Searcharxiv⌕ Search

arXiv subjects

Zhiyuan Wang

Publications and source records attributed to Zhiyuan Wang.

At least 127 records · Page 7Linked to original sources
arXiv2024

HyKGE: A Hypothesis Knowledge Graph Enhanced Framework for Accurate and Reliable Medical LLMs Responses

In this paper, we investigate the retrieval-augmented generation (RAG) based on Knowledge Graphs (KGs) to improve the accuracy and reliability of Large Language Models (LLMs). Recent approaches suffer from insufficient and repetitive knowledge retrieval, tedious and time-consuming query parsing, and monotonous knowledge utilization. To this end, we develop a Hypothesis Knowledge Graph Enhanced (HyKGE) framework, which leverages LLMs' powerful reasoning capacity to compensate for the incompleteness of user queries, optimizes the interaction process with LLMs, and provides diverse retrieved knowledge. Specifically, HyKGE explores the zero-shot capability and the rich knowledge of LLMs with Hypothesis Outputs to extend feasible exploration directions in the KGs, as well as the carefully curated prompt to enhance the density and efficiency of LLMs' responses. Furthermore, we introduce the HO Fragment Granularity-aware Rerank Module to filter out noise while ensuring the balance between diversity and relevance in retrieved knowledge. Experiments on two Chinese medical multiple-choice question datasets and one Chinese open-domain medical Q&A dataset with two LLM turbos demonstrate the superiority of HyKGE in terms of accuracy and explainability.

cs.CL↗
arXiv2024

Sensitivity Assessment of Multi-Criteria Decision-Making Methods in Chemical Engineering Optimization Applications

This chapter assesses the sensitivity of multi-criteria decision-making (MCDM) methods to modifications within the decision or objective matrix (DOM) in the context of chemical engineering optimization applications. Employing eight common or recent MCDM methods and three weighting methods, this study evaluates the impact of three specific DOM alterations: linear transformation of an objective (LTO), reciprocal objective reformulation (ROR), and the removal of alternatives (RA). Our comprehensive analysis reveals that the weights generated by entropy method are more sensitive to the examined modifications compared to the criteria importance through intercriteria correlation (CRITIC) and standard deviation (StDev) methods. ROR is found to have the largest effect on the ranking of alternatives. Moreover, certain methods, gray relational analysis (GRA) without any weights, multi-attributive border approximation area comparison (MABAC), combinative distance-based assessment (CODAS), and simple additive weighting (SAW) with entropy or CRITIC weights, and CODAS, SAW, and technique for order of preference by similarity to ideal solution (TOPSIS) with StDev weight are more robust to DOM modifications. This investigation not only corroborates the findings from the previous study, but also offers insights into the stability and reliability of MCDM methods in the context of chemical engineering.

physics.chem-ph↗
arXiv2024

Determinants of successful disease control through voluntary quarantine dynamics on social networks

In the wake of epidemics, quarantine measures are typically recommended by health authorities or governments to help control the spread of the disease. Compared with mandatory quarantine, voluntary quarantine offers individuals the liberty to decide whether to isolate themselves in case of infection exposure, driven by their personal assessment of the trade-off between economic loss and health risks as well as their own sense of social responsibility and concern for public health. To better understand self-motivated health behavior choices under these factors, here we incorporate voluntary quarantine into an endemic disease model -- the susceptible-infected-susceptible (SIS) model -- and perform comprehensive agent-based simulations to characterize the resulting behavior-disease interactions in structured populations. We quantify the conditions under which voluntary quarantine will be an effective intervention measure to mitigate disease burden. Furthermore, we demonstrate how individual decision-making factors, including the level of temptation to refrain from quarantine and the degree of social compassion, impact compliance levels of voluntary quarantines and the consequent collective disease mitigation efforts. We find that successful disease control requires either a sufficiently low level of temptation or a sufficiently high degree of social compassion, such that even complete containment of the epidemic is attainable. In addition to well-mixed populations, our simulation results are applicable to other more realistic social networks of contacts, including spatial lattices, small-world networks, and real social networks. Our work offers new insights into the fundamental social dilemma aspect of disease control through non-pharmaceutical interventions, such as voluntary quarantine and isolation, where the collective outcome of individual decision-making is crucial.

physics.soc-ph↗
arXiv2024

Hierarchical generalization of dual unitarity

Quantum dynamics with local interactions in lattice models display rich physics, but is notoriously hard to study. Dual-unitary circuits allow for exact answers to interesting physical questions in clean or disordered one- and higher-dimensional quantum systems. However, this family of models shows some non-universal features, like vanishing correlations inside the light-cone and instantaneous thermalization of local observables. In this work we propose a generalization of dual-unitary circuits where the exactly calculable spatial-temporal correlation functions display richer behavior, and have non-trivial thermalization of local observables. This is achieved by generalizing the single-gate condition to a hierarchy of multi-gate conditions, where the first level recovers dual-unitary models, and the second level exhibits these new interesting features. We also extend the discussion and provide exact solutions to correlators with few-site observables and discuss higher-orders, including the ones after a quantum quench. In addition, we provide exhaustive parametrizations for qubit cases, and propose a new family of models for local dimensions larger than two, which also provides a new family of dual-unitary models.

quant-ph↗
arXiv2024

Mapping electrostatic potential in electrolyte solution

Mapping the electrostatic potential (ESP) distribution around ions in electrolyte solution is crucial for the establishment of a microscopic understanding of electrolyte solution properties. For solutions in the bulk phase, it has not been possible to measure the ESP distribution on Angstrom scale. Here we show that liquid electron scattering experiment using state-of-the-art relativistic electron beam can be used to measure the Debye screening length of aqueous LiCl, KCl, and KI solutions across a wide range of concentrations. We observe that the Debye screening length is long-ranged at low concentration and short-ranged at high concentration, providing key insight into the decades-long debate over whether the impact of ions in water is long-ranged or short-ranged. In addition, we show that the measured ESP can be used to retrieve the non-local dielectric function of electrolyte solution, which can serve as a promising route to investigate the electrostatic origin of special ion effects. Our observations show that, interaction, as one of the two fundamental perspectives for understanding electrolyte solution, can provide much richer information than structure.

cond-mat.soft↗
arXiv2024

Möbius Inversion and Duality for Summations of Stable Graphs

Using the stratifications of Deligne-Mumford moduli spaces $\overline{\mathcal M}_{g,n}$ indexed by stable graphs, we introduce a partially ordered set of stable graphs by defining a partial ordering on the set of connected stable graphs of genus $g$ with $n$ external edges. By modifying the usual definition of zeta function and Möbius function of a poset, we introduce generalized ($\mathbb Q$-valued) zeta function and generalized ($\mathbb Q$-valued) Möbius function of the poset of stable graphs. We use them to proved a generalized Möbius inversion formula for functions on the poset of stable graphs. Two applications related to duality in earlier work are also presented.

math.CO↗
arXiv2023

Characterization of an $\rm ^{27}Al^+$ ion optical clock laser with three independent methods

We report on the development and performance evaluation of an ultra-stable clock laser for an $\rm ^{27}Al^+$ optical clock. The thermal noise limited ultra-stable laser is developed based on a 30 cm long ultra-stable cavity. Three independent evaluation methods, including the frequency noise summation method, the three-cornered hat (TCH) method, and the optical clock transition detection method, are used to evaluate the clock laser performance. The summation result of various frequency noise terms is compared with the result of the TCH method. In addition, the $\rm ^{27}Al^+$ ion optical clock transition with ultra-narrow linewidth is also used to detect the frequency noise of the laser at lower Fourier frequencies. The results of the three methods show good agreements, showing a frequency instability level of $1.3\times10^{-16}$, and giving us confidence that these evaluation methods may provides guidance for accurate evaluations of high stability laser sources.

physics.atom-ph↗
arXiv2023

Scalable CP Decomposition for Tensor Learning using GPU Tensor Cores

CP decomposition is a powerful tool for data science, especially gene analysis, deep learning, and quantum computation. However, the application of tensor decomposition is largely hindered by the exponential increment of the computational complexity and storage consumption with the size of tensors. While the data in our real world is usually presented as trillion- or even exascale-scale tensors, existing work can only support billion-scale scale tensors. In our work, we propose the Exascale-Tensor to mitigate the significant gap. Specifically, we propose a compression-based tensor decomposition framework, namely the exascale-tensor, to support exascale tensor decomposition. Then, we carefully analyze the inherent parallelism and propose a bag of strategies to improve computational efficiency. Last, we conduct experiments to decompose tensors ranging from million-scale to trillion-scale for evaluation. Compared to the baselines, the exascale-tensor supports 8,000x larger tensors and a speedup up to 6.95x. We also apply our method to two real-world applications, including gene analysis and tensor layer neural networks, of which the numeric results demonstrate the scalability and effectiveness of our method.

cs.LG↗
arXiv2023

Bright nonblinking photoluminescence with blinking lifetime from a nanocavity-coupled quantum dot

Colloidal semiconductor quantum dots (QDs) are excellent luminescent nanomaterials for a broad range of optoelectronic applications. Their photoluminescence blinking, however, hinders their practical use in many aspects. It has been shown that coupling QDs to plasmonic nanostructures may provide a viable way to suppress blinking. Nevertheless, the underlying mechanism of blinking suppression remains unclear and debated. Here, by deterministically coupling a single QD to a plasmonic nanocavity, we clarify the mechanism of blinking suppression, and demonstrate unprecedentedly bright emission from a single colloidal QD. In particular, we report for the first time that the coupled system exhibits nonblinking photoluminescence with blinking lifetime, which shows that the elimination of photoluminescence blinking originates from enhanced quantum yield of the charged states. We identify that the radiative decay rate is boosted from (48 ns)-1 to (0.7 ns)-1, which outcompetes Auger processes and enables similar quantum yields for charged and neutral excitons. Moreover, we demonstrate ultrabright photoluminescence of up to 17 million detected photons per second from a single QD. This work sheds new light on the goal of achieving ultrabright nonblinking QDs and may benefit a variety of QD-based applications.

physics.optics↗
arXiv2023

Diagonal Tau-Functions of 2D Toda Lattice Hierarchy, Connected $(n,m)$-Point Functions, and Double Hurwitz Numbers

We derive an explicit formula for the connected $(n,m)$-point functions associated to an arbitrary diagonal tau-function $τ_f(\boldsymbol{t}^+,\boldsymbol{t}^-)$ of the 2d Toda lattice hierarchy using fermionic computations and the boson-fermion correspondence. Then for fixed $\boldsymbol{t}^-$, we compute the KP-affine coordinates of $τ_f(\boldsymbol{t}^+,\boldsymbol{t}^-)$. As applications, we present a unified approach to compute various types of connected double Hurwitz numbers, including the ordinary double Hurwitz numbers, the double Hurwitz numbers with completed $r$-cycles, and the mixed double Hurwitz numbers. We also apply this method to the computation of the stationary Gromov-Witten invariants of $\mathbb P^1$ relative to two points.

nlin.SI↗
arXiv2023

LPN: Language-guided Prototypical Network for few-shot classification

Few-shot classification aims to adapt to new tasks with limited labeled examples. To fully use the accessible data, recent methods explore suitable measures for the similarity between the query and support images and better high-dimensional features with meta-training and pre-training strategies. However, the potential of multi-modality information has barely been explored, which may bring promising improvement for few-shot classification. In this paper, we propose a Language-guided Prototypical Network (LPN) for few-shot classification, which leverages the complementarity of vision and language modalities via two parallel branches to improve the classifier. Concretely, to introduce language modality with limited samples in the visual task, we leverage a pre-trained text encoder to extract class-level text features directly from class names while processing images with a conventional image encoder. Then, we introduce a language-guided decoder to obtain text features corresponding to each image by aligning class-level features with visual features. Additionally, we utilize class-level features and prototypes to build a refined prototypical head, which generates robust prototypes in the text branch for follow-up measurement. Furthermore, we leverage the class-level features to align the visual features, capturing more class-relevant visual features. Finally, we aggregate the visual and text logits to calibrate the deviation of a single modality, enhancing the overall performance. Extensive experiments demonstrate the competitiveness of LPN against state-of-the-art methods on benchmark datasets.

cs.CV↗
arXiv2023

A ModelOps-based Framework for Intelligent Medical Knowledge Extraction

Extracting medical knowledge from healthcare texts enhances downstream tasks like medical knowledge graph construction and clinical decision-making. However, the construction and application of knowledge extraction models lack automation, reusability and unified management, leading to inefficiencies for researchers and high barriers for non-AI experts such as doctors, to utilize knowledge extraction. To address these issues, we propose a ModelOps-based intelligent medical knowledge extraction framework that offers a low-code system for model selection, training, evaluation and optimization. Specifically, the framework includes a dataset abstraction mechanism based on multi-layer callback functions, a reusable model training, monitoring and management mechanism. We also propose a model recommendation method based on dataset similarity, which helps users quickly find potentially suitable models for a given dataset. Our framework provides convenience for researchers to develop models and simplifies model access for non-AI experts such as doctors.

cs.AI↗
arXiv2023

Diversifying the High-level Features for better Adversarial Transferability

Given the great threat of adversarial attacks against Deep Neural Networks (DNNs), numerous works have been proposed to boost transferability to attack real-world applications. However, existing attacks often utilize advanced gradient calculation or input transformation but ignore the white-box model. Inspired by the fact that DNNs are over-parameterized for superior performance, we propose diversifying the high-level features (DHF) for more transferable adversarial examples. In particular, DHF perturbs the high-level features by randomly transforming the high-level features and mixing them with the feature of benign samples when calculating the gradient at each iteration. Due to the redundancy of parameters, such transformation does not affect the classification performance but helps identify the invariant features across different models, leading to much better transferability. Empirical evaluations on ImageNet dataset show that DHF could effectively improve the transferability of existing momentum-based attacks. Incorporated into the input transformation-based attacks, DHF generates more transferable adversarial examples and outperforms the baselines with a clear margin when attacking several defense models, showing its generalization to various attacks and high effectiveness for boosting transferability. Code is available at https://github.com/Trustworthy-AI-Group/DHF.

cs.CV↗
arXiv2023

DiffuseGAE: Controllable and High-fidelity Image Manipulation from Disentangled Representation

Diffusion probabilistic models (DPMs) have shown remarkable results on various image synthesis tasks such as text-to-image generation and image inpainting. However, compared to other generative methods like VAEs and GANs, DPMs lack a low-dimensional, interpretable, and well-decoupled latent code. Recently, diffusion autoencoders (Diff-AE) were proposed to explore the potential of DPMs for representation learning via autoencoding. Diff-AE provides an accessible latent space that exhibits remarkable interpretability, allowing us to manipulate image attributes based on latent codes from the space. However, previous works are not generic as they only operated on a few limited attributes. To further explore the latent space of Diff-AE and achieve a generic editing pipeline, we proposed a module called Group-supervised AutoEncoder(dubbed GAE) for Diff-AE to achieve better disentanglement on the latent code. Our proposed GAE has trained via an attribute-swap strategy to acquire the latent codes for multi-attribute image manipulation based on examples. We empirically demonstrate that our method enables multiple-attributes manipulation and achieves convincing sample quality and attribute alignments, while significantly reducing computational requirements compared to pixel-based approaches for representational decoupling. Code will be released soon.

cs.CV↗
arXiv2023

Locality of gapped ground states in systems with power-law decaying interactions

It has been proved that in gapped ground states of locally-interacting quantum systems, the effect of local perturbations decays exponentially with distance. However, in systems with power-law ($1/r^α$) decaying interactions, no analogous statement has been shown, and there are serious mathematical obstacles to proving it with existing methods. In this paper we prove that when $α$ exceeds the spatial dimension $D$, the effect of local perturbations on local properties a distance $r$ away is upper bounded by a power law $1/r^{α_1}$ in gapped ground states, provided that the perturbations do not close the spectral gap. The power-law exponent $α_1$ is tight if $α>2D$ and interactions are two-body, where we have $α_1=α$. The proof is enabled by a method that avoids the use of quasiadiabatic continuation and incorporates techniques of complex analysis. This method also improves bounds on ground state correlation decay, even in short-range interacting systems. Our work generalizes the fundamental notion that local perturbations have local effects to power-law interacting systems, with broad implications for numerical simulations and experiments.

quant-ph↗
arXiv2023

A Novel Noise Injection-based Training Scheme for Better Model Robustness

Noise injection-based method has been shown to be able to improve the robustness of artificial neural networks in previous work. In this work, we propose a novel noise injection-based training scheme for better model robustness. Specifically, we first develop a likelihood ratio method to estimate the gradient with respect to both synaptic weights and noise levels for stochastic gradient descent training. Then, we design an approximation for the vanilla noise injection-based training method to reduce memory and improve computational efficiency. Next, we apply our proposed scheme to spiking neural networks and evaluate the performance of classification accuracy and robustness on MNIST and Fashion-MNIST datasets. Experiment results show that our proposed method achieves a much better performance on adversarial robustness and slightly better performance on original accuracy, compared with the conventional gradient-based training method.

cs.LG↗
arXiv2023

Personalized State Anxiety Detection: An Empirical Study with Linguistic Biomarkers and A Machine Learning Pipeline

Individuals high in social anxiety symptoms often exhibit elevated state anxiety in social situations. Research has shown it is possible to detect state anxiety by leveraging digital biomarkers and machine learning techniques. However, most existing work trains models on an entire group of participants, failing to capture individual differences in their psychological and behavioral responses to social contexts. To address this concern, in Study 1, we collected linguistic data from N=35 high socially anxious participants in a variety of social contexts, finding that digital linguistic biomarkers significantly differ between evaluative vs. non-evaluative social contexts and between individuals having different trait psychological symptoms, suggesting the likely importance of personalized approaches to detect state anxiety. In Study 2, we used the same data and results from Study 1 to model a multilayer personalized machine learning pipeline to detect state anxiety that considers contextual and individual differences. This personalized model outperformed the baseline F1-score by 28.0%. Results suggest that state anxiety can be more accurately detected with personalized machine learning approaches, and that linguistic biomarkers hold promise for identifying periods of state anxiety in an unobtrusive way.

cs.HC↗