arXiv subjects

Zihan Zhao

Publications and source records attributed to Zihan Zhao.

At least 55 records · Page 3Linked to original sources

arXiv2024

Learning Symmetry-Independent Jet Representations via Jet-Based Joint Embedding Predictive Architecture

In high energy physics, self-supervised learning (SSL) methods have the potential to aid in the creation of machine learning models without the need for labeled datasets for a variety of tasks, including those related to jets -- narrow sprays of particles produced by quarks and gluons in high energy particle collisions. This study introduces an approach to learning jet representations without hand-crafted augmentations using a jet-based joint embedding predictive architecture (J-JEPA), which aims to predict various physical targets from an informative context. As our method does not require hand-crafted augmentation like other common SSL techniques, J-JEPA avoids introducing biases that could harm downstream tasks. Since different tasks generally require invariance under different augmentations, this training without hand-crafted augmentation enables versatile applications, offering a pathway toward a cross-task foundation model. We finetune the representations learned by J-JEPA for jet tagging and benchmark them against task-specific representations.

hep-ph↗

arXiv2024

Strong interaction induced dimensional crossover in 1D quantum gas

We generated a one-dimensional quantum gas confined in an elongated optical dipole trap instead of 2D optical lattices. The sample, comprising thousands of atoms, spans several hundred micrometers and allows for independent control of temperature and chemical potential using Feshbach resonance. This allows us to directly observe and investigate the spatial distribution and associated excitation of 1D quantum gas without any ensemble averaging. In this system, we observed that the dimension of 1D gas will be popped up into 3D due to strong interaction without changing any trapping confinement. During the dimensional crossover, we found that increasing the scattering length leads to the failure of 1D theories, including 1D mean field, Yang-Yang equation, and 1D hydrodynamics. Specifically, the modified Yang-Yang equation effectively describes this 1D system at temperatures beyond the 1D threshold, but it does not account for the effects of stronger interactions. Meanwhile, we observe two possible quantized plateaus of breathing-mode oscillation frequencies predicted by 1D and 3D hydrodynamics, corresponding to weak and strong interactions respectively. And there is also a universal crossover connecting two different regimes where both hydrodynamics fail.

cond-mat.quant-gas↗

arXiv2024

SciDFM: A Large Language Model with Mixture-of-Experts for Science

Recently, there has been a significant upsurge of interest in leveraging large language models (LLMs) to assist scientific discovery. However, most LLMs only focus on general science, while they lack domain-specific knowledge, such as chemical molecules and amino acid sequences. To bridge these gaps, we introduce SciDFM, a mixture-of-experts LLM, which is trained from scratch and is able to conduct college-level scientific reasoning and understand molecules and amino acid sequences. We collect a large-scale training corpus containing numerous scientific papers and books from different disciplines as well as data from domain-specific databases. We further fine-tune the pre-trained model on lots of instruction data to improve performances on downstream benchmarks. From experiment results, we show that SciDFM achieves strong performance on general scientific benchmarks such as SciEval and SciQ, and it reaches a SOTA performance on domain-specific benchmarks among models of similar size. We further analyze the expert layers and show that the results of expert selection vary with data from different disciplines. To benefit the broader research community, we open-source SciDFM at https://huggingface.co/OpenDFM/SciDFM-MoE-A5.6B-v1.0.

cs.CL↗

arXiv2024

SEEKR: Selective Attention-Guided Knowledge Retention for Continual Learning of Large Language Models

Continual learning (CL) is crucial for language models to dynamically adapt to the evolving real-world demands. To mitigate the catastrophic forgetting problem in CL, data replay has been proven a simple and effective strategy, and the subsequent data-replay-based distillation can further enhance the performance. However, existing methods fail to fully exploit the knowledge embedded in models from previous tasks, resulting in the need for a relatively large number of replay samples to achieve good results. In this work, we first explore and emphasize the importance of attention weights in knowledge retention, and then propose a SElective attEntion-guided Knowledge Retention method (SEEKR) for data-efficient replay-based continual learning of large language models (LLMs). Specifically, SEEKR performs attention distillation on the selected attention heads for finer-grained knowledge retention, where the proposed forgettability-based and task-sensitivity-based measures are used to identify the most valuable attention heads. Experimental results on two continual learning benchmarks for LLMs demonstrate the superiority of SEEKR over the existing methods on both performance and efficiency. Explicitly, SEEKR achieves comparable or even better performance with only 1/10 of the replayed data used by other methods, and reduces the proportion of replayed data to 1%.

cs.CL↗

arXiv2024

SciEval: A Multi-Level Large Language Model Evaluation Benchmark for Scientific Research

Recently, there has been growing interest in using Large Language Models (LLMs) for scientific research. Numerous benchmarks have been proposed to evaluate the ability of LLMs for scientific research. However, current benchmarks are mostly based on pre-collected objective questions. This design suffers from data leakage problem and lacks the evaluation of subjective Q/A ability. In this paper, we propose SciEval, a comprehensive and multi-disciplinary evaluation benchmark to address these issues. Based on Bloom's taxonomy, SciEval covers four dimensions to systematically evaluate scientific research ability. In particular, we design a "dynamic" subset based on scientific principles to prevent evaluation from potential data leakage. Both objective and subjective questions are included in SciEval. These characteristics make SciEval a more effective benchmark for scientific research ability evaluation of LLMs. Comprehensive experiments on most advanced LLMs show that, although GPT-4 achieves SOTA performance compared to other LLMs, there is still substantial room for improvement, especially for dynamic questions. The codes and data are publicly available on https://github.com/OpenDFM/SciEval.

cs.CL↗

arXiv2024

Optimizing the image projection of spatially incoherent light from a multimode fiber

We study the spatially incoherent light generated by a multimode fiber(MMF) in the application of image projection designed for the ultracold-atom experiments. Inspired by previous half-analytic methods concerning the incoherent light, here a full-numerical model is established to provide more quantitative descriptions, and part of results is compared with experiments. Particularly, our model about the MMF is also compatible with light propagation in free space. Based on this, we study both the intrinsic speckle and the perturbation robustness of a MMF light field, under the influence of light propagation and fiber parameters. We point out several guidelines about choosing the suitable MMF in creating a spatially incoherent light source, which is useful in the context of the ultracold-atom experiments associating with the optical potential projection.

physics.optics↗

arXiv2024

Large-Scale Pretraining and Finetuning for Efficient Jet Classification in Particle Physics

This study introduces an innovative approach to analyzing unlabeled data in high-energy physics (HEP) through the application of self-supervised learning (SSL). Faced with the increasing computational cost of producing high-quality labeled simulation samples at the CERN LHC, we propose leveraging large volumes of unlabeled data to overcome the limitations of supervised learning methods, which heavily rely on detailed labeled simulations. By pretraining models on these vast, mostly untapped datasets, we aim to learn generic representations that can be finetuned with smaller quantities of labeled data. Our methodology employs contrastive learning with augmentations on jet datasets to teach the model to recognize common representations of jets, addressing the unique challenges of LHC physics. Building on the groundwork laid by previous studies, our work demonstrates the critical ability of SSL to utilize large-scale unlabeled data effectively. We showcase the scalability and effectiveness of our models by gradually increasing the size of the pretraining dataset and assessing the resultant performance enhancements. Our results, obtained from experiments on two datasets -- JetClass, representing unlabeled data, and Top Tagging, serving as labeled simulation data -- show significant improvements in data efficiency, computational efficiency, and overall performance. These findings suggest that SSL can greatly enhance the adaptability of ML models to the HEP domain. This work opens new avenues for the use of unlabeled data in HEP and contributes to a better understanding the potential of SSL for scientific discovery.

hep-ex↗

arXiv2024

HSDreport: Heart Sound Diagnosis with Echocardiography Reports

Heart sound auscultation holds significant importance in the diagnosis of congenital heart disease. However, existing methods for Heart Sound Diagnosis (HSD) tasks are predominantly limited to a few fixed categories, framing the HSD task as a rigid classification problem that does not fully align with medical practice and offers only limited information to physicians. Besides, such methods do not utilize echocardiography reports, the gold standard in the diagnosis of related diseases. To tackle this challenge, we introduce HSDreport, a new benchmark for HSD, which mandates the direct utilization of heart sounds obtained from auscultation to predict echocardiography reports. This benchmark aims to merge the convenience of auscultation with the comprehensive nature of echocardiography reports. First, we collect a new dataset for this benchmark, comprising 2,275 heart sound samples along with their corresponding reports. Subsequently, we develop a knowledge-aware query-based transformer to handle this task. The intent is to leverage the capabilities of medically pre-trained models and the internal knowledge of large language models (LLMs) to address the task's inherent complexity and variability, thereby enhancing the robustness and scientific validity of the method. Furthermore, our experimental results indicate that our method significantly outperforms traditional HSD approaches and existing multimodal LLMs in detecting key abnormalities in heart sounds.

cs.SD↗

arXiv2024

Boosting Medical Image Classification with Segmentation Foundation Model

The Segment Anything Model (SAM) exhibits impressive capabilities in zero-shot segmentation for natural images. Recently, SAM has gained a great deal of attention for its applications in medical image segmentation. However, to our best knowledge, no studies have shown how to harness the power of SAM for medical image classification. To fill this gap and make SAM a true ``foundation model'' for medical image analysis, it is highly desirable to customize SAM specifically for medical image classification. In this paper, we introduce SAMAug-C, an innovative augmentation method based on SAM for augmenting classification datasets by generating variants of the original images. The augmented datasets can be used to train a deep learning classification model, thereby boosting the classification performance. Furthermore, we propose a novel framework that simultaneously processes raw and SAMAug-C augmented image input, capitalizing on the complementary information that is offered by both. Experiments on three public datasets validate the effectiveness of our new approach.

cs.CV↗

arXiv2024

Mobile-Env: Building Qualified Evaluation Benchmarks for LLM-GUI Interaction

The Graphical User Interface (GUI) is pivotal for human interaction with the digital world, enabling efficient device control and the completion of complex tasks. Recent progress in Large Language Models (LLMs) and Vision Language Models (VLMs) offers the chance to create advanced GUI agents. To ensure their effectiveness, there's a pressing need for qualified benchmarks that provide trustworthy and reproducible evaluations -- a challenge current benchmarks often fail to address. To tackle this issue, we introduce Mobile-Env, a comprehensive toolkit tailored for creating GUI benchmarks in the Android mobile environment. Mobile-Env offers an isolated and controllable setting for reliable evaluations, and accommodates intermediate instructions and rewards to reflect real-world usage more naturally. Utilizing Mobile-Env, we collect an open-world task set across various real-world apps and a fixed world set, WikiHow, which captures a significant amount of dynamic online contents for fully controllable and reproducible evaluation. We conduct comprehensive evaluations of LLM agents using these benchmarks. Our findings reveal that even advanced models (e.g., GPT-4V and LLaMA-3) struggle with tasks that are relatively simple for humans. This highlights a crucial gap in current models and underscores the importance of developing more capable foundation models and more effective GUI agent frameworks.

cs.AI↗

arXiv2024

A Hessian-Based Field Deformer for Real-Time Topology-Aware Shape Editing

Shape manipulation is a central research topic in computer graphics. Topology editing, such as breaking apart connections, joining disconnected ends, and filling/opening a topological hole, is generally more challenging than geometry editing. In this paper, we observe that the saddle points of the signed distance function (SDF) provide useful hints for altering surface topology deliberately. Based on this key observation, we parameterize the SDF into a cubic trivariate tensor-product B-spline function $F$ whose saddle points $\{\boldsymbol{s}_i\}$ can be quickly exhausted based on a subdivision-based root-finding technique coupled with Newton's method. Users can select one of the candidate points, say $\boldsymbol{s}_i$, to edit the topology in real time. In implementation, we add a compactly supported B-spline function rooted at $\boldsymbol{s}_i$, which we call a \textit{deformer} in this paper, to $F$, with its local coordinate system aligning with the three eigenvectors of the Hessian. Combined with ray marching technique, our interactive system operates at 30 FPS. Additionally, our system empowers users to create desired bulges or concavities on the surface. An extensive user study indicates that our system is user-friendly and intuitive to operate. We demonstrate the effectiveness and usefulness of our system in a range of applications, including fixing surface reconstruction errors, artistic work design, 3D medical imaging and simulation, and antiquity restoration. Please refer to the attached video for a demonstration.

cs.GR↗

arXiv2024

LibriSQA: A Novel Dataset and Framework for Spoken Question Answering with Large Language Models

While Large Language Models (LLMs) have demonstrated commendable performance across a myriad of domains and tasks, existing LLMs still exhibit a palpable deficit in handling multimodal functionalities, especially for the Spoken Question Answering (SQA) task which necessitates precise alignment and deep interaction between speech and text features. To address the SQA challenge on LLMs, we initially curated the free-form and open-ended LibriSQA dataset from Librispeech, comprising Part I with natural conversational formats and Part II encompassing multiple-choice questions followed by answers and analytical segments. Both parts collectively include 107k SQA pairs that cover various topics. Given the evident paucity of existing speech-text LLMs, we propose a lightweight, end-to-end framework to execute the SQA task on the LibriSQA, witnessing significant results. By reforming ASR into the SQA format, we further substantiate our framework's capability in handling ASR tasks. Our empirical findings bolster the LLMs' aptitude for aligning and comprehending multimodal information, paving the way for the development of universal multimodal LLMs. The dataset and demo can be found at https://github.com/ZihanZhaoSJTU/LibriSQA.

cs.CL↗

arXiv2024

Hierarchical Multimodal Pre-training for Visually Rich Webpage Understanding

The growing prevalence of visually rich documents, such as webpages and scanned/digital-born documents (images, PDFs, etc.), has led to increased interest in automatic document understanding and information extraction across academia and industry. Although various document modalities, including image, text, layout, and structure, facilitate human information retrieval, the interconnected nature of these modalities presents challenges for neural networks. In this paper, we introduce WebLM, a multimodal pre-training network designed to address the limitations of solely modeling text and structure modalities of HTML in webpages. Instead of processing document images as unified natural images, WebLM integrates the hierarchical structure of document images to enhance the understanding of markup-language-based documents. Additionally, we propose several pre-training tasks to model the interaction among text, structure, and image modalities effectively. Empirical results demonstrate that the pre-trained WebLM significantly surpasses previous state-of-the-art pre-trained models across several webpage understanding tasks. The pre-trained models and code are available at https://github.com/X-LANCE/weblm.

cs.CL↗

arXiv2024

Computational Experiments Meet Large Language Model Based Agents: A Survey and Perspective

Computational experiments have emerged as a valuable method for studying complex systems, involving the algorithmization of counterfactuals. However, accurately representing real social systems in Agent-based Modeling (ABM) is challenging due to the diverse and intricate characteristics of humans, including bounded rationality and heterogeneity. To address this limitation, the integration of Large Language Models (LLMs) has been proposed, enabling agents to possess anthropomorphic abilities such as complex reasoning and autonomous learning. These agents, known as LLM-based Agent, offer the potential to enhance the anthropomorphism lacking in ABM. Nonetheless, the absence of explicit explainability in LLMs significantly hinders their application in the social sciences. Conversely, computational experiments excel in providing causal analysis of individual behaviors and complex phenomena. Thus, combining computational experiments with LLM-based Agent holds substantial research potential. This paper aims to present a comprehensive exploration of this fusion. Primarily, it outlines the historical development of agent structures and their evolution into artificial societies, emphasizing their importance in computational experiments. Then it elucidates the advantages that computational experiments and LLM-based Agents offer each other, considering the perspectives of LLM-based Agent for computational experiments and vice versa. Finally, this paper addresses the challenges and future trends in this research domain, offering guidance for subsequent related studies.

cs.AI↗

arXiv2023

Large Language Models Are Semi-Parametric Reinforcement Learning Agents

Inspired by the insights in cognitive science with respect to human memory and reasoning mechanism, a novel evolvable LLM-based (Large Language Model) agent framework is proposed as REMEMBERER. By equipping the LLM with a long-term experience memory, REMEMBERER is capable of exploiting the experiences from the past episodes even for different task goals, which excels an LLM-based agent with fixed exemplars or equipped with a transient working memory. We further introduce Reinforcement Learning with Experience Memory (RLEM) to update the memory. Thus, the whole system can learn from the experiences of both success and failure, and evolve its capability without fine-tuning the parameters of the LLM. In this way, the proposed REMEMBERER constitutes a semi-parametric RL agent. Extensive experiments are conducted on two RL task sets to evaluate the proposed framework. The average results with different initialization and training sets exceed the prior SOTA by 4% and 2% for the success rate on two task sets and demonstrate the superiority and robustness of REMEMBERER.

cs.CL↗

arXiv2023

Observation of universal dissipative dynamics in strongly correlated quantum gas

Dissipation is unavoidable in quantum systems. It usually induces decoherences and changes quantum correlations. To access the information of strongly correlated quantum matters, one has to overcome or suppress dissipation to extract out the underlying quantum phenomena. However, here we find an opposite effect that dissipation can be utilized as a powerful tool to probe the intrinsic correlations of quantum many-body systems. Applying highly-controllable dissipation in ultracold atomic systems, we observe a universal dissipative dynamics in strongly correlated one-dimensional quantum gases. The total particle number of this system follows a universal stretched-exponential decay, and the stretched exponent measures the anomalous dimension of the spectral function, a critical exponent characterizing strong quantum fluctuations of this system. This method could have broad applications in detecting strongly correlated features, including spin-charge separations and Fermi arcs in quantum materials.

cond-mat.quant-gas↗

arXiv2023

Calibrating the absorption imaging of cold atoms under high magnetic fields

We develop a theoretical model for calibrating the absorption imaging of cold atoms under high magnetic fields. Comparing to zero or low magnetic fields, the efficiency of the absorption imaging becomes lower while it requires an additional correction factor to obtain the absolute atom number under the Beer-Lambert law. Our model is based on the rate equations and can account many experimental imperfections such as Zeeman level crossing, failures of hyperfine structures, off-resonant couplings, and low repumping efficiency, etc. Based on this method, we can precisely calculate the correction factor for atom number measurement without any empirical or fitting parameters. Meanwhile, we use a cold-atom apparatus of rubidium-85 to experimentally verify our model. Besides these, we find our work can also serve as a benchmark to measure the polarization impurity of a circular-polarized laser beam with high sensitivities. We believe this work will bring convenience for most of cold-atom experiments using absorption imaging.

physics.atom-ph↗

arXiv2023

Knowledge-aware Bayesian Co-attention for Multimodal Emotion Recognition

Multimodal emotion recognition is a challenging research area that aims to fuse different modalities to predict human emotion. However, most existing models that are based on attention mechanisms have difficulty in learning emotionally relevant parts on their own. To solve this problem, we propose to incorporate external emotion-related knowledge in the co-attention based fusion of pre-trained models. To effectively incorporate this knowledge, we enhance the co-attention model with a Bayesian attention module (BAM) where a prior distribution is estimated using the emotion-related knowledge. Experimental results on the IEMOCAP dataset show that the proposed approach can outperform several state-of-the-art approaches by at least 0.7% unweighted accuracy (UA).

cs.CL↗