arXiv subjects
Publications and source records attributed to Hong Zhou.
For bicyclic hypergraphs, ordering by spectral moment ($S$-order) is investigated in this paper. We give the first and last hypergraphs in an $S$-order of linear bicyclic uniform hypergraphs with given girth and number of edges.
The Zagreb index of a hypergraph is defined as the sum of the squares of the degrees of its vertices. A connected $k$-uniform hypergraph with $n$ vertices and $m$ edges is called bicyclic if $n=m(k-1)-1$. In this paper, we determine the hypergraphs with the maximum and minimum Zagreb indices among all linear bicyclic uniform hypergraphs.
We investigate the temperature evolution in the three-dimensional skin tissue exposed to a millimeter-wave electromagnetic beam that is not necessarily perpendicular to the skin surface. This study examines the effect of the beam's incident angle. The incident angle influences the thermal heating in two aspects: (i) the beam spot projected onto the skin is elongated compared to the intrinsic beam spot in a perpendicular cross section, resulting in a lower power per skin area; and (ii) within the tissue, the beam propagates at the refracted angle relative to the depth direction. At millimeter-wavelength frequencies, the characteristic penetration depth is sub-millimeter, whereas the lateral extent of the beam spans at least several centimeters in applications. We explore the small ratio of the penetration depth to the lateral length scale in a non-dimensional formulation and derive a leading-term asymptotic solution for the temperature distribution. This analysis does not rely on a small incident angle and is therefore applicable to arbitrary angles of incidence. Based on the asymptotic solution, we establish scaling laws for the three-dimensional skin temperature, the skin surface temperature, and the skin volume in which thermal nociceptors are activated.
Visual autoregressive models typically adhere to a raster-order ``next-token prediction" paradigm, which overlooks the spatial and temporal locality inherent in visual content. Specifically, visual tokens exhibit significantly stronger correlations with their spatially or temporally adjacent tokens compared to those that are distant. In this paper, we propose Neighboring Autoregressive Modeling (NAR), a novel paradigm that formulates autoregressive visual generation as a progressive outpainting procedure, following a near-to-far ``next-neighbor prediction" mechanism. Starting from an initial token, the remaining tokens are decoded in ascending order of their Manhattan distance from the initial token in the spatial-temporal space, progressively expanding the boundary of the decoded region. To enable parallel prediction of multiple adjacent tokens in the spatial-temporal space, we introduce a set of dimension-oriented decoding heads, each predicting the next token along a mutually orthogonal dimension. During inference, all tokens adjacent to the decoded tokens are processed in parallel, substantially reducing the model forward steps for generation. Experiments on ImageNet$256\times 256$ and UCF101 demonstrate that NAR achieves 2.4$\times$ and 8.6$\times$ higher throughput respectively, while obtaining superior FID/FVD scores for both image and video generation tasks compared to the PAR-4X approach. When evaluating on text-to-image generation benchmark GenEval, NAR with 0.8B parameters outperforms Chameleon-7B while using merely 0.4 of the training data. Code is available at https://github.com/ThisisBillhe/NAR.
The efficiency of large vision-language models (LVLMs) is constrained by the computational bottleneck of the attention mechanism during the prefill phase and the memory bottleneck of fetching the key-value (KV) cache in the decoding phase, particularly in scenarios involving high-resolution images or videos. Visual content often exhibits substantial redundancy, resulting in highly sparse attention maps within LVLMs. This sparsity can be leveraged to accelerate attention computation or compress the KV cache through various approaches. However, most studies focus on addressing only one of these bottlenecks and do not adequately support dynamic adjustment of sparsity concerning distinct layers or tasks. In this paper, we present ZipVL, an efficient inference framework designed for LVLMs through a dynamic ratio allocation strategy of important tokens. This ratio is adaptively determined based on the layer-specific distribution of attention scores, rather than fixed hyper-parameters, thereby improving efficiency for less complex tasks while maintaining high performance for more challenging ones. Then we select important tokens based on their normalized attention scores and perform sparse attention mechanism solely on those important tokens, reducing the latency in the prefill phase. Tokens deemed less important will be discarded to reduce KV cache size, alleviating the memory bottleneck in the decoding phase. Our experiments demonstrate that ZipVL can accelerate the prefill phase by 2.3$\times$ and improve decoding throughput by 2.8$\times$, with a minimal accuracy reduction of only 0.5\% on VQAv2 benchmark over LLaVA-Next-13B model, effectively enhancing the generation efficiency of LVLMs.
We present a unified deterministic approach for experimental design problems using the method of interlacing polynomials. Our framework recovers the best-known approximation guarantees for the well-studied D/A/E-design problems with simple analysis. Furthermore, we obtain improved non-trivial approximation guarantee for E-design in the challenging small budget regime. Additionally, our approach provides an optimal approximation guarantee for a generalized ratio objective that generalizes both D-design and A-design.
With the increase of graph size, it becomes difficult or even impossible to visualize graph structures clearly within the limited screen space. Consequently, it is crucial to design effective visual representations for large graphs. In this paper, we propose AdaMotif, a novel approach that can capture the essential structure patterns of large graphs and effectively reveal the overall structures via adaptive motif designs. Specifically, our approach involves partitioning a given large graph into multiple subgraphs, then clustering similar subgraphs and extracting similar structural information within each cluster. Subsequently, adaptive motifs representing each cluster are generated and utilized to replace the corresponding subgraphs, leading to a simplified visualization. Our approach aims to preserve as much information as possible from the subgraphs while simplifying the graph efficiently. Notably, our approach successfully visualizes crucial community information within a large graph. We conduct case studies and a user study using real-world graphs to validate the effectiveness of our proposed approach. The results demonstrate the capability of our approach in simplifying graphs while retaining important structural and community information.
We assess the skin thermal injury risk in the situation where a test subject is exposed to an electromagnetic beam until the occurrence of flight action. The physical process is modeled as follows. The absorbed electromagnetic power increases the skin temperature. Wherever it is above a temperature threshold, thermal nociceptors are activated and transduce an electrical signal. When the activated skin volume reaches a threshold, the flight signal is initiated. After the delay of human reaction time, the flight action is materialized when the subject moves away or the beam power is turned off. The injury risk is quantified by the thermal damage parameter calculated in the Arrhenius equation. It depends on the beam power density absorbed into the skin, which is not measurable. In addition, the volume threshold for flight initiation is unknown. To circumference these difficulties, we normalize the formulation and write the thermal damage parameter in terms of the occurrence time of flight action, which is reliably observed in exposure tests. This thermal injury formulation provides a viable framework for investigating the effects of model parameters.
Spectral sparsification and discrepancy minimization are two well-studied areas that are closely related. Building on recent connections between these two areas, we generalize the "deterministic discrepancy walk" framework by Pesenti and Vladu [SODA~23] for vector discrepancy to matrix discrepancy, and use it to give a simpler proof of the matrix partial coloring theorem of Reis and Rothvoss [SODA~20]. Moreover, we show that this matrix discrepancy framework provides a unified approach for various spectral sparsification problems, from stronger notions including unit-circle approximation and singular-value approximation to weaker notions including graphical spectral sketching and effective resistance sparsification. In all of these applications, our framework produces improved results with a simpler and deterministic analysis.
KV cache stores key and value states from previous tokens to avoid re-computation, yet it demands substantial storage space, especially for long sequences. Adaptive KV cache compression seeks to discern the saliency of tokens, preserving vital information while aggressively compressing those of less importance. However, previous methods of this approach exhibit significant performance degradation at high compression ratios due to inaccuracies in identifying salient tokens. In this paper, we present ZipCache, an accurate and efficient KV cache quantization method for LLMs. First, we construct a strong baseline for quantizing KV cache. Through the proposed channel-separable tokenwise quantization scheme, the memory overhead of quantization parameters are substantially reduced compared to fine-grained groupwise quantization. To enhance the compression ratio, we propose normalized attention score as an effective metric for identifying salient tokens by considering the lower triangle characteristics of the attention matrix. Moreover, we develop an efficient approximation method that decouples the saliency metric from full attention scores, enabling compatibility with fast attention implementations like FlashAttention. Extensive experiments demonstrate that ZipCache achieves superior compression ratios, fast generation speed and minimal performance losses compared with previous KV cache compression methods. For instance, when evaluating Mistral-7B model on GSM8k dataset, ZipCache is capable of compressing the KV cache by $4.98\times$, with only a $0.38\%$ drop in accuracy. In terms of efficiency, ZipCache also showcases a $37.3\%$ reduction in prefill-phase latency, a $56.9\%$ reduction in decoding-phase latency, and a $19.8\%$ reduction in GPU memory usage when evaluating LLaMA3-8B model with a input length of $4096$.
Predicting future human pose is a fundamental application for machine intelligence, which drives robots to plan their behavior and paths ahead of time to seamlessly accomplish human-robot collaboration in real-world 3D scenarios. Despite encouraging results, existing approaches rarely consider the effects of the external scene on the motion sequence, leading to pronounced artifacts and physical implausibilities in the predictions. To address this limitation, this work introduces a novel multi-modal sense-informed motion prediction approach, which conditions high-fidelity generation on two modal information: external 3D scene, and internal human gaze, and is able to recognize their salience for future human activity. Furthermore, the gaze information is regarded as the human intention, and combined with both motion and scene features, we construct a ternary intention-aware attention to supervise the generation to match where the human wants to reach. Meanwhile, we introduce semantic coherence-aware attention to explicitly distinguish the salient point clouds and the underlying ones, to ensure a reasonable interaction of the generated sequence with the 3D scene. On two real-world benchmarks, the proposed method achieves state-of-the-art performance both in 3D human pose and trajectory prediction.
Cloud computing (cloud computing) is a kind of distributed computing, referring to the network "cloud" will be a huge data calculation and processing program into countless small programs, and then, through the system composed of multiple servers to process and analyze these small programs to get the results and return to the user. This report explores the intersection of cloud computing and financial information processing, identifying risks and challenges faced by financial institutions in adopting cloud technology. It discusses the need for intelligent solutions to enhance data processing efficiency and accuracy while addressing security and privacy concerns. Drawing on regulatory frameworks, the report proposes policy recommendations to mitigate concentration risks associated with cloud computing in the financial industry. By combining intelligent forecasting and evaluation technologies with cloud computing models, the study aims to provide effective solutions for financial data processing and management, facilitating the industry's transition towards digital transformation.
Diffusion models have demonstrated remarkable capabilities in image synthesis and related generative tasks. Nevertheless, their practicality for real-world applications is constrained by substantial computational costs and latency issues. Quantization is a dominant way to compress and accelerate diffusion models, where post-training quantization (PTQ) and quantization-aware training (QAT) are two main approaches, each bearing its own properties. While PTQ exhibits efficiency in terms of both time and data usage, it may lead to diminished performance in low bit-width. On the other hand, QAT can alleviate performance degradation but comes with substantial demands on computational and data resources. In this paper, we introduce a data-free and parameter-efficient fine-tuning framework for low-bit diffusion models, dubbed EfficientDM, to achieve QAT-level performance with PTQ-like efficiency. Specifically, we propose a quantization-aware variant of the low-rank adapter (QALoRA) that can be merged with model weights and jointly quantized to low bit-width. The fine-tuning process distills the denoising capabilities of the full-precision model into its quantized counterpart, eliminating the requirement for training data. We also introduce scale-aware optimization and temporal learned step-size quantization to further enhance performance. Extensive experimental results demonstrate that our method significantly outperforms previous PTQ-based diffusion models while maintaining similar time and data efficiency. Specifically, there is only a 0.05 sFID increase when quantizing both weights and activations of LDM-4 to 4-bit on ImageNet 256x256. Compared to QAT-based methods, our EfficientDM also boasts a 16.2x faster quantization speed with comparable generation quality. Code is available at \href{https://github.com/ThisisBillhe/EfficientDM}{this hrl}.
With the application of hematopoietic stem cell transplantation and new drugs, the progression-free survival rate and overall survival rate of multiple myeloma have been greatly improved, but it is still considered as a kind of disease that cannot be completely cured. Many patients have disease recurrence after complete remission, which is rooted in the presence of minimal residual disease MRD in patients. Studies have shown that positive MRD is an independent adverse prognostic factor affecting survival, so MRD detection is an important indicator to judge the prognosis of patients and guide clinical treatment. At present, multipa-rameter flow cytometry (MFC), polymerase chain reaction (PCR), positron emission tomography (positron emission) Several techniques, such as PET/computer tomography (CT), have been used for MRD detection of multiple myeloma.However, there is still no cure for the disease. "IFM2013-04" four clinical studies confirmed for the first time that proteasome inhibitors (PIs) and immunomodulatory drugs, The synergism and importance of the combination of IMiDs in the treatment of MM, the large Phase 3 clinical study SWOG SO777 compared the combination of bortezomib plus lenalidomide and dexamethasone. The efficacy of VRD and D established the status of VRD first-line treatment of MM, and due to the good efficacy of CD38 monoclonal antibody in large clinical studies, combination therapy with VRD has been recommended as the first-line treatment of MM. However, to explore the clinical value and problems of applying artificial intelligence bone marrow cell recognition system Morphogo in the detection of multiple myeloma minimal residual disease (MRD)
Model reparameterization is a widely accepted technique for improving inference speed without compromising performance. However, current Post-training Quantization (PTQ) methods often lead to significant accuracy degradation when applied to reparameterized models. This is primarily caused by channel-specific and sample-specific outliers, which appear only at specific samples and channels and impact on the selection of quantization parameters. To address this issue, we propose RepAPQ, a novel framework that preserves the accuracy of quantized reparameterization models. Different from previous frameworks using Mean Squared Error (MSE) as a measurement, we utilize Mean Absolute Error (MAE) to mitigate the influence of outliers on quantization parameters. Our framework comprises two main components: Quantization Protecting Reparameterization and Across-block Calibration. For effective calibration, Quantization Protecting Reparameterization combines multiple branches into a single convolution with an affine layer. During training, the affine layer accelerates convergence and amplifies the output of the convolution to better accommodate samples with outliers. Additionally, Across-block Calibration leverages the measurement of stage output as supervision to address the gradient problem introduced by MAE and enhance the interlayer correlation with quantization parameters. Comprehensive experiments demonstrate the effectiveness of RepAPQ across various models and tasks. Our framework outperforms previous methods by approximately 1\% for 8-bit PTQ and 2\% for 6-bit PTQ, showcasing its superior performance. The code is available at \url{https://github.com/ilur98/DLMC-QUANT}.
Nowadays, Vision Transformer (ViT) is widely utilized in various computer vision tasks, owing to its unique self-attention mechanism. However, the model architecture of ViT is complex and often challenging to comprehend, leading to a steep learning curve. ViT developers and users frequently encounter difficulties in interpreting its inner workings. Therefore, a visualization system is needed to assist ViT users in understanding its functionality. This paper introduces EL-VIT, an interactive visual analytics system designed to probe the Vision Transformer and facilitate a better understanding of its operations. The system consists of four layers of visualization views. The first three layers include model overview, knowledge background graph, and model detail view. These three layers elucidate the operation process of ViT from three perspectives: the overall model architecture, detailed explanation, and mathematical operations, enabling users to understand the underlying principles and the transition process between layers. The fourth interpretation view helps ViT users and experts gain a deeper understanding by calculating the cosine similarity between patches. Our two usage scenarios demonstrate the effectiveness and usability of EL-VIT in helping ViT users understand the working mechanism of ViT.
Collecting and annotating images with pixel-wise labels is time-consuming and laborious. In contrast, synthetic data can be freely available using a generative model (e.g., DALL-E, Stable Diffusion). In this paper, we show that it is possible to automatically obtain accurate semantic masks of synthetic images generated by the Off-the-shelf Stable Diffusion model, which uses only text-image pairs during training. Our approach, called DiffuMask, exploits the potential of the cross-attention map between text and image, which is natural and seamless to extend the text-driven image synthesis to semantic mask generation. DiffuMask uses text-guided cross-attention information to localize class/word-specific regions, which are combined with practical techniques to create a novel high-resolution and class-discriminative pixel-wise mask. The methods help to reduce data collection and annotation costs obviously. Experiments demonstrate that the existing segmentation methods trained on synthetic data of DiffuMask can achieve a competitive performance over the counterpart of real data (VOC 2012, Cityscapes). For some classes (e.g., bird), DiffuMask presents promising performance, close to the stateof-the-art result of real data (within 3% mIoU gap). Moreover, in the open-vocabulary segmentation (zero-shot) setting, DiffuMask achieves a new SOTA result on Unseen class of VOC 2012. The project website can be found at https://weijiawu.github.io/DiffusionMask/.
Image segmentation based on continual learning exhibits a critical drop of performance, mainly due to catastrophic forgetting and background shift, as they are required to incorporate new classes continually. In this paper, we propose a simple, yet effective Continual Image Segmentation method with incremental Dynamic Query (CISDQ), which decouples the representation learning of both old and new knowledge with lightweight query embedding. CISDQ mainly includes three contributions: 1) We define dynamic queries with adaptive background class to exploit past knowledge and learn future classes naturally. 2) CISDQ proposes a class/instance-aware Query Guided Knowledge Distillation strategy to overcome catastrophic forgetting by capturing the inter-class diversity and intra-class identity. 3) Apart from semantic segmentation, CISDQ introduce the continual learning for instance segmentation in which instance-wise labeling and supervision are considered. Extensive experiments on three datasets for two tasks (i.e., continual semantic and instance segmentation are conducted to demonstrate that CISDQ achieves the state-of-the-art performance, specifically, obtaining 4.4% and 2.9% mIoU improvements for the ADE 100-10 (6 steps) setting and ADE 100-5 (11 steps) setting.