Jiangmen Underground Neutrino Observatory (JUNO), in a model independent manner by using three distinct channels of the charged-current (CC), neutral-current (NC) and elastic scattering (ES) interactions. Due to the largest-ever mass of $^{13}$C nuclei in the liquid-scintillator detectors and the {expected} low background level, $^8$B solar neutrinos would be observable in the CC and NC interactions on $^{13}$C for the first time. By virtue of optimized event selections and muon veto strategies, backgrounds from the accidental coincidence, muon-induced isotopes, and external backgrounds can be greatly suppressed. Excellent signal-to-background ratios can be achieved in the CC, NC and ES channels to guarantee the $^8$B solar neutrino observation. From the sensitivity studies performed in this work, we show that JUNO, with ten years of data, can reach the {1$σ$} precision levels of 5%, 8% and 20% for the $^8$B neutrino flux, $\sin^2θ_{12}$, and $Δm^2_{21}$, respectively. It would be unique and helpful to probe the details of both solar physics and neutrino physics. In addition, when combined with SNO, the world-best precision of 3% is expected for the $^8$B neutrino flux measurement.

hep-ex↗

arXiv2024

BSDP: Brain-inspired Streaming Dual-level Perturbations for Online Open World Object Detection

Humans can easily distinguish the known and unknown categories and can recognize the unknown object by learning it once instead of repeating it many times without forgetting the learned object. Hence, we aim to make deep learning models simulate the way people learn. We refer to such a learning manner as OnLine Open World Object Detection(OLOWOD). Existing OWOD approaches pay more attention to the identification of unknown categories, while the incremental learning part is also very important. Besides, some neuroscience research shows that specific noises allow the brain to form new connections and neural pathways which may improve learning speed and efficiency. In this paper, we take the dual-level information of old samples as perturbations on new samples to make the model good at learning new knowledge without forgetting the old knowledge. Therefore, we propose a simple plug-and-play method, called Brain-inspired Streaming Dual-level Perturbations(BSDP), to solve the OLOWOD problem. Specifically, (1) we first calculate the prototypes of previous categories and use the distance between samples and the prototypes as the sample selecting strategy to choose old samples for replay; (2) then take the prototypes as the streaming feature-level perturbations of new samples, so as to improve the plasticity of the model through revisiting the old knowledge; (3) and also use the distribution of the features of the old category samples to generate adversarial data in the form of streams as the data-level perturbations to enhance the robustness of the model to new categories. We empirically evaluate BSDP on PASCAL VOC and MS-COCO, and the excellent results demonstrate the promising performance of our proposed method and learning manner.

cs.CV↗

arXiv2024

Provably Efficient Partially Observable Risk-Sensitive Reinforcement Learning with Hindsight Observation

This work pioneers regret analysis of risk-sensitive reinforcement learning in partially observable environments with hindsight observation, addressing a gap in theoretical exploration. We introduce a novel formulation that integrates hindsight observations into a Partially Observable Markov Decision Process (POMDP) framework, where the goal is to optimize accumulated reward under the entropic risk measure. We develop the first provably efficient RL algorithm tailored for this setting. We also prove by rigorous analysis that our algorithm achieves polynomial regret $\tilde{O}\left(\frac{e^{|γ|H}-1}{|γ|H}H^2\sqrt{KHS^2OA}\right)$, which outperforms or matches existing upper bounds when the model degenerates to risk-neutral or fully observable settings. We adopt the method of change-of-measure and develop a novel analytical tool of beta vectors to streamline mathematical derivations. These techniques are of particular interest to the theoretical study of reinforcement learning.

cs.LG↗

arXiv2024

Provable Risk-Sensitive Distributional Reinforcement Learning with General Function Approximation

In the realm of reinforcement learning (RL), accounting for risk is crucial for making decisions under uncertainty, particularly in applications where safety and reliability are paramount. In this paper, we introduce a general framework on Risk-Sensitive Distributional Reinforcement Learning (RS-DisRL), with static Lipschitz Risk Measures (LRM) and general function approximation. Our framework covers a broad class of risk-sensitive RL, and facilitates analysis of the impact of estimation functions on the effectiveness of RSRL strategies and evaluation of their sample complexity. We design two innovative meta-algorithms: \texttt{RS-DisRL-M}, a model-based strategy for model-based function approximation, and \texttt{RS-DisRL-V}, a model-free approach for general value function approximation. With our novel estimation techniques via Least Squares Regression (LSR) and Maximum Likelihood Estimation (MLE) in distributional RL with augmented Markov Decision Process (MDP), we derive the first $\widetilde{\mathcal{O}}(\sqrt{K})$ dependency of the regret upper bound for RSRL with static LRM, marking a pioneering contribution towards statistically efficient algorithms in this domain.

cs.LG↗

arXiv2024

Asymptotics of the determinant of the modified Bessel functions and the second Painlevé equation

In the paper, we consider the extended Gross-Witten-Wadia unitary matrix model by introducing a logarithmic term in the potential. The partition function of the model can be expressed equivalently in terms of the Toeplitz determinant with the $(i,j)$-entry being the modified Bessel functions of order $i-j-ν$, $ν\in\mathbb{C}$. When the degree $n$ is finite, we show that the Toeplitz determinant is described by the isomonodromy $τ$-function of the Painlevé III equation. As a double scaling limit, %In the double scaling limit as the degree $n\to\infty$, we establish an asymptotic approximation of the logarithmic derivative of the Toeplitz determinant, expressed in terms of the Hastings-McLeod solution of the inhomogeneous Painlevé II equation with parameter $ν+\frac{1}{2}$. The asymptotics of the leading coefficient and recurrence coefficient of the associated orthogonal polynomials are also derived. We obtain the results by applying the Deift-Zhou nonlinear steepest descent method to the Riemann-Hilbert problem for orthogonal polynomials on the Hankel loop. The main concern here is the construction of a local parametrix at the critical point $z=-1$, where the $ψ$-function of the Jimbo-Miwa Lax pair for the inhomogeneous Painlevé II equation is involved.

math-ph↗

arXiv2024

Scanning tunnelling microscopy and X-ray photoemission studies of NdNiO2 infinite-layer nickelates films

We report scanning tunnelling microscopy (STM) and x-ray photoemission spectroscopy (XPS) measurements on uncapped and SrTiO3 (STO) capped NdNiO2 realized by pulsed-laser deposition and topotactic reduction process. We find that untreated NdNiO2 surfaces are insulating and contain Ni mostly in a nominal Ni2+ oxidation state. Room temperature STM shows signatures of a striped-like topographic pattern, possibly compatible with recent reports of ordered oxygen vacancies in uncapped nickelates due to incomplete oxygen de-intercalation of the upper layers. A metallic surface and a full Ni1+ oxidation state is recovered by ultra high vacuum annealing at 250 C, as shown by STM and XPS. STO-capped NdNiO2 films, on the other hand, show Ni mostly in Ni1+ configuration, but Nd 3d5/2 core level spectra have a relevant contribution from ligand 4f 4L states, suggesting the formation of a NdTiNiOx layer at the interface with the STO. By in situ unit-cell by unit-cell Ar-ion sputtering removal of the STO capping and of the non stoichiometric interface layer, we were able to address the surface electronic properties of these samples, as shown by high resolution valence band photoemission spectroscopy. The results provide insights in the properties of infinite-layer NdNiO2 thin films prepared by the CaH2 topotactic reduction of perovskite NdNiO3 and suggest methods to improve their surface quality.

cond-mat.supr-con↗

arXiv2024

Floorplanning of VLSI by Mixed-Variable Optimization

By formulating the floorplanning of VLSI as a mixed-variable optimization problem, this paper proposes to solve it by memetic algorithms, where the discrete orientation variables are addressed by the distribution evolutionary algorithm based on a population of probability model (DEA-PPM), and the continuous coordination variables are optimized by the conjugate sub-gradient algorithm (CSA). Accordingly, the fixed-outline floorplanning algorithm based on CSA and DEA-PPM (FFA-CD) and the floorplanning algorithm with golden section strategy (FA-GSS) are proposed for the floorplanning problems with and without fixed-outline constraint. %FF-CD is committed to optimizing wirelength targets within a fixed profile. FA-GSS uses the Golden Section strategy to optimize both wirelength and area targets. The CSA is used to solve the proposed non-smooth optimization model, and the DEA-PPM is used to explore the module rotation scheme to enhance the flexibility of the algorithm. Numerical experiments on GSRC test circuits show that the proposed algorithms are superior to some celebrated B*-tree based floorplanning algorithms, and are expected to be applied to large-scale floorplanning problems due to their low time complexity.

cs.NE↗

arXiv2024

Learning based numerical methods for Helmholtz equation with high frequency

High-frequency issues have been remarkably challenges in numerical methods for partial differential equations. In this paper, a learning based numerical method (LbNM) is proposed for Helmholtz equation with high frequency. The main novelty is using Tikhonov regularization method to stably learn the solution operator by utilizing relevant information especially the fundamental solutions. Then applying the solution operator to a new boundary input could quickly update the solution. Based on the method of fundamental solutions and the quantitative Runge approximation, we give the error estimate. This indicates interpretability and generalizability of the present method. Numerical results validates the error analysis and demonstrates the high-precision and high-efficiency features.

math.NA↗

arXiv2024

Lower Bounds on $0$-Extension with Steiner Nodes

In the $0$-Extension problem, we are given an edge-weighted graph $G=(V,E,c)$, a set $T\subseteq V$ of its vertices called terminals, and a semi-metric $D$ over $T$, and the goal is to find an assignment $f$ of each non-terminal vertex to a terminal, minimizing the sum, over all edges $(u,v)\in E$, the product of the edge weight $c(u,v)$ and the distance $D(f(u),f(v))$ between the terminals that $u,v$ are mapped to. Current best approximation algorithms on $0$-Extension are based on rounding a linear programming relaxation called the \emph{semi-metric LP relaxation}. The integrality gap of this LP, with best upper bound $O(\log |T|/\log\log |T|)$ and best lower bound $Ω((\log |T|)^{2/3})$, has been shown to be closely related to the best quality of cut and flow vertex sparsifiers. We study a variant of the $0$-Extension problem where Steiner vertices are allowed. Specifically, we focus on the integrality gap of the same semi-metric LP relaxation to this new problem. Following from previous work, this new integrality gap turns out to be closely related to the quality achievable by cut/flow vertex sparsifiers with Steiner nodes, a major open problem in graph compression. Our main result is that the new integrality gap stays superconstant $Ω(\log\log |T|)$ even if we allow a super-linear $O(|T|\log^{1-\varepsilon}|T|)$ number of Steiner nodes.

cs.DS↗

arXiv2024

FedEmb: A Vertical and Hybrid Federated Learning Algorithm using Network And Feature Embedding Aggregation

Federated learning (FL) is an emerging paradigm for decentralized training of machine learning models on distributed clients, without revealing the data to the central server. The learning scheme may be horizontal, vertical or hybrid (both vertical and horizontal). Most existing research work with deep neural network (DNN) modelling is focused on horizontal data distributions, while vertical and hybrid schemes are much less studied. In this paper, we propose a generalized algorithm FedEmb, for modelling vertical and hybrid DNN-based learning. The idea of our algorithm is characterised by higher inference accuracy, stronger privacy-preserving properties, and lower client-server communication bandwidth demands as compared with existing work. The experimental results show that FedEmb is an effective method to tackle both split feature & subject space decentralized problems, shows 0.3% to 4.2% inference accuracy improvement with limited privacy revealing for datasets stored in local clients, and reduces 88.9 % time complexity over vertical baseline method.

cs.LG↗