arXiv subjects

Mingjie Li

Publications and source records attributed to Mingjie Li.

At least 73 records · Page 4Linked to original sources

arXiv2024

Moving impedance profiles make one--way, spectrum--reshaping mirrors

We find a new set of exact solutions to Maxwell's equations in space--time varying materials, where the refractive index is constant, while the impedance exhibits effective motion, i.e. it is a function of $x-vt$. We find that waves co--propagating with the modulation are not reflected within the material, while counter--propagating waves are continually reflected by the changing impedance. For a finite section of such a material we find analogues of transmission resonances, where specially shaped `eigenpulses' enter without reflection. We also find that there is a strong asymmetry in reflection from the medium when the impedance modulation is small but rapid, the material reflecting strongly from one side, and negligibly from the other. Unlike stationary media, the spectrum of the reflected wave can be significantly different from the incident one.

physics.optics↗

arXiv2024

Stability of Stationary Solutions to the Nonisentropic Euler-Poisson System in a Perturbed Half Space

The main concern of this paper is to mathematically investigate the formation of a plasma sheath near the surface of nonplanar walls. We study the existence and asymptotic stability of stationary solutions for the nonisentropic Euler-Poisson equations in a domain of which boundary is drawn by a graph, by employing a space weighted energy method. Moreover, the convergence rate of the solution toward the stationary solution is obtained, provided that the initial perturbation belongs to the weighted Sobolev space. Because the domain is the perturbed half space, we first show the time-global solvability of the nonisentropic Euler-Poisson equations, then construct stationary solutions by using the time-global solutions.

math.AP↗

arXiv2024

Multiple plateaus of high-sideband generation from Floquet matters

We theoretically report that high-order sideband generation (HSG) from Floquet matters driven by a strong terahertz light while engineered by weak infrared light can achieve multiple plateau HSG. The Floquet-engineering systems exhibit distinctive spectroscopic characteristics that go beyond the HSG processes in field-free band-structure systems. The spatial-temporal dynamics analyses under Floquet-Bloch and time-reversal-symmetry theories clarify the spectral and its odd-even characteristics in the HSG spectrum. Our work demonstrates the HSG of Floquet matters via Floquet engineering and indicates a promising way to extract Floquet material parameters in future experiments.

physics.optics↗

arXiv2024

Black-box Adversarial Attacks Against Image Quality Assessment Models

The goal of No-Reference Image Quality Assessment (NR-IQA) is to predict the perceptual quality of an image in line with its subjective evaluation. To put the NR-IQA models into practice, it is essential to study their potential loopholes for model refinement. This paper makes the first attempt to explore the black-box adversarial attacks on NR-IQA models. Specifically, we first formulate the attack problem as maximizing the deviation between the estimated quality scores of original and perturbed images, while restricting the perturbed image distortions for visual quality preservation. Under such formulation, we then design a Bi-directional loss function to mislead the estimated quality scores of adversarial examples towards an opposite direction with maximum deviation. On this basis, we finally develop an efficient and effective black-box attack method against NR-IQA models. Extensive experiments reveal that all the evaluated NR-IQA models are vulnerable to the proposed attack method. And the generated perturbations are not transferable, enabling them to serve the investigation of specialities of disparate IQA models.

cs.CV↗

arXiv2024

EMBRE: Entity-aware Masking for Biomedical Relation Extraction

Information extraction techniques, including named entity recognition (NER) and relation extraction (RE), are crucial in many domains to support making sense of vast amounts of unstructured text data by identifying and connecting relevant information. Such techniques can assist researchers in extracting valuable insights. In this paper, we introduce the Entity-aware Masking for Biomedical Relation Extraction (EMBRE) method for biomedical relation extraction, as applied in the context of the BioRED challenge Task 1, in which human-annotated entities are provided as input. Specifically, we integrate entity knowledge into a deep neural network by pretraining the backbone model with an entity masking objective. We randomly mask named entities for each instance and let the model identify the masked entity along with its type. In this way, the model is capable of learning more specific knowledge and more robust representations. Then, we utilize the pre-trained model as our backbone to encode language representations and feed these representations into two multilayer perceptron (MLPs) to predict the logits for relation and novelty, respectively. The experimental results demonstrate that our proposed method can improve the performances of entity pair, relation and novelty extraction over our baseline.

cs.CL↗

arXiv2023

Interpreting and Disentangling Feature Components of Various Complexity from DNNs

This paper aims to define, quantify, and analyze the feature complexity that is learned by a DNN. We propose a generic definition for the feature complexity. Given the feature of a certain layer in the DNN, our method disentangles feature components of different complexity orders from the feature. We further design a set of metrics to evaluate the reliability, the effectiveness, and the significance of over-fitting of these feature components. Furthermore, we successfully discover a close relationship between the feature complexity and the performance of DNNs. As a generic mathematical tool, the feature complexity and the proposed metrics can also be used to analyze the success of network compression and knowledge distillation.

cs.LG↗

arXiv2023

Mask Propagation for Efficient Video Semantic Segmentation

Video Semantic Segmentation (VSS) involves assigning a semantic label to each pixel in a video sequence. Prior work in this field has demonstrated promising results by extending image semantic segmentation models to exploit temporal relationships across video frames; however, these approaches often incur significant computational costs. In this paper, we propose an efficient mask propagation framework for VSS, called MPVSS. Our approach first employs a strong query-based image segmentor on sparse key frames to generate accurate binary masks and class predictions. We then design a flow estimation module utilizing the learned queries to generate a set of segment-aware flow maps, each associated with a mask prediction from the key frame. Finally, the mask-flow pairs are warped to serve as the mask predictions for the non-key frames. By reusing predictions from key frames, we circumvent the need to process a large volume of video frames individually with resource-intensive segmentors, alleviating temporal redundancy and significantly reducing computational costs. Extensive experiments on VSPW and Cityscapes demonstrate that our mask propagation framework achieves SOTA accuracy and efficiency trade-offs. For instance, our best model with Swin-L backbone outperforms the SOTA MRCFA using MiT-B5 by 4.0% mIoU, requiring only 26% FLOPs on the VSPW dataset. Moreover, our framework reduces up to 4x FLOPs compared to the per-frame Mask2Former baseline with only up to 2% mIoU degradation on the Cityscapes validation set. Code is available at https://github.com/ziplab/MPVSS.

cs.CV↗

arXiv2023

Explaining How a Neural Network Play the Go Game and Let People Learn

The AI model has surpassed human players in the game of Go, and it is widely believed that the AI model has encoded new knowledge about the Go game beyond human players. In this way, explaining the knowledge encoded by the AI model and using it to teach human players represent a promising-yet-challenging issue in explainable AI. To this end, mathematical supports are required to ensure that human players can learn accurate and verifiable knowledge, rather than specious intuitive analysis. Thus, in this paper, we extract interaction primitives between stones encoded by the value network for the Go game, so as to enable people to learn from the value network. Experiments show the effectiveness of our method.

cs.AI↗

arXiv2023

The Electronic and Electromagnetic Dirac Equations

Maxwell's equations and the Dirac equation are the first-order differential relativistic wave equation for electromagnetic waves and electronic waves respectively. Hence, there is a notable similarity between these two wave equations, which has been widely researched since the Dirac equation was proposed. In this paper, we show that the Maxwell equations can be written in an exact form of the Dirac equation by representing the four Dirac operators with $8\times8$ matrices. Unlike the ordinary $4\times4$ Dirac equation, both spin--1/2 and spin--1 operators can be derived from the $8\times8$ Dirac equation, manifesting that the $8\times8$ Dirac equation is able to describe both electrons and photons. As a result of the restrictions that the electromagnetic wave is a transverse wave, the photon is a spin--1 particle. The four--current in the Maxwell equations and the mass in the electronic Dirac equation also force the electromagnetic field to transform differently to the electronic field. We use this $8\times8$ representation to find that the Zitterbewegung of the photon is actually the oscillatory part of the Poynting vector, often neglected upon time averaging.

quant-ph↗

arXiv2023

Data navigation on the ENCODE portal

Spanning two decades, the Encyclopaedia of DNA Elements (ENCODE) is a collaborative research project that aims to identify all the functional elements in the human and mouse genomes. To best serve the scientific community, all data generated by the consortium is shared through a web-portal (https://www.encodeproject.org/) with no access restrictions. The fourth and final phase of the project added a diverse set of new samples (including those associated with human disease), and a wide range of new assays aimed at detection, characterization and validation of functional genomic elements. The ENCODE data portal hosts results from over 23,000 functional genomics experiments, over 800 functional elements characterization experiments (including in vivo transgenic enhancer assays, reporter assays and CRISPR screens) along with over 60,000 results of computational and integrative analyses (including imputations, predictions and genome annotations). The ENCODE Data Coordination Center (DCC) is responsible for development and maintenance of the data portal, along with the implementation and utilisation of the ENCODE uniform processing pipelines to generate uniformly processed data. Here we report recent updates to the data portal. Specifically, we have completely redesigned the home page, improved search interface, added several new pages to highlight collections of biologically related data (deeply profiled cell lines, immune cells, Alzheimer's Disease, RNA-Protein interactions, degron matrix and a matrix of experiments organised by human donors), added single-cell experiments, and enhanced the cart interface for visualisation and download of user-selected datasets.

q-bio.GN↗

arXiv2023

Towards Medical Artificial General Intelligence via Knowledge-Enhanced Multimodal Pretraining

Medical artificial general intelligence (MAGI) enables one foundation model to solve different medical tasks, which is very practical in the medical domain. It can significantly reduce the requirement of large amounts of task-specific data by sufficiently sharing medical knowledge among different tasks. However, due to the challenges of designing strongly generalizable models with limited and complex medical data, most existing approaches tend to develop task-specific models. To take a step towards MAGI, we propose a new paradigm called Medical-knOwledge-enhanced mulTimOdal pretRaining (MOTOR). In MOTOR, we combine two kinds of basic medical knowledge, i.e., general and specific knowledge, in a complementary manner to boost the general pretraining process. As a result, the foundation model with comprehensive basic knowledge can learn compact representations from pretraining radiographic data for better cross-modal alignment. MOTOR unifies the understanding and generation, which are two kinds of core intelligence of an AI system, into a single medical foundation model, to flexibly handle more diverse medical tasks. To enable a comprehensive evaluation and facilitate further research, we construct a medical multimodal benchmark including a wide range of downstream tasks, such as chest x-ray report generation and medical visual question answering. Extensive experiments on our benchmark show that MOTOR obtains promising results through simple task-oriented adaptation. The visualization shows that the injected knowledge successfully highlights key information in the medical data, demonstrating the excellent interpretability of MOTOR. Our MOTOR successfully mimics the human practice of fulfilling a "medical student" to accelerate the process of becoming a "specialist". We believe that our work makes a significant stride in realizing MAGI.

cs.AI↗

arXiv2023

Defining and Quantifying the Emergence of Sparse Concepts in DNNs

This paper aims to illustrate the concept-emerging phenomenon in a trained DNN. Specifically, we find that the inference score of a DNN can be disentangled into the effects of a few interactive concepts. These concepts can be understood as causal patterns in a sparse, symbolic causal graph, which explains the DNN. The faithfulness of using such a causal graph to explain the DNN is theoretically guaranteed, because we prove that the causal graph can well mimic the DNN's outputs on an exponential number of different masked samples. Besides, such a causal graph can be further simplified and re-written as an And-Or graph (AOG), without losing much explanation accuracy.

cs.LG↗

arXiv2023

Can the Inference Logic of Large Language Models be Disentangled into Symbolic Concepts?

In this paper, we explain the inference logic of large language models (LLMs) as a set of symbolic concepts. Many recent studies have discovered that traditional DNNs usually encode sparse symbolic concepts. However, because an LLM has much more parameters than traditional DNNs, whether the LLM also encodes sparse symbolic concepts is still an open problem. Therefore, in this paper, we propose to disentangle the inference score of LLMs for dialogue tasks into a small number of symbolic concepts. We verify that we can use those sparse concepts to well estimate all inference scores of the LLM on all arbitrarily masking states of the input sentence. We also evaluate the transferability of concepts encoded by an LLM and verify that symbolic concepts usually exhibit high transferability across similar input sentences. More crucially, those symbolic concepts can be used to explain the exact reasons accountable for the LLM's prediction errors.

cs.CL↗

arXiv2023

Dynamic Graph Enhanced Contrastive Learning for Chest X-ray Report Generation

Automatic radiology reporting has great clinical potential to relieve radiologists from heavy workloads and improve diagnosis interpretation. Recently, researchers have enhanced data-driven neural networks with medical knowledge graphs to eliminate the severe visual and textual bias in this task. The structures of such graphs are exploited by using the clinical dependencies formed by the disease topic tags via general knowledge and usually do not update during the training process. Consequently, the fixed graphs can not guarantee the most appropriate scope of knowledge and limit the effectiveness. To address the limitation, we propose a knowledge graph with Dynamic structure and nodes to facilitate medical report generation with Contrastive Learning, named DCL. In detail, the fundamental structure of our graph is pre-constructed from general knowledge. Then we explore specific knowledge extracted from the retrieved reports to add additional nodes or redefine their relations in a bottom-up manner. Each image feature is integrated with its very own updated graph before being fed into the decoder module for report generation. Finally, this paper introduces Image-Report Contrastive and Image-Report Matching losses to better represent visual features and textual information. Evaluated on IU-Xray and MIMIC-CXR datasets, our DCL outperforms previous state-of-the-art models on these two benchmarks.

cs.CV↗

arXiv2023

Category $\mathcal{O}$ for the Lie algebra of vector fields on the line

Let $\mathfrak{W}$ be the Lie algebra of vector fields on the line. Via computing extensions between all simple modules in the category $\mathcal{O}$, we give the block decomposition of $\mathcal{O}$, and show that the representation type of each block of $\mathcal{O}$ is wild using the Ext-quiver. Each block of $\mathcal{O}$ has infinite simple objects. This result is very different from that of $\mathcal{O}$ for complex semisimple Lie algebras. To find a connection between $\mathcal{O}$ and the module category over some associative algebra, we define a subalgebra $H_1$ of $U(\mathfrak{b})$. We give an exact functor from $\mathcal{O}$ to the category $Ω$ of finite dimensional modules over $H_1$. We also construct new simple $\mathfrak{W}$-modules from Weyl modules and modules over the Borel subalgebra $\mathfrak{b}$ of $\mathfrak{W}$.

math.RT↗

arXiv2023

Entropy-bounded solutions to the Cauchy problem of compressible planar non-resistive magnetohydrodynamics equations with far field vacuum

We investigate the Cauchy problem to the compressible planar non-resistive magnetohydrodynamic equations with zero heat conduction. The global existence of strong solutions to such a model has been established by Li and Li (J. Differential Equations 316: 136--157, 2022). However, to our best knowledge, so far there is no result on the behavior of the entropy near the vacuum region for this model. The main novelty of this paper is to give a positive response to this problem. More precisely, by a series of a priori estimates, especially the singular type estimates, we show that the boundedness of the entropy can be propagated up to any finite time provided that the initial vacuum presents only at far fields with sufficiently slow decay of the initial density.

math.AP↗

arXiv2022

Actionable and Interpretable Fault Localization for Recurring Failures in Online Service Systems

Fault localization is challenging in an online service system due to its monitoring data's large volume and variety and complex dependencies across or within its components (e.g., services or databases). Furthermore, engineers require fault localization solutions to be actionable and interpretable, which existing research approaches cannot satisfy. Therefore, the common industry practice is that, for a specific online service system, its experienced engineers focus on localization for recurring failures based on the knowledge accumulated about the system and historical failures. Although the above common practice is actionable and interpretable, it is largely manual, thus slow and sometimes inaccurate. In this paper, we aim to automate this practice through machine learning. That is, we propose an actionable and interpretable fault localization approach, DejaVu, for recurring failures in online service systems. For a specific online service system, DejaVu takes historical failures and dependencies in the system as input and trains a localization model offline; for an incoming failure, the trained model online recommends where the failure occurs (i.e., the faulty components) and which kind of failure occurs (i.e., the indicative group of metrics) (thus actionable), which are further interpreted by both global and local interpretation methods (thus interpretable). Based on the evaluation on 601 failures from three production systems and one open-source benchmark, in less than one second, DejaVu can on average rank the ground truths at 1.66-th to 5.03-th among a long candidate list, outperforming baselines by at least 51.51%.

cs.SE↗

arXiv2022

Towards Theoretical Analysis of Transformation Complexity of ReLU DNNs

This paper aims to theoretically analyze the complexity of feature transformations encoded in piecewise linear DNNs with ReLU layers. We propose metrics to measure three types of complexities of transformations based on the information theory. We further discover and prove the strong correlation between the complexity and the disentanglement of transformations. Based on the proposed metrics, we analyze two typical phenomena of the change of the transformation complexity during the training process, and explore the ceiling of a DNN's complexity. The proposed metrics can also be used as a loss to learn a DNN with the minimum complexity, which also controls the over-fitting level of the DNN and influences adversarial robustness, adversarial transferability, and knowledge consistency. Comprehensive comparative studies have provided new perspectives to understand the DNN.

cs.LG↗