Search

We establish a uniform-in-time estimate for the mean-field convergence of the Consensus-Based Optimization (CBO) algorithm by rescaling the consensus point in the dynamics with a small parameter $\kappa \in (0,1)$. This uniform-in-time estimate is essential, as CBO convergence relies on a sufficiently large time horizon and is crucial for ensuring stable, reliable long-term convergence, the latter being key to the practical effectiveness of CBO methods.

Long-tail learning has garnered widespread attention and achieved significant progress in recent times. However, even with pre-trained prior knowledge, models still exhibit weaker generalization performance on tail classes. The promising Sharpness-Aware Minimization (SAM) can effectively improve the generalization capability of models by seeking out flat minima in the loss landscape, which, however, comes at the cost of doubling the computational time. Since the update rule of SAM necessitates two consecutive (non-parallelizable) forward and backpropagation at each step. To address this issue, we propose a novel method called Random SAM prompt tuning (RSAM-PT) to improve the model generalization, requiring only one-step gradient computation at each step. Specifically, we search for the gradient descent direction within a random neighborhood of the parameters during each gradient update. To amplify the impact of tail-class samples and avoid overfitting, we employ the deferred re-weight scheme to increase the significance of tail-class samples. The classification accuracy of long-tailed data can be significantly improved by the proposed RSAM-PT, particularly for tail classes. RSAM-PT achieves the state-of-the-art performance of 90.3\%, 76.5\%, and 50.1\% on benchmark datasets CIFAR100-LT (IF 100), iNaturalist 2018, and Places-LT, respectively. The source code is temporarily available at https://github.com/Keke921/GNM-PT., Comment: NeurIPS 2024

Recent advancements in Direct Preference Optimization (DPO) have significantly enhanced the alignment of Large Language Models (LLMs) with human preferences, owing to its simplicity and effectiveness. However, existing methods typically optimize a scalar score or ranking reward, thereby overlooking the multi-dimensional nature of human preferences. In this work, we propose to extend the preference of DPO to two dimensions: segments and aspects. We first introduce a 2D supervision dataset called HelpSteer-2D. For the segment dimension, we divide the response into sentences and assign scores to each segment. For the aspect dimension, we meticulously design several criteria covering the response quality rubrics. With the 2-dimensional signals as feedback, we develop a 2D-DPO framework, decomposing the overall objective into multi-segment and multi-aspect objectives. Extensive experiments on popular benchmarks demonstrate that 2D-DPO performs better than methods that optimize for scalar or 1-dimensional preferences., Comment: The first four authors contributed equally, 25 pages

Consensus-based optimization (CBO) is a versatile multi-particle optimization method for performing nonconvex and nonsmooth global optimizations in high dimensions. Proofs of global convergence in probability have been achieved for a broad class of objective functions in unconstrained optimizations. In this work we adapt the algorithm for solving constrained optimizations on compact and unbounded domains with boundary by leveraging emerging reflective boundary conditions. In particular, we close a relevant gap in the literature by providing a global convergence proof for the many-particle regime comprehensive of convergence rates. On the one hand, for the sake of minimizing running cost, it is desirable to keep the number of particles small. On the other hand, reducing the number of particles implies a diminished capability of exploration of the algorithm. Hence numerical heuristics are needed to ensure convergence of CBO in the few-particle regime. In this work, we also significantly improve the convergence and complexity of CBO by utilizing an adaptive region control mechanism and by choosing geometry-specific random noise. In particular, by combining a hierarchical noise structure with a multigrid finite element method, we are able to compute global minimizers for a constrained $p$-Allen-Cahn problem with obstacles, a very challenging variational problem.

While numerous indices of inter-coder reliability exist, Krippendorff's {\alpha} and Cohen's \{kappa} have long dominated in communication studies and other fields, respectively. The near consensus, however, may be near the end. Recent theoretical and mathematical analyses reveal that these indices assume intentional and maximal random coding, leading to paradoxes and inaccuracies. A controlled experiment with one-way golden standard and Monte Carlo simulations supports these findings, showing that \{kappa} and {\alpha} are poor predictors and approximators of true intercoder reliability. As consensus on a perfect index remains elusive, more authors recommend selecting the best available index for specific situations (BAFS). To make informed choices, researchers, reviewers, and educators need to understand the liberal-conservative hierarchy of indices, i.e., which indices produce higher or lower scores. This study extends previous efforts by expanding the math-based hierarchies to include 23 indices and constructing six additional hierarchies using Monte Carlo simulations. These simulations account for factors like the number of categories and distribution skew. The resulting eight hierarchies display a consistent pattern and reveal a previously undetected paradox in the Ir index., Comment: 30 pages

Recently, there has been a trend of evaluating the Large Language Model (LLM) quality in the flavor of LLM-as-a-Judge, namely leveraging another LLM to evaluate the current output quality. However, existing judges are proven to be biased, namely they would favor answers which present better superficial quality (such as verbosity, fluency) while ignoring the instruction following ability. In this work, we propose systematic research about the bias of LLM-as-a-Judge. Specifically, for closed-source judge models, we apply calibration to mitigate the significance of superficial quality, both on probability level and prompt level. For open-source judge models, we propose to mitigate the bias by contrastive training, with curated negative samples that deviate from instruction but present better superficial quality. We apply our methods on the bias evaluation benchmark, and experiment results show our methods mitigate the bias by a large margin while maintaining a satisfactory evaluation accuracy.

Managing the level-of-detail (LOD) in architectural models is crucial yet challenging, particularly for effective representation and visualization of buildings. Traditional approaches often fail to deliver controllable detail alongside semantic consistency, especially when dealing with noisy and inconsistent inputs. We address these limitations with \emph{Co-LOD}, a new approach specifically designed for effective LOD management in architectural modeling. Co-LOD employs shape co-analysis to standardize geometric structures across multiple buildings, facilitating the progressive and consistent generation of LODs. This method allows for precise detailing in both individual models and model collections, ensuring semantic integrity. Extensive experiments demonstrate that Co-LOD effectively applies accurate LOD across a variety of architectural inputs, consistently delivering superior detail and quality in LOD representations., Comment: ACM Transactions on Graphics (SIGGRAPH Aisa 2024); Project page: https://vcc.tech/research/2024/CoLOD

Legal facts refer to the facts that can be proven by acknowledged evidence in a trial. They form the basis for the determination of court judgments. This paper introduces a novel NLP task: legal fact prediction, which aims to predict the legal fact based on a list of evidence. The predicted facts can instruct the parties and their lawyers involved in a trial to strengthen their submissions and optimize their strategies during the trial. Moreover, since real legal facts are difficult to obtain before the final judgment, the predicted facts also serve as an important basis for legal judgment prediction. We construct a benchmark dataset consisting of evidence lists and ground-truth legal facts for real civil loan cases, LFPLoan. Our experiments on this dataset show that this task is non-trivial and requires further considerable research efforts.

Diffusion models have achieved notable success in image generation, but they remain highly vulnerable to backdoor attacks, which compromise their integrity by producing specific undesirable outputs when presented with a pre-defined trigger. In this paper, we investigate how to protect diffusion models from this dangerous threat. Specifically, we propose TERD, a backdoor defense framework that builds unified modeling for current attacks, which enables us to derive an accessible reversed loss. A trigger reversion strategy is further employed: an initial approximation of the trigger through noise sampled from a prior distribution, followed by refinement through differential multi-step samplers. Additionally, with the reversed trigger, we propose backdoor detection from the noise space, introducing the first backdoor input detection approach for diffusion models and a novel model detection algorithm that calculates the KL divergence between reversed and benign distributions. Extensive evaluations demonstrate that TERD secures a 100% True Positive Rate (TPR) and True Negative Rate (TNR) across datasets of varying resolutions. TERD also demonstrates nice adaptability to other Stochastic Differential Equation (SDE)-based models. Our code is available at https://github.com/PKU-ML/TERD.

Multi-view crowd localization predicts the ground locations of all people in the scene. Typical methods usually estimate the crowd density maps on the ground plane first, and then obtain the crowd locations. However, the performance of existing methods is limited by the ambiguity of the density maps in crowded areas, where local peaks can be smoothed away. To mitigate the weakness of density map supervision, optimal transport-based point supervision methods have been proposed in the single-image crowd localization tasks, but have not been explored for multi-view crowd localization yet. Thus, in this paper, we propose a novel Mahalanobis distance-based multi-view optimal transport (M-MVOT) loss specifically designed for multi-view crowd localization. First, we replace the Euclidean-based transport cost with the Mahalanobis distance, which defines elliptical iso-contours in the cost function whose long-axis and short-axis directions are guided by the view ray direction. Second, the object-to-camera distance in each view is used to adjust the optimal transport cost of each location further, where the wrong predictions far away from the camera are more heavily penalized. Finally, we propose a strategy to consider all the input camera views in the model loss (M-MVOT) by computing the optimal transport cost for each ground-truth point based on its closest camera. Experiments demonstrate the advantage of the proposed method over density map-based or common Euclidean distance-based optimal transport loss on several multi-view crowd localization datasets. Project page: https://vcc.tech/research/2024/MVOT., Comment: ECCV 2024

Irregularly observed functional time series (FTS) are increasingly available in many real-world applications. To analyze FTS, it is crucial to account for both serial dependencies and the irregularly observed nature of functional data. However, existing methods for FTS often rely on specific model assumptions in capturing serial dependencies, or cannot handle the irregular observational scheme of functional data. To solve these issues, one can perform dimension reduction on FTS via functional principal component analysis (FPCA) or dynamic FPCA. Nonetheless, these methods may either be not theoretically optimal or too redundant to represent serially dependent functional data. In this article, we introduce a novel dimension reduction method for FTS based on dynamic FPCA. Through a new concept called optimal functional filters, we unify the theories of FPCA and dynamic FPCA, providing a parsimonious and optimal representation for FTS adapting to its serial dependence structure. This framework is referred to as principal analysis via dependency-adaptivity (PADA). Under a hierarchical Bayesian model, we establish an estimation procedure for dimension reduction via PADA. Our method can be used for both sparsely and densely observed FTS, and is capable of predicting future functional data. We investigate the theoretical properties of PADA and demonstrate its effectiveness through extensive simulation studies. Finally, we illustrate our method via dimension reduction and prediction of daily PM2.5 data.

Polygonal mesh reconstruction of a raw point cloud is a valuable topic in the field of computer graphics and 3D vision. Especially to 3D architectural models, polygonal mesh provides concise expressions for fundamental geometric structures while effectively reducing data volume. However, there are some limitations of traditional reconstruction methods: normal vector dependency, noisy points and defective parts sensitivity, and internal geometric structure lost, which reduce the practicality in real scene. In this paper, we propose a robust and efficient polygonal mesh reconstruction method to address the issues in architectural point cloud reconstruction task. It is an iterative adaptation process to detect planar shapes from scattered points. The initial structural polygonal mesh can be established in the constructed convex polyhedral space without assistance of normal vectors. Then, we develop an efficient polygon-based winding number strategy to orient polygonal mesh with global consistency. The significant advantage of our method is to provide a structural reconstruction for architectural point clouds and avoid point-based normal vector analysis. It effectively improves the robustness to noisy points and defective parts. More geometric details can be preserved in the reconstructed polygonal mesh. Experimental results show that our method can reconstruct concise, oriented and faithfully polygonal mesh that are better than results of state-of-the-art methods. More results and details can be found on https://vcc.tech/research/2024/WindPoly, Comment: European Conference on Computer Vision (Proceedings of ECCV 2024)

We propose a zero-order optimization method for sequential min-max problems based on two populations of interacting particles. The systems are coupled so that one population aims to solve the inner maximization problem, while the other aims to solve the outer minimization problem. The dynamics are characterized by a consensus-type interaction with additional stochasticity to promote exploration of the objective landscape. Without relying on convexity or concavity assumptions, we establish theoretical convergence guarantees of the algorithm via a suitable mean-field approximation of the particle systems. Numerical experiments illustrate the validity of the proposed approach. In particular, the algorithm is able to identify a global min-max solution, in contrast to gradient-based methods, which typically converge to possibly suboptimal stationary points.

Pre-trained large-scale models have exhibited remarkable efficacy in computer vision, particularly for 2D image analysis. However, when it comes to 3D point clouds, the constrained accessibility of data, in contrast to the vast repositories of images, poses a challenge for the development of 3D pre-trained models. This paper therefore attempts to directly leverage pre-trained models with 2D prior knowledge to accomplish the tasks for 3D point cloud analysis. Accordingly, we propose the Adaptive PointFormer (APF), which fine-tunes pre-trained 2D models with only a modest number of parameters to directly process point clouds, obviating the need for mapping to images. Specifically, we convert raw point clouds into point embeddings for aligning dimensions with image tokens. Given the inherent disorder in point clouds, in contrast to the structured nature of images, we then sequence the point embeddings to optimize the utilization of 2D attention priors. To calibrate attention across 3D and 2D domains and reduce computational overhead, a trainable PointFormer with a limited number of parameters is subsequently concatenated to a frozen pre-trained image model. Extensive experiments on various benchmarks demonstrate the effectiveness of the proposed APF. The source code and more details are available at https://vcc.tech/research/2024/PointFormer., Comment: ECAI 2024 main conference paper

We present a new approach for generating 3D house wireframes with semantic enrichment using an autoregressive model. Unlike conventional generative models that independently process vertices, edges, and faces, our approach employs a unified wire-based representation for improved coherence in learning 3D wireframe structures. By re-ordering wire sequences based on semantic meanings, we facilitate seamless semantic integration during sequence generation. Our two-phase technique merges a graph-based autoencoder with a transformer-based decoder to learn latent geometric tokens and generate semantic-aware wireframes. Through iterative prediction and decoding during inference, our model produces detailed wireframes that can be easily segmented into distinct components, such as walls, roofs, and rooms, reflecting the semantic essence of the shape. Empirical results on a comprehensive house dataset validate the superior accuracy, novelty, and semantic fidelity of our model compared to existing generative models. More results and details can be found on https://vcc.tech/research/2024/3DWire., Comment: European Conference on Computer Vision (Proceedings of ECCV 2024); Project page: https://vcc.tech/research/2024/3DWire; GitHub repository: https://github.com/3d-house-wireframe/3d-house-wireframe-dataset

In this paper, we introduce a novel method called FRI-Net for 2D floorplan reconstruction from 3D point cloud. Existing methods typically rely on corner regression or box regression, which lack consideration for the global shapes of rooms. To address these issues, we propose a novel approach using a room-wise implicit representation with structural regularization to characterize the shapes of rooms in floorplans. By incorporating geometric priors of room layouts in floorplans into our training strategy, the generated room polygons are more geometrically regular. We have conducted experiments on two challenging datasets, Structured3D and SceneCAD. Our method demonstrates improved performance compared to state-of-the-art methods, validating the effectiveness of our proposed representation for floorplan reconstruction., Comment: ECCV 2024

In this paper, we present a rigorous derivation of the mean-field limit for a moderately interacting particle system in $\R^d$ $(d\geq 2)$. For stochastic initial data, we demonstrate that the solution to the interacting particle model, with an appropriately applied cut-off, converges in probabilistic sense to the solution of the characteristics of the regularized Vlasov models featuring local alignments and Newtonian potential. Notably, the cutoff parameter for the singular potential is selected to scale polynomially with the number of particles, representing an improvement over the logarithmic cut-off obtained in [38].

Understanding the longitudinally changing associations between Social Determinants of Health (SDOH) and stroke mortality is essential for effective stroke management. Previous studies have uncovered significant regional disparities in the relationships between SDOH and stroke mortality. However, existing studies have not utilized longitudinal associations to develop data-driven methods for regional division in stroke control. To fill this gap, we propose a novel clustering method to analyze SDOH -- stroke mortality associations in US counties. To enhance the interpretability of the clustering outcomes, we introduce a novel regularized expectation-maximization algorithm equipped with various sparsity-and-smoothness-pursued penalties, aiming at simultaneous clustering and variable selection in longitudinal associations. As a result, we can identify crucial SDOH that contribute to longitudinal changes in stroke mortality. This facilitates the clustering of US counties into different regions based on the relationships between these SDOH and stroke mortality. The effectiveness of our proposed method is demonstrated through extensive numerical studies. By applying our method to longitudinal data on SDOH and stroke mortality at the county level, we identify 18 important SDOH for stroke mortality and divide the US counties into two clusters based on these selected SDOH. Our findings unveil complex regional heterogeneity in the longitudinal associations between SDOH and stroke mortality, providing valuable insights into region-specific SDOH adjustments for mitigating stroke mortality.

We introduce a novel method for acquiring boundary representations (B-Reps) of 3D CAD models which involves a two-step process: it first applies a spatial partitioning, referred to as the ``split``, followed by a ``fit`` operation to derive a single primitive within each partition. Specifically, our partitioning aims to produce the classical Voronoi diagram of the set of ground-truth (GT) B-Rep primitives. In contrast to prior B-Rep constructions which were bottom-up, either via direct primitive fitting or point clustering, our Split-and-Fit approach is top-down and structure-aware, since a Voronoi partition explicitly reveals both the number of and the connections between the primitives. We design a neural network to predict the Voronoi diagram from an input point cloud or distance field via a binary classification. We show that our network, coined NVD-Net for neural Voronoi diagrams, can effectively learn Voronoi partitions for CAD models from training data and exhibits superior generalization capabilities. Extensive experiments and evaluation demonstrate that the resulting B-Reps, consisting of parametric surfaces, curves, and vertices, are more plausible than those obtained by existing alternatives, with significant improvements in reconstruction quality. Code will be released on https://github.com/yilinliu77/NVDNet., Comment: ACM Transactions on Graphics (SIGGRAPH 2024); Project page: https://vcc.tech/research/2024/BRepVP; Code: https://github.com/yilinliu77/NVDNet

Recent deep learning-based multi-view people detection (MVD) methods have shown promising results on existing datasets. However, current methods are mainly trained and evaluated on small, single scenes with a limited number of multi-view frames and fixed camera views. As a result, these methods may not be practical for detecting people in larger, more complex scenes with severe occlusions and camera calibration errors. This paper focuses on improving multi-view people detection by developing a supervised view-wise contribution weighting approach that better fuses multi-camera information under large scenes. Besides, a large synthetic dataset is adopted to enhance the model's generalization ability and enable more practical evaluation and comparison. The model's performance on new testing scenes is further improved with a simple domain adaptation technique. Experimental results demonstrate the effectiveness of our approach in achieving promising cross-scene multi-view people detection performance. See code here: https://vcc.tech/research/2024/MVD., Comment: AAAI 2024

Affective Image Manipulation (AIM) seeks to modify user-provided images to evoke specific emotional responses. This task is inherently complex due to its twofold objective: significantly evoking the intended emotion, while preserving the original image composition. Existing AIM methods primarily adjust color and style, often failing to elicit precise and profound emotional shifts. Drawing on psychological insights, we extend AIM by incorporating content modifications to enhance emotional impact. We introduce EmoEdit, a novel two-stage framework comprising emotion attribution and image editing. In the emotion attribution stage, we leverage a Vision-Language Model (VLM) to create hierarchies of semantic factors that represent abstract emotions. In the image editing stage, the VLM identifies the most relevant factors for the provided image, and guides a generative editing model to perform affective modifications. A ranking technique that we developed selects the best edit, balancing between emotion fidelity and structure integrity. To validate EmoEdit, we assembled a dataset of 416 images, categorized into positive, negative, and neutral classes. Our method is evaluated both qualitatively and quantitatively, demonstrating superior performance compared to existing state-of-the-art techniques. Additionally, we showcase EmoEdit's potential in various manipulation tasks, including emotion-oriented and semantics-oriented editing.

In this paper, we introduce LGTM, a novel Local-to-Global pipeline for Text-to-Motion generation. LGTM utilizes a diffusion-based architecture and aims to address the challenge of accurately translating textual descriptions into semantically coherent human motion in computer animation. Specifically, traditional methods often struggle with semantic discrepancies, particularly in aligning specific motions to the correct body parts. To address this issue, we propose a two-stage pipeline to overcome this challenge: it first employs large language models (LLMs) to decompose global motion descriptions into part-specific narratives, which are then processed by independent body-part motion encoders to ensure precise local semantic alignment. Finally, an attention-based full-body optimizer refines the motion generation results and guarantees the overall coherence. Our experiments demonstrate that LGTM gains significant improvements in generating locally accurate, semantically-aligned human motion, marking a notable advancement in text-to-motion applications. Code and data for this paper are available at https://github.com/L-Sun/LGTM, Comment: 9 pages,7 figures, SIGGRAPH 2024

In this paper, we introduce a new method for the task of interaction transfer. Given an example interaction between a source object and an agent, our method can automatically infer both surface and spatial relationships for the agent and target objects within the same category, yielding more accurate and valid transfers. Specifically, our method characterizes the example interaction using a combined spatial and surface representation. We correspond the agent points and object points related to the representation to the target object space using a learned spatial and surface correspondence field, which represents objects as deformed and rotated signed distance fields. With the corresponded points, an optimization is performed under the constraints of our spatial and surface interaction representation and additional regularization. Experiments conducted on human-chair and hand-mug interaction transfer tasks show that our approach can handle larger geometry and topology variations between source and target shapes, significantly outperforming state-of-the-art methods., Comment: Accepted to SIGGRAPH 2024, project page at https://vcc.tech/research/2024/InterTransfer

In this paper, we present an innovative particle system characterized by moderate interactions, designed to accurately approximate kinetic flocking models that incorporate singular interaction forces and local alignment mechanisms. We establish the existence of weak solutions to the corresponding flocking equations and provide an error estimate for the mean-field limit. This is achieved through the regularization of singular forces and a nonlocal approximation strategy for local alignments. We show that, by selecting the regularization and localization parameters logarithmically with respect to the number of particles, the particle system effectively approximates the mean-field equation.

Over the past decade, a series of unflagging efforts have been dedicated to developing highly expressive and controllable text-to-speech (TTS) systems. In general, the holistic TTS comprises two interconnected components: the frontend module and the backend module. The frontend excels in capturing linguistic representations from the raw text input, while the backend module converts linguistic cues to speech. The research community has shown growing interest in the study of the frontend component, recognizing its pivotal role in text-to-speech systems, including Text Normalization (TN), Prosody Boundary Prediction (PBP), and Polyphone Disambiguation (PD). Nonetheless, the limitations posed by insufficient annotated textual data and the reliance on homogeneous text signals significantly undermine the effectiveness of its supervised learning. To evade this obstacle, a novel two-stage TTS frontend prediction pipeline, named TAP-FM, is proposed in this paper. Specifically, during the first learning phase, we present a Multi-scale Contrastive Text-audio Pre-training protocol (MC-TAP), which hammers at acquiring richer insights via multi-granularity contrastive pre-training in an unsupervised manner. Instead of mining homogeneous features in prior pre-training approaches, our framework demonstrates the ability to delve deep into both global and local text-audio semantic and acoustic representations. Furthermore, a parallelized TTS frontend model is delicately devised to execute TN, PD, and PBP prediction tasks, respectively in the second stage. Finally, extensive experiments illustrate the superiority of our proposed method, achieving state-of-the-art performance.

We present a novel topology-preserving 3D medial axis computation framework based on volumetric restricted power diagram (RPD), while preserving the medial features and geometric convergence simultaneously, for both 3D CAD and organic shapes. The volumetric RPD discretizes the input 3D volume into sub-regions given a set of medial spheres. With this intermediate structure, we convert the homotopy equivalency between the generated medial mesh and the input 3D shape into a localized contractibility checking for each restricted element (power cell, power face, power edge), by checking their connected components and Euler characteristics. We further propose a fractional Euler characteristic algorithm for efficient GPU-based computation of Euler characteristic for each restricted element on the fly while computing the volumetric RPD. Compared with existing voxel-based or point-cloud-based methods, our approach is the first to adaptively and directly revise the medial mesh without globally modifying the dependent structure, such as voxel size or sampling density, while preserving its topology and medial features. In comparison with the feature preservation method MATFP, our method provides geometrically comparable results with fewer spheres and more robustly captures the topology of the input 3D shape.

In this study, we tackle the complex task of generating 3D human-object interactions (HOI) from textual descriptions in a zero-shot text-to-3D manner. We identify and address two key challenges: the unsatisfactory outcomes of direct text-to-3D methods in HOI, largely due to the lack of paired text-interaction data, and the inherent difficulties in simultaneously generating multiple concepts with complex spatial relationships. To effectively address these issues, we present InterFusion, a two-stage framework specifically designed for HOI generation. InterFusion involves human pose estimations derived from text as geometric priors, which simplifies the text-to-3D conversion process and introduces additional constraints for accurate object generation. At the first stage, InterFusion extracts 3D human poses from a synthesized image dataset depicting a wide range of interactions, subsequently mapping these poses to interaction descriptions. The second stage of InterFusion capitalizes on the latest developments in text-to-3D generation, enabling the production of realistic and high-quality 3D HOI scenes. This is achieved through a local-global optimization process, where the generation of human body and object is optimized separately, and jointly refined with a global optimization of the entire scene, ensuring a seamless and contextually coherent integration. Our experimental results affirm that InterFusion significantly outperforms existing state-of-the-art methods in 3D HOI generation., Comment: ECCV 2024; Project page https://sisidai.github.io/InterFusion/; Code available at https://github.com/sisidai/InterFusion

Parameters of differential equations are essential to characterize intrinsic behaviors of dynamic systems. Numerous methods for estimating parameters in dynamic systems are computationally and/or statistically inadequate, especially for complex systems with general-order differential operators, such as motion dynamics. This article presents Green's matching, a computationally tractable and statistically efficient two-step method, which only needs to approximate trajectories in dynamic systems but not their derivatives due to the inverse of differential operators by Green's function. This yields a statistically optimal guarantee for parameter estimation in general-order equations, a feature not shared by existing methods, and provides an efficient framework for broad statistical inferences in complex dynamic systems., Comment: 40 pages, 4 figures

Synchronized dual-arm rearrangement is widely studied as a common scenario in industrial applications. It often faces scalability challenges due to the computational complexity of robotic arm rearrangement and the high-dimensional nature of dual-arm planning. To address these challenges, we formulated the problem as cooperative mTSP, a variant of mTSP where agents share cooperative costs, and utilized reinforcement learning for its solution. Our approach involved representing rearrangement tasks using a task state graph that captured spatial relationships and a cooperative cost matrix that provided details about action costs. Taking these representations as observations, we designed an attention-based network to effectively combine them and provide rational task scheduling. Furthermore, a cost predictor is also introduced to directly evaluate actions during both training and planning, significantly expediting the planning process. Our experimental results demonstrate that our approach outperforms existing methods in terms of both performance and planning efficiency.

The proliferation of open-source Large Language Models (LLMs) underscores the pressing need for evaluation methods. Existing works primarily rely on external evaluators, focusing on training and prompting strategies. However, a crucial aspect, model-aware glass-box features, is overlooked. In this study, we explore the utility of glass-box features under the scenario of self-evaluation, namely applying an LLM to evaluate its own output. We investigate various glass-box feature groups and discovered that the softmax distribution serves as a reliable quality indicator for self-evaluation. Experimental results on public benchmarks validate the feasibility of self-evaluation of LLMs using glass-box features., Comment: accepted as Findings of EMNLP2024

Recently, there has been a growing trend of utilizing Large Language Model (LLM) to evaluate the quality of other LLMs. Many studies have employed proprietary close-sourced models, especially GPT-4, as the evaluator. Alternatively, other works have fine-tuned judge models based on open-source LLMs as the evaluator. While the fine-tuned judge models are claimed to achieve comparable evaluation capability with GPT-4, in this work, we conduct an empirical study of judge models. Our findings indicate that although the fine-tuned judge models achieve high performance on in-domain test sets, even surpassing GPT-4, they underperform GPT-4 across several dimensions, including generalizability, fairness, aspect-specific evaluation, and scalability. We also reveal that the fine-tuned judge model inherently operates as a task-specific classifier, consequently imposing the limitations. Finally, we introduce a integrated method, leveraging GPT-4 to compensate for the limitations and improve the fine-tuned judges. Experiment results show our method achieves accuracy on par with GPT-4 with only 50% of the API expense.

This paper addresses the complex issue of one-shot face stylization, focusing on the simultaneous consideration of appearance and structure, where previous methods have fallen short. We explore deformation-aware face stylization that diverges from traditional single-image style reference, opting for a real-style image pair instead. The cornerstone of our method is the utilization of a self-supervised vision transformer, specifically DINO-ViT, to establish a robust and consistent facial structure representation across both real and style domains. Our stylization process begins by adapting the StyleGAN generator to be deformation-aware through the integration of spatial transformers (STN). We then introduce two innovative constraints for generator fine-tuning under the guidance of DINO semantics: i) a directional deformation loss that regulates directional vectors in DINO space, and ii) a relative structural consistency constraint based on DINO token self-similarities, ensuring diverse generation. Additionally, style-mixing is employed to align the color generation with the reference, minimizing inconsistent correspondences. This framework delivers enhanced deformability for general one-shot face stylization, achieving notable efficiency with a fine-tuning duration of approximately 10 minutes. Extensive qualitative and quantitative comparisons demonstrate our superiority over state-of-the-art one-shot face stylization methods. Code is available at https://github.com/zichongc/DoesFS, Comment: Accepted to CVPR 2024. Project page: https://github.com/zichongc/DoesFS