论文检索

Reinforcement Learning · Deep RL

Dohyeong Kim, Mineui Hong, Jeongho Park, Songhwai Oh

In real-world applications, a reinforcement learning (RL) agent should consider multiple objectives and adhere to safety guidelines.To address these considerations, we propose a constrained multi-objective RL algorithm named constrained multi-objective gradient aggregator (CoMOGA).In the field of multi-objective optimization, managing conflicts between the gradients of the multiple objectives is crucial to prevent policies from converging to local optima.It is also essential to efficiently handle safety constraints for stable training and constraint satisfaction.We address these challenges straightforwardly by treating the maximization of multiple objectives as a constrained optimization problem (COP), where the constraints are defined to improve the original objectives.Existing safety constraints are then integrated into the COP, and the policy is updated by solving the COP, which ensures the avoidance of gradient conflicts.Despite its simplicity, CoMOGA guarantees convergence to global optima in a tabular setting.Through various experiments, we have confirmed that preventing gradient conflicts is critical, and the proposed method achieves constraint satisfaction across all tasks.

Optimization · Non-Convex

Ryan D'Orazio, Danilo Vucetic, Zichu Liu, Junhyung Lyle Kim, Ioannis Mitliagkas, Gauthier Gidel

Deep learning has proven to be effective in a wide variety of loss minimization problems.However, many applications of interest, like minimizing projected Bellman error and min-max optimization, cannot be modelled as minimizing a scalar loss function but instead correspond to solving a variational inequality (VI) problem.This difference in setting has caused many practical challenges as naive gradient-based approaches from supervised learning tend to diverge and cycle in the VI case.In this work, we propose a principled surrogate-based approach compatible with deep learning to solve VIs.We show that our surrogate-based approach has three main benefits: (1) under assumptions that are realistic in practice (when hidden monotone structure is present, interpolation, and sufficient optimization of the surrogates), it guarantees convergence, (2) it provides a unifying perspective of existing methods, and (3) is amenable to existing deep learning optimizers like ADAM.Experimentally, we demonstrate our surrogate-based approach is effective in min-max optimization and minimizing projected Bellman error. Furthermore, in the deep reinforcement learning case, we propose a novel variant of TD(0) which is more compute and sample efficient.

Reinforcement Learning · Planning

Navdeep Kumar, Yashaswini Murthy, Itai Shufaro, Kfir Y Levy, R. Srikant, Shie Mannor

We present the first comprehensive finite-time global convergence analysis of policy gradient for infinite horizon average reward Markov decision processes (MDPs). Specifically, we focus on ergodic tabular MDPs with finite state and action spaces. Our analysis shows that the policy gradient iterates converge to the optimal policy at a sublinear rate of $O(\frac{1}{T})$, where $T$ represents the number of iterations. Performance bounds for discounted reward MDPs cannot be easily extended to average reward MDPs as the bounds grow proportional to the fifth power of the effective horizon. Recent work on such extensions makes a smoothness assumption that has not been verified. Thus, our primary contribution is in providing the first complete proof that the policy gradient algorithm converges globally for average-reward MDPs, without such an assumption. We also obtain the corresponding finite-time performance guarantees. In contrast to the existing discounted reward performance bounds, our performance bounds have an explicit dependence on constants that capture the complexity of the underlying MDP. Motivated by this observation, we reexamine and improve the existing performance bounds for discounted reward MDPs. We also present simulations that empirically validate the result.

Optimization · Non-Convex

Bingcong Li, Liang Zhang, Aryan Mokhtari, Niao He

This work revisits the classical low-rank matrix factorization problem and unveils the critical role of initialization in shaping convergence rates for such nonconvex and nonsmooth optimization. We introduce Nystrom initialization, which significantly improves the global convergence of Scaled Gradient Descent (ScaledGD) in both symmetric and asymmetric matrix factorization tasks. Specifically, we prove that ScaledGD with Nystrom initialization achieves quadratic convergence in cases where only linear rates were previously known. Furthermore, we extend this initialization to low-rank adapters (LoRA) commonly used for finetuning foundation models. Our approach, NoRA, i.e., LoRA with Nystrom initialization, demonstrates superior performance across various downstream tasks and model scales, from 1B to 7B parameters, in large language and diffusion models.

Ryan Lucas, Rahul Mazumder

We present SNOWS, a one-shot post-training pruning framework aimed at reducing the cost of vision network inference without retraining. Current leading one-shot pruning methods minimize layer-wise least squares reconstruction error which does not take into account deeper network representations. We propose to optimize a more global reconstruction objective. This objective accounts for nonlinear activations deep in the network to obtain a better proxy for the network loss. This nonlinear objective leads to a more challenging optimization problem---we demonstrate it can be solved efficiently using a specialized second-order optimization framework. A key innovation of our framework is the use of Hessian-free optimization to compute exact Newton descent steps without needing to compute or store the full Hessian matrix. A distinct advantage of SNOWS is that it can be readily applied on top of any sparse mask derived from prior methods, readjusting their weights to preserve deep feature representations. SNOWS obtains state-of-the-art results on various one-shot pruning benchmarks including residual networks and Vision Transformers (ViT/B-16 and ViT/L-16, 86m and 304m parameters respectively). Our open-source implementation is available at https://github.com/mazumder-lab/SNOWS.

Applications · Computer Vision

Qihang Zhang, Yinghao Xu, Chaoyang Wang, Hsin-Ying Lee, Gordon Wetzstein, Bolei Zhou, Ceyuan Yang

Scene image editing is crucial for entertainment, photography, and advertising design. Existing methods solely focus on either 2D individual object or 3D global scene editing. This results in a lack of a unified approach to effectively control and manipulate scenes at the 3D level with different levels of granularity. In this work, we propose 3DitScene, a novel and unified scene editing framework leveraging language-guided disentangled Gaussian Splatting that enables seamless editing from 2D to 3D, allowing precise control over scene composition and individual objects. We first incorporate 3D Gaussians that are refined through generative priors and optimization techniques. Language features from CLIP then introduce semantics into 3D geometry for object disentanglement. With the disentangled Gaussians, 3DitScene allows for manipulation at both the global and individual levels, revolutionizing creative expression and empowering control over scenes and objects. Experimental results demonstrate the effectiveness and versatility of 3DitScene in scene image editing.

Yingying Ren 0001, Julian Panetta, Seiichi Suzuki, Uday Kusupati, Florin Isvoranu, Mark Pauly

Surface-based inflatables are composed of two thin layers of nearly inextensible sheet material joined together along carefully selected fusing curves. During inflation, pressure forces separate the two sheets to maximize the enclosed volume. The fusing curves restrict this expansion, leading to a spatially varying in-plane contraction and hence metric frustration. The inflated structure settles into a 3D equilibrium that balances pressure forces with the internal elastic forces of the sheets. We present a computational framework for analyzing and designing surface-based inflatable structures with arbitrary fusing patterns. Our approach employs numerical homogenization to characterize the behavior of parametric families of periodic inflatable patch geometries, which can then be combined to tessellate the sheet with smoothly varying patterns. We propose a novel parametrization of the underlying deformation space that allows accurate, efficient, and systematical analysis of the stretching and bending behavior of inflated patches with potentially open boundaries. We apply our homogenization algorithm to create a database of geometrically diverse fusing patterns spanning a wide range of material properties and deformation characteristics. This database is employed in an inverse design algorithm that solves for fusing curves to best approximate a given input target surface. Local patches are selected and blended to form a global network of curves based on a geometric flattening algorithm. These fusing curves are then further optimized to minimize the distance of the deployed structure to target surface. We show that this approach offers greater flexibility to approximate given target geometries compared to previous work while significantly improving structural performance.

Jingyu Hu 0001, Ka-Hei Hui, Zhengzhe Liu, Hao Zhang 0002, Chi-Wing Fu

This paper introduces a new approach based on a coupled representation and a neural volume optimization to implicitly perform 3D shape editing in latent space. This work has three innovations. First, we design the coupled neural shape (CNS) representation for supporting 3D shape editing. This representation includes a latent code, which captures high-level global semantics of the shape, and a 3D neural feature volume, which provides a spatial context to associate with the local shape changes given by the editing. Second, we formulate the coupled neural shape optimization procedure to co-optimize the two coupled components in the representation subject to the editing operation. Last, we offer various 3D shape editing operators, i.e., copy, resize, delete, and drag, and derive each into an objective for guiding the CNS optimization, such that we can iteratively co-optimize the latent code and neural feature volume to match the editing target. With our approach, we can achieve a rich variety of editing results that are not only aware of the shape semantics but are also not easy to achieve by existing approaches. Both quantitative and qualitative evaluations demonstrate the strong capabilities of our approach over the state-of-the-art solutions.

Haowen Sun 0001, Ruikun Zheng, Haibin Huang, Chongyang Ma, Hui Huang 0004, Ruizhen Hu

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

Weizhou Liu, Xingce Wang, Haichuan Zhao, Xingfei Xue, Zhongke Wu, Xuequan Lu, Ying He 0001

This paper introduces a new approach for generating globally consistent normals for point clouds sampled from manifold surfaces. Given that the generalized winding number (GWN) field generated by a point cloud with globally consistent normals is a solution to a PDE with jump boundary conditions and possesses harmonic properties, and the Dirichlet energy of the GWN field can be defined as an integral over the boundary surface, we formulate a boundary energy derived from the Dirichlet energy of the GWN. Taking as input a point cloud with randomly oriented normals, we optimize this energy to restore the global harmonicity of the GWN field, thereby recovering the globally consistent normals. Experiments show that our method outperforms state-of-the-art approaches, exhibiting enhanced robustness to noise, outliers, complex topologies, and thin structures. Our code can be found at https://github.com/liuweizhou319/BIM.

Nicholas Milef, John Keyser, Shu Kong

Optical motion capture (mocap) requires accurately reconstructing the human body from retroreflective markers, including pose and shape. In a typical mocap setting, marker labeling is an important but tedious and error-prone step. Previous work has shown that marker labeling can be automated by using a structured template defining specific marker placements, but this places additional recording constraints. We propose to relax these constraints and solve for Unstructured Unlabeled Optical (UUO) mocap. Compared to the typical mocap setting that either labels markers or places them w.r.t a structured layout, markers in UUO mocap can be placed anywhere on the body and even on one specific limb (e.g., right leg for biomechanics research), hence it is of more practical significance. It is also more challenging. To solve UUO mocap, we exploit a monocular video captured by a single RGB camera, which does not require camera calibration. On this video, we run an off-the-shelf method to reconstruct and track a human individual, giving strong visual priors of human body pose and shape. With both the video and UUO markers, we propose an optimization pipeline towards marker identification, marker labeling, human pose estimation, and human body reconstruction. Our technical novelties include multiple hypothesis testing to optimize global orientation, and marker localization and marker-part matching to better optimize for body surface. We conduct extensive experiments to quantitatively compare our method against state-of-the-art approaches, including marker-only mocap and video-only human body/shape reconstruction. Experiments demonstrate that our method resoundingly outperforms existing methods on three established benchmark datasets for both full-body and partial-body reconstruction.

Seungjae Lee 0004, Seung-Woo Nam 0002, Kevin W. Rio, Renate Landig, Hsien-Hui Cheng, Lu Lu, Barry Silverstein

Steered retinal projection (SRP) is an emerging display technology that combines retinal projection and pupil steering to achieve exceptional light efficiency and a consistent viewing experience. Retinal projection enables most photons from a display projector to reach the retina, and pupil steering dynamically aligns the narrow viewing window of the retinal projection with the eye. While SRP holds considerable promise, its development has been stagnant due to a lack of understanding how human vision reacts to the dynamic steering movement of the viewing window. To delve into these areas, this study introduces the first SRP system testbed specifically designed for perceptual studies on the viewing experience of pupil steering. The testbed replicates the SRP viewing experience and offers the flexibility in adjusting several parameters including steering resolution, accuracy, and latency. We conducted two perceptual studies utilizing the testbed. The first study investigates the impact of saccadic suppression, a phenomenon that reduces visual sensitivity during rapid eye movements, on the SRP viewing experience. The second study explores the trade space between eye-tracking and pupil steering performance, providing insights into the optimal balance between these factors. Additionally, we introduce a numerical model to predict the detection probability for SRP artifacts considering the temporal characteristics of global luminance and the human vision system. This model enables a more comprehensive interpretation of user study and provides preliminary hardware requirements for SRP systems. The findings from this study offer invaluable research directions that may help determine component-level development milestones for SRP development, paving the way for the practical implementation of this promising technology.

Rui Li, Dong Liu

Recent studies on motion estimation have advocated an optimized motion representation that is globally consistent across the entire video, preferably for every pixel. This is challenging as a uniform representation may not account for the complex and diverse motion and appearance of natural videos. We address this problem and propose a new test-time optimization method, named DecoMotion, for estimating per-pixel and long-range motion. DecoMotion explicitly decomposes video content into static scenes and dynamic objects, either of which uses a quasi-3D canonical volume to represent. DecoMotion separately coordinates the transformations between local and canonical spaces, facilitating an affine transformation for the static scene that corresponds to camera motion. For the dynamic volume, DecoMotion leverages discriminative and temporally consistent features to rectify the non-rigid transformation. The two volumes are finally fused to fully represent motion and appearance. This divide-and-conquer strategy leads to more robust tracking through occlusions and deformations and meanwhile obtains decomposed appearances. We conduct evaluations on the TAP-Vid benchmark. The results demonstrate our method boosts the point-tracking accuracy by a large margin and performs on par with some state-of-the-art dedicated point-tracking solutions. Code and models will be available upon acceptance of the paper.

Fei Wang

3D reconstruction from a sketch offers an efficient means of boosting the productivity of 3D modeling. However, such a task remains largely under-explored due to the difficulties caused by the inherent abstractive representation and diversity of sketches. In this paper, we introduce a novel deep neural network model, Sketch2Vox, for 3D reconstruction from a single monocular sketch. Taking a sketch as input, the proposed model first converts it into two different representations, i.e., a binary image and a 2D point cloud. Second, we extract semantic features from them using two newly-developed processing modules, including the SktConv module designed for hierarchical abstract features learning from the binary image and the SktMPFM designed for local and global context feature extraction from the 2D point cloud. Prior to feeding features into the 3D-decoder-refiner module for fine-grained reconstruction, the resultant image-based and point-based feature maps are fused together according to their internal correlation using the proposed cross-modal fusion attention module. Finally, we use an optimization module to refine the details of the generated 3D model. To evaluate the efficiency of our method, we collect a large dataset consisting of more than 12,000 Sketch-Voxel pairs and compare the proposed Sketch2Vox against several state-of-the-art methods. The experimental results demonstrate the proposed method is superior to peer ones with regard to reconstruction quality. The dataset is publicly available on https://drive.google.com/file/d/1aXug8PcLnWaDZiWZrcmhvVNFC4n_eAih/view?usp=sharing.

Kihwan Yoon, Yong Han Kim, Sungjei Kim, Jinwoo Jeong

Within the video, different regions have varying motion complexity, with simple regions containing static or global motion and complex regions containing fast motion or lots of local motion. In recent years, the performance of flow-based Video Frame Interpolation (VFI) algorithms has improved significantly. However, existing training methods train on randomly cropped regions of train data without considering the complexity of the motion. As a result, they cannot handle all regions of the frame that contain varying motion complexity. To solve this problem, we propose a novel VFI approach (IAM-VFI) that can interpolate any motion by considering the motion complexity of all regions in the frame. First, we propose a training data classification method for motion optimization based on each motion complexity. Then, using the proposed data, a flow estimation network generates optimized results for each complexity. Finally, we propose a Motion Complexity Estimation Network (MCENet) to generate a Motion Complexity Map (MCM) that can estimate the motion complexity of each region. Our proposed methods can be easily applied to most flow-based VFI algorithms. Experimental results show that the proposed method can interpolate any motion and significantly improve the performance of existing VFI algorithms.

Juncheng Ma, Peiwen Sun, Yaoting Wang, Di Hu

Audio-Visual Segmentation (AVS) aims to achieve pixel-level localization of sound sources in videos, while Audio-Visual Semantic Segmentation (AVSS), as an extension of AVS, further pursues semantic understanding of audio-visual scenes. However, since the AVSS task requires the establishment of audio-visual correspondence and semantic understanding simultaneously, we observe that previous methods have struggled to handle this mashup of objectives in end-to-end training, resulting in insufficient learning and sub-optimization. Therefore, we propose a two-stage training strategy called Stepping Stones, which decomposes the AVSS task into two simple subtasks from localization to semantic understanding, which are fully optimized in each stage to achieve step-by-step global optimization. This training strategy has also proved its generalization and effectiveness on existing methods. To further improve the performance of AVS tasks, we propose a novel framework Adaptive Audio Visual Segmentation, in which we incorporate an adaptive audio query generator and integrate masked attention into the transformer decoder, facilitating the adaptive fusion of visual and audio features. Extensive experiments demonstrate that our methods achieve state-of-the-art results on all three AVS benchmarks. The code will be released soon.

Rolandos Alexandros Potamias, Michael Tarasiou, Stylianos Ploumpis, Stefanos Zafeiriou

In the realm of 3D computer vision, parametric models have emerged as a ground-breaking methodology for the creation of realistic and expressive 3D avatars. Traditionally, they rely on Principal Component Analysis (PCA), given its ability to decompose data to an orthonormal space that maximally captures shape variations. However, due to the orthogonality constraints and the global nature of PCA's decomposition, these models struggle to perform localized and disentangled editing of 3D shapes, which severely affects their use in applications requiring fine control such as face sculpting. In this paper, we leverage diffusion models to enable diverse and fully localized edits on 3D meshes, while completely preserving the un-edited regions. We propose an effective diffusion masking training strategy that, by design, facilitates localized manipulation of any shape region, without being limited to predefined regions or to sparse sets of predefined control vertices. Following our framework, a user can explicitly set their manipulation region of choice and define an arbitrary set of vertices as handles to edit a 3D mesh. Compared to the current state-of-the-art our method leads to more interpretable shape manipulations than methods relying on latent code state, greater localization and generation diversity while offering faster inference than optimization based approaches. Code has been included in the supplementary material.

Ning Gao, xingyu jiang, Xiuhui Zhang, Yue Deng

Most current single image-deraining (SID) methods are based on the Transformer, which brings global modeling capabilities and is critical for high-quality reconstruction. However, their architectures only consider constructing long-range dependencies from the spatial domain, which suffers from a significant computational burden to keep effectiveness. Besides, these methods either overlook negative sample information in the optimization pipeline or underutilize the rain streak characteristics present in the negative ones. To tackle these problems, we propose a Frequency-Aware Deraining Transformer Framework (FADformer) that fully captures frequency domain features to achieve efficient rain removal. Specifically, we first construct the FADBlock, which comprises the Fused Fourier Convolution Mixer (FFCM) and Prior-Gated Feed-forward Network (PGFN). Unlike self-attention mechanisms, the FFCM exclusively conducts convolution operations in both spatial and frequency domains, endowing it with local-global capturing capabilities and efficiency. Simultaneously, the PGFN introduces residue channel prior in a gate-controlled manner to both enhance local details and retain the structure of features. Furthermore, we introduce a Frequency-domain Contrastive Regularization (FCR) during the training phase. The FCR facilitates contrastive learning in the frequency domain, enhancing the contribution of rain streak patterns in negative samples to improve performance. Extensive experiments on synthetic and real-world datasets show that the proposed method significantly outperforms the state-of-the-art approaches. We will release the code soon after the paper is accepted.

Gyeongrok Oh, Jaehwan Jeong, Sieun Kim, Wonmin Byeon, Jinkyu Kim, Sungwoong Kim, Sangpil Kim

We introduce a novel diffusion-based video generation method, generating a video showing multiple events given multiple individual sentences from the user. Our method does not require a large-scale video dataset since our method uses a pre-trained diffusion-based text-to-video generative model without a fine-tuning process. Specifically, we propose a last frame-aware diffusion process to preserve visual coherence between consecutive videos where each video consists of different events by initializing the latent and simultaneously adjusting noise in the latent to enhance the motion dynamic in a generated video. Furthermore, we find that the iterative update of latent vectors by referring to all the preceding frames maintains the global appearance across the frames in a video clip. To handle dynamic text input for video generation, we utilize a novel prompt generator that transfers course text messages from the user into the multiple optimal prompts for the text-to-video diffusion model. Extensive experiments and user studies show that our proposed method is superior to other video-generative models in terms of temporal coherency of content and semantics.

Junpeng Jing, Ye Mao, Krystian Mikolajczyk

Dynamic stereo matching is the task of estimating consistent disparities from stereo videos with dynamic objects. Recent learning-based methods prioritize optimal performance on a single stereo pair, resulting in temporal inconsistencies. Existing video methods apply per-frame matching and window-based cost aggregation across the time dimension, leading to low-frequency oscillations at the scale of the window size. Towards this challenge, we develop a bidirectional alignment mechanism for adjacent frames as a fundamental operation. We further propose a novel framework, BiDAStereo, that achieves consistent dynamic stereo matching. Unlike the existing methods, we model this task as local matching and global aggregation. Locally, we consider correlation in a triple-frame manner to pool information from adjacent frames and improve the temporal consistency. Globally, to exploit the entire sequence's consistency and extract dynamic scene cues for aggregation, we develop a motion-propagation recurrent unit. Extensive experiments demonstrate the performance of our method, showcasing improvements in prediction quality and achieving state-of-the-art results on various commonly used benchmarks. The code is available in supplemental material.