arXiv subjects

Kai Cheng

Publications and source records attributed to Kai Cheng.

At least 37 records · Page 2Linked to original sources

arXiv2024

Scaling New Frontiers: Insights into Large Recommendation Models

Recommendation systems are essential for filtering data and retrieving relevant information across various applications. Recent advancements have seen these systems incorporate increasingly large embedding tables, scaling up to tens of terabytes for industrial use. However, the expansion of network parameters in traditional recommendation models has plateaued at tens of millions, limiting further benefits from increased embedding parameters. Inspired by the success of large language models (LLMs), a new approach has emerged that scales network parameters using innovative structures, enabling continued performance improvements. A significant development in this area is Meta's generative recommendation model HSTU, which illustrates the scaling laws of recommendation systems by expanding parameters to thousands of billions. This new paradigm has achieved substantial performance gains in online experiments. In this paper, we aim to enhance the understanding of scaling laws by conducting comprehensive evaluations of large recommendation models. Firstly, we investigate the scaling laws across different backbone architectures of the large recommendation models. Secondly, we conduct comprehensive ablation studies to explore the origins of these scaling laws. We then further assess the performance of HSTU, as the representative of large recommendation models, on complex user behavior modeling tasks to evaluate its applicability. Notably, we also analyze its effectiveness in ranking tasks for the first time. Finally, we offer insights into future directions for large recommendation models. Supplementary materials for our research are available on GitHub at https://github.com/USTC-StarTeam/Large-Recommendation-Models.

cs.IR↗

arXiv2024

DC-Gaussian: Improving 3D Gaussian Splatting for Reflective Dash Cam Videos

We present DC-Gaussian, a new method for generating novel views from in-vehicle dash cam videos. While neural rendering techniques have made significant strides in driving scenarios, existing methods are primarily designed for videos collected by autonomous vehicles. However, these videos are limited in both quantity and diversity compared to dash cam videos, which are more widely used across various types of vehicles and capture a broader range of scenarios. Dash cam videos often suffer from severe obstructions such as reflections and occlusions on the windshields, which significantly impede the application of neural rendering techniques. To address this challenge, we develop DC-Gaussian based on the recent real-time neural rendering technique 3D Gaussian Splatting (3DGS). Our approach includes an adaptive image decomposition module to model reflections and occlusions in a unified manner. Additionally, we introduce illumination-aware obstruction modeling to manage reflections and occlusions under varying lighting conditions. Lastly, we employ a geometry-guided Gaussian enhancement strategy to improve rendering details by incorporating additional geometry priors. Experiments on self-captured and public dash cam videos show that our method not only achieves state-of-the-art performance in novel view synthesis, but also accurately reconstructing captured scenes getting rid of obstructions. See the project page for code, data: https://linhanwang.github.io/dcgaussian/.

cs.CV↗

arXiv2024

GlossyGS: Inverse Rendering of Glossy Objects with 3D Gaussian Splatting

Reconstructing objects from posed images is a crucial and complex task in computer graphics and computer vision. While NeRF-based neural reconstruction methods have exhibited impressive reconstruction ability, they tend to be time-comsuming. Recent strategies have adopted 3D Gaussian Splatting (3D-GS) for inverse rendering, which have led to quick and effective outcomes. However, these techniques generally have difficulty in producing believable geometries and materials for glossy objects, a challenge that stems from the inherent ambiguities of inverse rendering. To address this, we introduce GlossyGS, an innovative 3D-GS-based inverse rendering framework that aims to precisely reconstruct the geometry and materials of glossy objects by integrating material priors. The key idea is the use of micro-facet geometry segmentation prior, which helps to reduce the intrinsic ambiguities and improve the decomposition of geometries and materials. Additionally, we introduce a normal map prefiltering strategy to more accurately simulate the normal distribution of reflective surfaces. These strategies are integrated into a hybrid geometry and material representation that employs both explicit and implicit methods to depict glossy objects. We demonstrate through quantitative analysis and qualitative visualization that the proposed method is effective to reconstruct high-fidelity geometries and materials of glossy objects, and performs favorably against state-of-the-arts.

cs.CV↗

arXiv2024

Denoising Pre-Training and Customized Prompt Learning for Efficient Multi-Behavior Sequential Recommendation

In the realm of recommendation systems, users exhibit a diverse array of behaviors when interacting with items. This phenomenon has spurred research into learning the implicit semantic relationships between these behaviors to enhance recommendation performance. However, these methods often entail high computational complexity. To address concerns regarding efficiency, pre-training presents a viable solution. Its objective is to extract knowledge from extensive pre-training data and fine-tune the model for downstream tasks. Nevertheless, previous pre-training methods have primarily focused on single-behavior data, while multi-behavior data contains significant noise. Additionally, the fully fine-tuning strategy adopted by these methods still imposes a considerable computational burden. In response to this challenge, we propose DPCPL, the first pre-training and prompt-tuning paradigm tailored for Multi-Behavior Sequential Recommendation. Specifically, in the pre-training stage, we commence by proposing a novel Efficient Behavior Miner (EBM) to filter out the noise at multiple time scales, thereby facilitating the comprehension of the contextual semantics of multi-behavior sequences. Subsequently, we propose to tune the pre-trained model in a highly efficient manner with the proposed Customized Prompt Learning (CPL) module, which generates personalized, progressive, and diverse prompts to fully exploit the potential of the pre-trained model effectively. Extensive experiments on three real-world datasets have unequivocally demonstrated that DPCPL not only exhibits high efficiency and effectiveness, requiring minimal parameter adjustments but also surpasses the state-of-the-art performance across a diverse range of downstream tasks.

cs.IR↗

arXiv2024

PET: Multi-agent Independent PPO-based Automatic ECN Tuning for High-Speed Data Center Networks

Explicit Congestion Notification (ECN)-based congestion control schemes have been widely adopted in high-speed data center networks (DCNs), where the ECN marking threshold plays a determinant role in guaranteeing a packet lossless DCN. However, existing approaches either employ static settings with immutable thresholds that cannot be dynamically self-adjusted to adapt to network dynamics, or fail to take into account many-to-one traffic patterns and different requirements of different types of traffic, resulting in relatively poor performance. To address these problems, this paper proposes a novel learning-based automatic ECN tuning scheme, named PET, based on the multi-agent Independent Proximal Policy Optimization (IPPO) algorithm. PET dynamically adjusts ECN thresholds by fully considering pivotal congestion-contributing factors, including queue length, output data rate, output rate of ECN-marked packets, current ECN threshold, the extent of incast, and the ratio of mice and elephant flows. PET adopts the Decentralized Training and Decentralized Execution (DTDE) paradigm and combines offline and online training to accommodate network dynamics. PET is also fair and readily deployable with commodity hardware. Comprehensive experimental results demonstrate that, compared with state-of-the-art static schemes and the learning-based automatic scheme, our PET achieves better performance in terms of flow completion time, convergence rate, queue length variance, and system robustness.

cs.NI↗

arXiv2024

Smart Fitting Room: A One-stop Framework for Matching-aware Virtual Try-on

The development of virtual try-on has revolutionized online shopping by allowing customers to visualize themselves in various fashion items, thus extending the in-store try-on experience to the cyber space. Although virtual try-on has attracted considerable research initiatives, existing systems only focus on the quality of image generation, overlooking whether the fashion item is a good match to the given person and clothes. Recognizing this gap, we propose to design a one-stop Smart Fitting Room, with the novel formulation of matching-aware virtual try-on. Following this formulation, we design a Hybrid Matching-aware Virtual Try-On Framework (HMaVTON), which combines retrieval-based and generative methods to foster a more personalized virtual try-on experience. This framework integrates a hybrid mix-and-match module and an enhanced virtual try-on module. The former can recommend fashion items available on the platform to boost sales and generate clothes that meets the diverse tastes of consumers. The latter provides high-quality try-on effects, delivering a one-stop shopping service. To validate the effectiveness of our approach, we enlist the expertise of fashion designers for a professional evaluation, assessing the rationality and diversity of the clothes combinations and conducting an evaluation matrix analysis. Our method significantly enhances the practicality of virtual try-on. The code is available at https://github.com/Yzcreator/HMaVTON.

cs.MM↗

arXiv2024

LaserHuman: Language-guided Scene-aware Human Motion Generation in Free Environment

Language-guided scene-aware human motion generation has great significance for entertainment and robotics. In response to the limitations of existing datasets, we introduce LaserHuman, a pioneering dataset engineered to revolutionize Scene-Text-to-Motion research. LaserHuman stands out with its inclusion of genuine human motions within 3D environments, unbounded free-form natural language descriptions, a blend of indoor and outdoor scenarios, and dynamic, ever-changing scenes. Diverse modalities of capture data and rich annotations present great opportunities for the research of conditional motion generation, and can also facilitate the development of real-life applications. Moreover, to generate semantically consistent and physically plausible human motions, we propose a multi-conditional diffusion model, which is simple but effective, achieving state-of-the-art performance on existing datasets.

cs.CV↗

arXiv2024

GaussianPro: 3D Gaussian Splatting with Progressive Propagation

The advent of 3D Gaussian Splatting (3DGS) has recently brought about a revolution in the field of neural rendering, facilitating high-quality renderings at real-time speed. However, 3DGS heavily depends on the initialized point cloud produced by Structure-from-Motion (SfM) techniques. When tackling with large-scale scenes that unavoidably contain texture-less surfaces, the SfM techniques always fail to produce enough points in these surfaces and cannot provide good initialization for 3DGS. As a result, 3DGS suffers from difficult optimization and low-quality renderings. In this paper, inspired by classical multi-view stereo (MVS) techniques, we propose GaussianPro, a novel method that applies a progressive propagation strategy to guide the densification of the 3D Gaussians. Compared to the simple split and clone strategies used in 3DGS, our method leverages the priors of the existing reconstructed geometries of the scene and patch matching techniques to produce new Gaussians with accurate positions and orientations. Experiments on both large-scale and small-scale scenes validate the effectiveness of our method, where our method significantly surpasses 3DGS on the Waymo dataset, exhibiting an improvement of 1.15dB in terms of PSNR.

cs.CV↗

arXiv2024

Unveiling the Depths: A Multi-Modal Fusion Framework for Challenging Scenarios

Monocular depth estimation from RGB images plays a pivotal role in 3D vision. However, its accuracy can deteriorate in challenging environments such as nighttime or adverse weather conditions. While long-wave infrared cameras offer stable imaging in such challenging conditions, they are inherently low-resolution, lacking rich texture and semantics as delivered by the RGB image. Current methods focus solely on a single modality due to the difficulties to identify and integrate faithful depth cues from both sources. To address these issues, this paper presents a novel approach that identifies and integrates dominant cross-modality depth features with a learning-based framework. Concretely, we independently compute the coarse depth maps with separate networks by fully utilizing the individual depth cues from each modality. As the advantageous depth spreads across both modalities, we propose a novel confidence loss steering a confidence predictor network to yield a confidence map specifying latent potential depth areas. With the resulting confidence map, we propose a multi-modal fusion network that fuses the final depth in an end-to-end manner. Harnessing the proposed pipeline, our method demonstrates the ability of robust depth estimation in a variety of difficult scenarios. Experimental results on the challenging MS$^2$ and ViViD++ datasets demonstrate the effectiveness and robustness of our method.

cs.CV↗

arXiv2024

Sliced gradient-enhanced Kriging for high-dimensional function approximation

Gradient-enhanced Kriging (GE-Kriging) is a well-established surrogate modelling technique for approximating expensive computational models. However, it tends to get impractical for high-dimensional problems due to the size of the inherent correlation matrix and the associated high-dimensional hyper-parameter tuning problem. To address these issues, a new method, called sliced GE-Kriging (SGE-Kriging), is developed in this paper for reducing both the size of the correlation matrix and the number of hyper-parameters. We first split the training sample set into multiple slices, and invoke Bayes' theorem to approximate the full likelihood function via a sliced likelihood function, in which multiple small correlation matrices are utilized to describe the correlation of the sample set rather than one large one. Then, we replace the original high-dimensional hyper-parameter tuning problem with a low-dimensional counterpart by learning the relationship between the hyper-parameters and the derivative-based global sensitivity indices. The performance of SGE-Kriging is finally validated by means of numerical experiments with several benchmarks and a high-dimensional aerodynamic modeling problem. The results show that the SGE-Kriging model features an accuracy and robustness that is comparable to the standard one but comes at much less training costs. The benefits are most evident for high-dimensional problems with tens of variables.

stat.ML↗

arXiv2023

UC-NeRF: Neural Radiance Field for Under-Calibrated Multi-view Cameras in Autonomous Driving

Multi-camera setups find widespread use across various applications, such as autonomous driving, as they greatly expand sensing capabilities. Despite the fast development of Neural radiance field (NeRF) techniques and their wide applications in both indoor and outdoor scenes, applying NeRF to multi-camera systems remains very challenging. This is primarily due to the inherent under-calibration issues in multi-camera setup, including inconsistent imaging effects stemming from separately calibrated image signal processing units in diverse cameras, and system errors arising from mechanical vibrations during driving that affect relative camera poses. In this paper, we present UC-NeRF, a novel method tailored for novel view synthesis in under-calibrated multi-view camera systems. Firstly, we propose a layer-based color correction to rectify the color inconsistency in different image regions. Second, we propose virtual warping to generate more viewpoint-diverse but color-consistent virtual views for color correction and 3D recovery. Finally, a spatiotemporally constrained pose refinement is designed for more robust and accurate pose calibration in multi-camera systems. Our method not only achieves state-of-the-art performance of novel view synthesis in multi-camera setups, but also effectively facilitates depth estimation in large-scale outdoor scenes with the synthesized novel views.

cs.CV↗

arXiv2023

FrozenRecon: Pose-free 3D Scene Reconstruction with Frozen Depth Models

3D scene reconstruction is a long-standing vision task. Existing approaches can be categorized into geometry-based and learning-based methods. The former leverages multi-view geometry but can face catastrophic failures due to the reliance on accurate pixel correspondence across views. The latter was proffered to mitigate these issues by learning 2D or 3D representation directly. However, without a large-scale video or 3D training data, it can hardly generalize to diverse real-world scenarios due to the presence of tens of millions or even billions of optimization parameters in the deep network. Recently, robust monocular depth estimation models trained with large-scale datasets have been proven to possess weak 3D geometry prior, but they are insufficient for reconstruction due to the unknown camera parameters, the affine-invariant property, and inter-frame inconsistency. Here, we propose a novel test-time optimization approach that can transfer the robustness of affine-invariant depth models such as LeReS to challenging diverse scenes while ensuring inter-frame consistency, with only dozens of parameters to optimize per video frame. Specifically, our approach involves freezing the pre-trained affine-invariant depth model's depth predictions, rectifying them by optimizing the unknown scale-shift values with a geometric consistency alignment module, and employing the resulting scale-consistent depth maps to robustly obtain camera poses and achieve dense scene reconstruction, even in low-texture regions. Experiments show that our method achieves state-of-the-art cross-dataset reconstruction on five zero-shot testing datasets.

cs.CV↗

arXiv2023

Piezoelectric altermagnetism and spin-valley polarization in Janus monolayer $\mathrm{Cr_2SO}$

The altermagnetism can achieve spin-split bands in collinear symmetry-compensated antiferromagnets. Here, we predict altermagnetic order in Janus monolayer $\mathrm{Cr_2SO}$ with eliminated inversion symmetry, which can realize the combination of piezoelectricity and altermagnetism in a two-dimensional material, namely 2D piezoelectric altermagnetism. It is found that $\mathrm{Cr_2SO}$ is an altermagnetic semiconductor, and the spin-split bands of both valence and conduction bands are near the Fermi level. The $\mathrm{Cr_2SO}$ has large out-of-plane piezoelectricity ($|d_{31}|$$=$0.97 pm/V), which is highly desirable for ultrathin piezoelectric device application. Due to spin-valley locking, both spin and valley can be polarized by simply breaking the corresponding crystal symmetry with uniaxial strain. Our findings provide a platform to integrate spin, piezoelectricity and valley in a single material, which is useful for multi-functional device applications.

cond-mat.mtrl-sci↗

arXiv2023

The Second Monocular Depth Estimation Challenge

This paper discusses the results for the second edition of the Monocular Depth Estimation Challenge (MDEC). This edition was open to methods using any form of supervision, including fully-supervised, self-supervised, multi-task or proxy depth. The challenge was based around the SYNS-Patches dataset, which features a wide diversity of environments with high-quality dense ground-truth. This includes complex natural environments, e.g. forests or fields, which are greatly underrepresented in current benchmarks. The challenge received eight unique submissions that outperformed the provided SotA baseline on any of the pointcloud- or image-based metrics. The top supervised submission improved relative F-Score by 27.62%, while the top self-supervised improved it by 16.61%. Supervised submissions generally leveraged large collections of datasets to improve data diversity. Self-supervised submissions instead updated the network architecture and pretrained backbones. These results represent a significant progress in the field, while highlighting avenues for future research, such as reducing interpolation artifacts at depth boundaries, improving self-supervised indoor performance and overall natural image accuracy.

cs.CV↗

arXiv2023

Towards 3D Scene Reconstruction from Locally Scale-Aligned Monocular Video Depth

Existing monocular depth estimation methods have achieved excellent robustness in diverse scenes, but they can only retrieve affine-invariant depth, up to an unknown scale and shift. However, in some video-based scenarios such as video depth estimation and 3D scene reconstruction from a video, the unknown scale and shift residing in per-frame prediction may cause the depth inconsistency. To solve this problem, we propose a locally weighted linear regression method to recover the scale and shift with very sparse anchor points, which ensures the scale consistency along consecutive frames. Extensive experiments show that our method can boost the performance of existing state-of-the-art approaches by 50% at most over several zero-shot benchmarks. Besides, we merge over 6.3 million RGBD images to train strong and robust depth models. Our produced ResNet50-backbone model even outperforms the state-of-the-art DPT ViT-Large model. Combining with geometry-based reconstruction methods, we formulate a new dense 3D scene reconstruction pipeline, which benefits from both the scale consistency of sparse points and the robustness of monocular methods. By performing the simple per-frame prediction over a video, the accurate 3D scene shape can be recovered.

cs.CV↗

arXiv2023

Janus monolayer ScXY (X$\neq$Y=Cl, Br and I) for piezoelectric and valleytronic application: a first-principle prediction

Coexistence of ferromagnetism, piezoelectricity and valley in two-dimensional (2D) materials is crucial to advance multifunctional electronic technologies. Here, Janus ScXY (X$\neq$Y=Cl, Br and I) monolayers are predicted to be in-plane piezoelectric ferromagnetic (FM) semiconductors with dynamical, mechanical and thermal stabilities. The predicted piezoelectric strain coefficients $d_{11}$ and $d_{31}$ (absolute values) are higher than ones of most 2D materials. Moreover, the $d_{31}$ (absolute value) of ScClI reaches up to 1.14 pm/V, which is highly desirable for ultrathin piezoelectric device application. To obtain spontaneous valley polarization, charge doping are explored to tune the direction of magnetization of ScXY. By appropriate hole doping, their easy magnetization axis can change from in-plane to out-of-plane, resulting in spontaneous valley polarization. Taking ScBrI with 0.20 holes per f.u. as a example, under the action of an in-plane electric field, the hole carriers of K valley turn towards one edge of the sample, which will produce anomalous valley Hall effect (AVHE), and the hole carriers of $Γ$ valley move in a straight line. These findings could pave the way for designing piezoelectric and valleytronic devices.

cond-mat.mtrl-sci↗

arXiv2022

Electric-field induced magnetic-anisotropy transformation to achieve spontaneous valley polarization

Valleytronics has been widely investigated for providing new degrees of freedom to future information coding and processing. Here, it is proposed that valley polarization can be achieved by electric field induced magnetic anisotropy (MA) transformation. Through the first-principle calculations, our idea is illustrated by a concrete example of $\mathrm{VSi_2P_4}$ monolayer. The increasing electric field can induce a transition of MA from in-plane to out-of-plane by changing magnetic anisotropy energy (MAE) from negative to positive value, which is mainly due to increasing magnetocrystalline anisotropy (MCA) energy. The out-of-plane magnetization is in favour of spontaneous valley polarization in $\mathrm{VSi_2P_4}$. Within considered electric field range, $\mathrm{VSi_2P_4}$ is always ferromagnetic (FM) ground state. In a certain range of electric field, the coexistence of semiconductor and out-of-plane magnetization makes $\mathrm{VSi_2P_4}$ become a true ferrovalley (FV) material. The anomalous valley Hall effect (AVHE) can be observed under in-plane and out-of-plane electrical field in $\mathrm{VSi_2P_4}$. Our works pave the way to design the ferrovalley material by electric field.

cond-mat.mtrl-sci↗

arXiv2022

Importance of magnetic shape anisotropy in determining magnetic and electronic properties of monolayer $\mathrm{VSi_2P_4}$

Two-dimensional (2D) ferromagnets have been a fascinating subject of research, and magnetic anisotropy (MA) is indispensable for stabilizing the 2D magnetic order. Here, we investigate magnetic anisotropy energy (MAE), magnetic and electronic properties of $\mathrm{VSi_2P_4}$ by using the generalized gradient approximation plus $U$ (GGA+$U$) approach. For large $U$, the magnetic shape anisotropy (MSA) energy has a more pronounced contribution to the MAE, which can overcome the magnetocrystalline anisotropy (MCA) energy to evince an easy-plane. For fixed out-of-plane MA, monolayer $\mathrm{VSi_2P_4}$ undergoes ferrovalley (FV), half-valley-metal (HVM), valley-polarized quantum anomalous Hall insulator (VQAHI), HVM and FV states with increasing $U$. However, for assumptive in-plane MA, there is no special quantum anomalous Hall (QAH) state and spontaneous valley polarization within considered $U$ range. According to the MAE and electronic structure with fixed out-of-plane or in-plane MA, the intrinsic phase diagram shows common magnetic semiconductor (CMS), FV and VQAHI in monolayer $\mathrm{VSi_2P_4}$. At representative $U$$=$3 eV widely used in references, $\mathrm{VSi_2P_4}$ can be regarded as a 2D-$XY$ magnet, not Ising-like 2D long-range order magnets predicted in previous works with only considering MCA energy. Our findings shed light on importance of MSA in determining magnetic and electronic properties of monolayer $\mathrm{VSi_2P_4}$.

cond-mat.mtrl-sci↗