arXiv subjects
Publications and source records attributed to Cheng Chen.
In this paper, the authors study the matrix-valued harmonic functions and characterize them by the Poisson integral of functions in non-commutative BMO (bounded mean oscillation) spaces. This provides a very satisfactory non-commutative analogue of the beautiful result due to Fabes, Johnson and Neri [Indiana Univ. Math. J. {\bf25} (1976) 159-170; MR0394172].
Diffusion models are emerging models that generate images by iteratively denoising random Gaussian noise using deep neural networks. These models typically exhibit high computational and memory demands, necessitating effective post-training quantization for high-performance inference. Recent works propose low-bitwidth (e.g., 8-bit or 4-bit) quantization for diffusion models, however 4-bit integer quantization typically results in low-quality images. We observe that on several widely used hardware platforms, there is little or no difference in compute capability between floating-point and integer arithmetic operations of the same bitwidth (e.g., 8-bit or 4-bit). Therefore, we propose an effective floating-point quantization method for diffusion models that provides better image quality compared to integer quantization methods. We employ a floating-point quantization method that was effective for other processing tasks, specifically computer vision and natural language tasks, and tailor it for diffusion models by integrating weight rounding learning during the mapping of the full-precision values to the quantized values in the quantization process. We comprehensively study integer and floating-point quantization methods in state-of-the-art diffusion models. Our floating-point quantization method not only generates higher-quality images than that of integer quantization methods, but also shows no noticeable degradation compared to full-precision models (32-bit floating-point), when both weights and activations are quantized to 8-bit floating-point values, while has minimal degradation with 4-bit weights and 8-bit activations.
In this paper, we establish a noncommutative maximal inequality for ergodic averages with respect to the set $\{k^t|k=1,2,3,...\}$ acting on noncommutative $L_p$ spaces for $p>\frac{\sqrt{5}+1}{2}$.
We use a Rydberg quantum simulator to demonstrate a new form of spectroscopy, called quench spectroscopy, which probes the low-energy excitations of a many-body system. We illustrate the method on a two-dimensional simulation of the spin-1/2 dipolar XY model. Through microscopic measurements of the spatial spin correlation dynamics following a quench, we extract the dispersion relation of the elementary excitations for both ferro- and anti-ferromagnetic couplings. We observe qualitatively different behaviors between the two cases that result from the long-range nature of the interactions, and the frustration inherent in the antiferromagnet. In particular, the ferromagnet exhibits elementary excitations behaving as linear spin waves. In the anti-ferromagnet, spin waves appear to decay, suggesting the presence of strong nonlinearities. Our demonstration highlights the importance of power-law interactions on the excitation spectrum of a many-body system.
One-shot detection of anatomical landmarks is gaining significant attention for its efficiency in using minimal labeled data to produce promising results. However, the success of current methods heavily relies on the employment of extensive unlabeled data to pre-train an effective feature extractor, which limits their applicability in scenarios where a substantial amount of unlabeled data is unavailable. In this paper, we propose the first foundation model-enabled one-shot landmark detection (FM-OSD) framework for accurate landmark detection in medical images by utilizing solely a single template image without any additional unlabeled data. Specifically, we use the frozen image encoder of visual foundation models as the feature extractor, and introduce dual-branch global and local feature decoders to increase the resolution of extracted features in a coarse to fine manner. The introduced feature decoders are efficiently trained with a distance-aware similarity learning loss to incorporate domain knowledge from the single template image. Moreover, a novel bidirectional matching strategy is developed to improve both robustness and accuracy of landmark detection in the case of scattered similarity map obtained by foundation models. We validate our method on two public anatomical landmark detection datasets. By using solely a single template image, our method demonstrates significant superiority over strong state-of-the-art one-shot landmark detection methods.
Semi-supervised learning (SSL) has achieved notable progress in medical image segmentation. To achieve effective SSL, a model needs to be able to efficiently learn from limited labeled data and effectively exploiting knowledge from abundant unlabeled data. Recent developments in visual foundation models, such as the Segment Anything Model (SAM), have demonstrated remarkable adaptability with improved sample efficiency. To harness the power of foundation models for application in SSL, we propose a cross prompting consistency method with segment anything model (CPC-SAM) for semi-supervised medical image segmentation. Our method employs SAM's unique prompt design and innovates a cross-prompting strategy within a dual-branch framework to automatically generate prompts and supervisions across two decoder branches, enabling effectively learning from both scarce labeled and valuable unlabeled data. We further design a novel prompt consistency regularization, to reduce the prompt position sensitivity and to enhance the output invariance under different prompts. We validate our method on two medical image segmentation tasks. The extensive experiments with different labeled-data ratios and modalities demonstrate the superiority of our proposed method over the state-of-the-art SSL methods, with more than 9% Dice improvement on the breast cancer segmentation task.
Although Federated Learning (FL) enables global model training across clients without compromising their raw data, due to the unevenly distributed data among clients, existing Federated Averaging (FedAvg)-based methods suffer from the problem of low inference performance. Specifically, different data distributions among clients lead to various optimization directions of local models. Aggregating local models usually results in a low-generalized global model, which performs worse on most of the clients. To address the above issue, inspired by the observation from a geometric perspective that a well-generalized solution is located in a flat area rather than a sharp area, we propose a novel and heuristic FL paradigm named FedMR (Federated Model Recombination). The goal of FedMR is to guide the recombined models to be trained towards a flat area. Unlike conventional FedAvg-based methods, in FedMR, the cloud server recombines collected local models by shuffling each layer of them to generate multiple recombined models for local training on clients rather than an aggregated global model. Since the area of the flat area is larger than the sharp area, when local models are located in different areas, recombined models have a higher probability of locating in a flat area. When all recombined models are located in the same flat area, they are optimized towards the same direction. We theoretically analyze the convergence of model recombination. Experimental results show that, compared with state-of-the-art FL methods, FedMR can significantly improve the inference accuracy without exposing the privacy of each client.
The Rossiter-McLaughlin effect measures the misalignment between a planet's orbital plane and its host star's rotation plane. Around 10$\%$ of planets exhibit misalignments in the approximate range $80 - 125^\circ$, with their origin remaining a mystery. On the other hand, large misalignments may be common in eccentric circumbinary systems due to misaligned discs undergoing polar alignment. If the binary subsequently merges, a polar circumbinary disc -- along with any planets that form within it -- may remain inclined near 90$^{\circ}$ to the merged star's rotation. To test this hypothesis, we present $N$-body simulations of the evolution of a polar circumbinary debris disc comprised of test particles around an eccentric binary during a binary merger that is induced by tidal dissipation. After the merger, the disc particles remain on near-polar orbits. Interaction of the binary with the polar-aligned gas disc may be required to bring the binary to the small separations that trigger the merger by tides. Our findings imply that planets forming in discs that are polar-aligned to the orbit of a high-eccentricity binary may, following the merger of the binary, provide a possible origin for the population of near-polar planets around single stars.
Light-fidelity (LiFi) is an emerging technology for high-speed short-range mobile communications. Inter-cell interference (ICI) is an important issue that limits the system performance in an optical attocell network. Angle diversity receivers (ADRs) have been proposed to mitigate ICI. In this paper, the structure of pyramid receivers (PRs) and truncated pyramid receivers (TPRs) are studied. The coverage problems of PRs and TPRs are defined and investigated, and the lower bound of field of view (FOV) for each PD is given analytically. The impact of random device orientation and diffuse link signal propagation are taken into consideration. The performances of PRs and TPRs are compared and then optimized ADR structures are proposed. The performance comparison between the select best combining (SBC) and maximum ratio combining (MRC) is given under different noise levels. It is shown that SBC will outperform MRC in an interference limited system, otherwise, MRC is a preferred scheme. In addition, the double source system, where each LiFi AP consists of two sources transmitting the same information signals but with opposite polarity, is proved to outperform the single source (SS) system under certain conditions.
Topologically trivial insulators can be classified into atomic insulators (AIs) and obstructed atomic insulators (OAIs) depending on whether the Wannier charge centers are localized or not at spatial positions occupied by atoms. An OAI can possess unusual properties such as surface states along certain crystalline surfaces, which advantageously appear in materials with much larger bulk energy gap than topological insulators, making them more attractive for potential applications. In this work, we show that a well-known crystal, silicon (Si) is a model OAI, which naturally explains some of Si's unusual properties such as its famous (111) surface states. On this surface, using angle resolved photoemission spectroscopy (ARPES), we reveal sharp quasi-1D massive Dirac line dispersions; we also observe, using scanning tunneling microscopy/spectroscopy (STM/STS), topological solitons at the interface of the two atomic chains. Remarkably, we show that the different chain domains can be reversibly switched at the nanometer scale, suggesting the application potential in ultra-high density storage devices.
Addressing missing modalities presents a critical challenge in multimodal learning. Current approaches focus on developing models that can handle modality-incomplete inputs during inference, assuming that the full set of modalities are available for all the data during training. This reliance on full-modality data for training limits the use of abundant modality-incomplete samples that are often encountered in practical settings. In this paper, we propose a robust universal model with modality reconstruction and model personalization, which can effectively tackle the missing modality at both training and testing stages. Our method leverages a multimodal masked autoencoder to reconstruct the missing modality and masked patches simultaneously, incorporating an innovative distribution approximation mechanism to fully utilize both modality-complete and modality-incomplete data. The reconstructed modalities then contributes to our designed data-model co-distillation scheme to guide the model learning in the presence of missing modalities. Moreover, we propose a CLIP-driven hyper-network to personalize partial model parameters, enabling the model to adapt to each distinct missing modality scenario. Our method has been extensively validated on two brain tumor segmentation benchmarks. Experimental results demonstrate the promising performance of our method, which consistently exceeds previous state-of-the-art approaches under the all-stage missing modality settings with different missing ratios. Code will be available.
In two-dimensional van der Waals magnetic materials, the interplay between magnetism and electron correlation can give rise to new ground states and lead to novel transport and optical properties. A fundamental question in these materials is how the electron correlation manifests and interacts with the magnetic orders. In this study, we demonstrate that the recently discovered 2D antiferromagnetic material, CrSBr, is a Mott insulator, through the combined use of resonant and temperature-dependent angle-resolved photoemission spectroscopy techiniques, supplemented by dynamical mean-field theory analysis. Intriguingly, we found that as the system transitions from the antiferromagnetic to the paramagnetic phases, its Mott bands undergo a reconfiguration, and a coherent-incoherent crossover, driven by the dissolution of the magnetic order. Our findings reveal a distinctive evolution of band structure associated with magnetic phase transitions, shedding light on the investigation of intricate interplay between correlation and magnetic orders in strongly correlated van der Waals magnetic materials.
Landslide susceptibility mapping (LSM) is crucial for identifying high-risk areas and informing prevention strategies. This study investigates the interpretability of statistical, machine learning (ML), and deep learning (DL) models in predicting landslide susceptibility. This is achieved by incorporating various relevant interpretation methods and two types of input factors: a comprehensive set of 19 contributing factors that are statistically relevant to landslides, as well as a dedicated set of 9 triggering factors directly associated with triggering landslides. Given that model performance is a crucial metric in LSM, our investigations into interpretability naturally involve assessing and comparing LSM accuracy across different models considered. In our investigation, the convolutional neural network model achieved the highest accuracy (0.8447 with 19 factors; 0.8048 with 9 factors), while Extreme Gradient Boosting and Support Vector Machine also demonstrated strong predictive capabilities, outperforming conventional statistical models. These findings indicate that DL and sophisticated ML algorithms can effectively capture the complex relationships between input factors and landslide occurrence. However, the interpretability of predictions varied among different models, particularly when using the broader set of 19 contributing factors. Explanation methods like SHAP, LIME, and DeepLIFT also led to variations in interpretation results. Using a comprehensive set of 19 contributing factors improved prediction accuracy but introduced complexities and inconsistency in model interpretations. Focusing on a dedicated set of 9 triggering factors sacrificed some predictive power but enhanced interpretability, as evidenced by more consistent key factors identified across various models and alignment with the findings of field investigation reports....
Over the years, great efforts have been devoted in introducing a sizable and tunable band gap in graphene for its potential application in next-generation electronic devices. The primary challenge in modulating this gap has been the absence of a direct method for observing changes of the band gap in momentum space. In this study, we employ advanced spatial- and angle-resolved photoemission spectroscopy technique to directly visualize the gap formation in bilayer graphene, modulated by both displacement fields and moiré potentials. The application of displacement field via in-situ electrostatic gating introduces a sizable and tunable electronic bandgap, proportional to the field strength up to 100 meV. Meanwhile, the moiré potential, induced by aligning the underlying hexagonal boron nitride substrate, extends the bandgap by ~ 20 meV. Theoretical calculations, effectively capture the experimental observations. Our investigation provides a quantitative understanding of how these two mechanisms collaboratively modulate the band gap in bilayer graphene, offering valuable guidance for the design of graphene-based electronic devices.
Deep learning methods, especially Convolutional Neural Networks (CNN) and Vision Transformer (ViT), are frequently employed to perform semantic segmentation of high-resolution remotely sensed images. However, CNNs are constrained by their restricted receptive fields, while ViTs face challenges due to their quadratic complexity. Recently, the Mamba model, featuring linear complexity and a global receptive field, has gained extensive attention for vision tasks. In such tasks, images need to be serialized to form sequences compatible with the Mamba model. Numerous research efforts have explored scanning strategies to serialize images, aiming to enhance the Mamba model's understanding of images. However, the effectiveness of these scanning strategies remains uncertain. In this research, we conduct a comprehensive experimental investigation on the impact of mainstream scanning directions and their combinations on semantic segmentation of remotely sensed images. Through extensive experiments on the LoveDA, ISPRS Potsdam, and ISPRS Vaihingen datasets, we demonstrate that no single scanning strategy outperforms others, regardless of their complexity or the number of scanning directions involved. A simple, single scanning direction is deemed sufficient for semantic segmentation of high-resolution remotely sensed images. Relevant directions for future research are also recommended.
Dynamic graph random walk (DGRW) emerges as a practical tool for capturing structural relations within a graph. Effectively executing DGRW on GPU presents certain challenges. First, existing sampling methods demand a pre-processing buffer, causing substantial space complexity. Moreover, the power-law distribution of graph vertex degrees introduces workload imbalance issues, rendering DGRW embarrassed to parallelize. In this paper, we propose FlowWalker, a GPU-based dynamic graph random walk framework. FlowWalker implements an efficient parallel sampling method to fully exploit the GPU parallelism and reduce space complexity. Moreover, it employs a sampler-centric paradigm alongside a dynamic scheduling strategy to handle the huge amounts of walking queries. FlowWalker stands as a memory-efficient framework that requires no auxiliary data structures in GPU global memory. We examine the performance of FlowWalker extensively on ten datasets, and experiment results show that FlowWalker achieves up to 752.2x, 72.1x, and 16.4x speedup compared with existing CPU, GPU, and FPGA random walk frameworks, respectively. Case study shows that FlowWalker diminishes random walk time from 35% to 3% in a pipeline of ByteDance friend recommendation GNN training.
Large Language Models (LLMs) have demonstrated impressive capabilities to solve a wide range of tasks without being explicitly fine-tuned on task-specific datasets. However, deploying LLMs in the real world is not trivial, as it requires substantial computing resources. In this paper, we investigate whether smaller, compact LLMs are a good alternative to the comparatively Larger LLMs2 to address significant costs associated with utilizing LLMs in the real world. In this regard, we study the meeting summarization task in a real-world industrial environment and conduct extensive experiments by comparing the performance of fine-tuned compact LLMs (e.g., FLAN-T5, TinyLLaMA, LiteLLaMA) with zero-shot larger LLMs (e.g., LLaMA-2, GPT-3.5, PaLM-2). We observe that most smaller LLMs, even after fine-tuning, fail to outperform larger zero-shot LLMs in meeting summarization datasets. However, a notable exception is FLAN-T5 (780M parameters), which performs on par or even better than many zero-shot Larger LLMs (from 7B to above 70B parameters), while being significantly smaller. This makes compact LLMs like FLAN-T5 a suitable cost-efficient solution for real-world industrial deployment.
High-resolution remotely sensed images pose a challenge for commonly used semantic segmentation methods such as Convolutional Neural Network (CNN) and Vision Transformer (ViT). CNN-based methods struggle with handling such high-resolution images due to their limited receptive field, while ViT faces challenges in handling long sequences. Inspired by Mamba, which adopts a State Space Model (SSM) to efficiently capture global semantic information, we propose a semantic segmentation framework for high-resolution remotely sensed images, named Samba. Samba utilizes an encoder-decoder architecture, with Samba blocks serving as the encoder for efficient multi-level semantic information extraction, and UperNet functioning as the decoder. We evaluate Samba on the LoveDA, ISPRS Vaihingen, and ISPRS Potsdam datasets, comparing its performance against top-performing CNN and ViT methods. The results reveal that Samba achieved unparalleled performance on commonly used remote sensing datasets for semantic segmentation. Our proposed Samba demonstrates for the first time the effectiveness of SSM in semantic segmentation of remotely sensed images, setting a new benchmark in performance for Mamba-based techniques in this specific application. The source code and baseline implementations are available at https://github.com/zhuqinfeng1999/Samba.