Xu Chen

Jie Chen, Zhipeng Chen, Jiapeng Wang et al.

Continual pre-training (CPT) has been an important approach for adapting language models to specific domains or tasks. To make the CPT approach more traceable, this paper presents a technical report for continually pre-training Llama-3 (8B), which significantly enhances the Chinese language ability and scientific reasoning ability of the backbone model. To enhance the new abilities while retaining the original abilities, we design specific data mixture and curriculum strategies by utilizing existing datasets and synthesizing high-quality datasets. Specifically, we synthesize multidisciplinary scientific question and answer (QA) pairs based on related web pages, and subsequently incorporate these synthetic data to improve the scientific reasoning ability of Llama-3. We refer to the model after CPT as Llama-3-SynE (Synthetic data Enhanced Llama-3). We also present the tuning experiments with a relatively small model -- TinyLlama, and employ the derived findings to train the backbone model. Extensive experiments on a number of evaluation benchmarks show that our approach can largely improve the performance of the backbone models, including both the general abilities (+8.81 on C-Eval and +6.31 on CMMLU) and the scientific reasoning abilities (+12.00 on MATH and +4.13 on SciEval), without hurting the original capacities. Our model, data, and codes are available at https://github.com/RUC-GSAI/Llama-3-SynE.

Yunkui Pang, Yilin Liu, Xu Chen et al.

Cone Beam Computed Tomography (CBCT) finds diverse applications in medicine. Ensuring high image quality in CBCT scans is essential for accurate diagnosis and treatment delivery. Yet, the susceptibility of CBCT images to noise and artifacts undermines both their usefulness and reliability. Existing methods typically address CBCT artifacts through image-to-image translation approaches. These methods, however, are limited by the artifact types present in the training data, which may not cover the complete spectrum of CBCT degradations stemming from variations in imaging protocols. Gathering additional data to encompass all possible scenarios can often pose a challenge. To address this, we present SinoSynth, a physics-based degradation model that simulates various CBCT-specific artifacts to generate a diverse set of synthetic CBCT images from high-quality CT images without requiring pre-aligned data. Through extensive experiments, we demonstrate that several different generative networks trained on our synthesized data achieve remarkable results on heterogeneous multi-institutional datasets, outperforming even the same networks trained on actual data. We further show that our degradation model conveniently provides an avenue to enforce anatomical constraints in conditional generative models, yielding high-quality and structure-preserving synthetic CT images.

Ilya A. Petrov, Vladimir Guzov, Riccardo Marin et al.

Modeling human-object interactions (HOI) from an egocentric perspective is a largely unexplored yet important problem due to the increasing adoption of wearable devices, such as smart glasses and watches. We investigate how much information about interaction can be recovered from only head and wrists tracking. Our answer is ECHO (Ego-Centric modeling of Human-Object interactions), which, for the first time, proposes a unified framework to recover three modalities: human pose, object motion, and contact from such minimal observation. ECHO employs a Diffusion Transformer architecture and a unique three-variate diffusion process, which jointly models human motion, object trajectory, and contact sequence, allowing for flexible input configurations. Our method operates in a head-centric canonical space, enhancing robustness to global orientation. We propose a conveyor-based inference, which progressively increases the diffusion timestamp with the frame position, allowing us to process sequences of any length. Through extensive evaluation, we demonstrate that ECHO outperforms existing methods that do not offer the same flexibility, setting a state-of-the-art in egocentric HOI reconstruction.

Zijian Dong, Xu Chen, Jinlong Yang et al.

While progress in 2D generative models of human appearance has been rapid, many applications require 3D avatars that can be animated and rendered. Unfortunately, most existing methods for learning generative models of 3D humans with diverse shape and appearance require 3D training data, which is limited and expensive to acquire. The key to progress is hence to learn generative models of 3D avatars from abundant unstructured 2D image collections. However, learning realistic and complete 3D appearance and geometry in this under-constrained setting remains challenging, especially in the presence of loose clothing such as dresses. In this paper, we propose a new adversarial generative model of realistic 3D people from 2D images. Our method captures shape and deformation of the body and loose clothing by adopting a holistic 3D generator and integrating an efficient and flexible articulation module. To improve realism, we train our model using multiple discriminators while also integrating geometric cues in the form of predicted 2D normal maps. We experimentally find that our method outperforms previous 3D- and articulation-aware methods in terms of geometry and appearance. We validate the effectiveness of our model and the importance of each component via systematic ablation studies.

Hsuan-I Ho, Xu Chen, Jie Song et al.

Upsampling videos of human activity is an interesting yet challenging task with many potential applications ranging from gaming to entertainment and sports broadcasting. The main difficulty in synthesizing video frames in this setting stems from the highly complex and non-linear nature of human motion and the complex appearance and texture of the body. We propose to address these issues in a motion-guided frame-upsampling framework that is capable of producing realistic human motion and appearance. A novel motion model is trained to inference the non-linear skeletal motion between frames by leveraging a large-scale motion-capture dataset (AMASS). The high-frame-rate pose predictions are then used by a neural rendering pipeline to produce the full-frame output, taking the pose and background consistency into consideration. Our pipeline only requires low-frame-rate videos and unpaired human motion data but does not require high-frame-rate videos for training. Furthermore, we contribute the first evaluation dataset that consists of high-quality and high-frame-rate videos of human activities for this task. Compared with state-of-the-art video interpolation techniques, our method produces in-between frames with better quality and accuracy, which is evident by state-of-the-art results on pixel-level, distributional metrics and comparative user evaluations. Our code and the collected dataset are available at https://git.io/Render-In-Between.

Xu Chen, Yufeng Zheng, Michael J. Black et al.

Neural implicit surface representations have emerged as a promising paradigm to capture 3D shapes in a continuous and resolution-independent manner. However, adapting them to articulated shapes is non-trivial. Existing approaches learn a backward warp field that maps deformed to canonical points. However, this is problematic since the backward warp field is pose dependent and thus requires large amounts of data to learn. To address this, we introduce SNARF, which combines the advantages of linear blend skinning (LBS) for polygonal meshes with those of neural implicit surfaces by learning a forward deformation field without direct supervision. This deformation field is defined in canonical, pose-independent space, allowing for generalization to unseen poses. Learning the deformation field from posed meshes alone is challenging since the correspondences of deformed points are defined implicitly and may not be unique under changes of topology. We propose a forward skinning model that finds all canonical correspondences of any deformed point using iterative root finding. We derive analytical gradients via implicit differentiation, enabling end-to-end training from 3D meshes with bone transformations. Compared to state-of-the-art neural implicit representations, our approach generalizes better to unseen poses while preserving accuracy. We demonstrate our method in challenging scenarios on (clothed) 3D humans in diverse and unseen poses.

Xu Chen, Jie Song, Otmar Hilliges

We present an approach that learns to synthesize high-quality, novel views of 3D objects or scenes, while providing fine-grained and precise control over the 6-DOF viewpoint. The approach is self-supervised and only requires 2D images and associated view transforms for training. Our main contribution is a network architecture that leverages a transforming auto-encoder in combination with a depth-guided warping procedure to predict geometrically accurate unseen views. Leveraging geometric constraints renders direct supervision via depth or flow maps unnecessary. If large parts of the object are occluded in the source view, a purely learning based prior is used to predict the values for dis-occluded pixels. Our network furthermore predicts a per-pixel mask, used to fuse depth-guided and pixel-based predictions. The resulting images reflect the desired 6-DOF transformation and details are preserved. We thoroughly evaluate our architecture on synthetic and real scenes and under fine-grained and fixed-view settings. Finally, we demonstrate that the approach generalizes to entirely unseen images such as product images downloaded from the internet.

Xu Chen, Saratendu Sethi

The classification of electrocardiographic (ECG) signals is a challenging problem for healthcare industry. Traditional supervised learning methods require a large number of labeled data which is usually expensive and difficult to obtain for ECG signals. Active learning is well-suited for ECG signal classification as it aims at selecting the best set of labeled data in order to maximize the classification performance. Motivated by the fact that ECG data are usually heavily unbalanced among different classes and the class labels are noisy as they are manually labeled, this paper proposes a novel solution based on robust active learning for addressing these challenges. The key idea is to first apply the clustering of the data in a low dimensional embedded space and then select the most information instances within local clusters. By selecting the most informative instances relying on local average minimal distances, the algorithm tends to select the data for labelling in a more diversified way. Finally, the robustness of the model is further enhanced by adding a novel noisy label reduction scheme after the selection of the labeled data. Experiments on the ECG signal classification from the MIT-BIH arrhythmia database demonstrate the effectiveness of the proposed algorithm.

Yun Wang, Xu Chen, Peter J. Ramadge

One way to solve lasso problems when the dictionary does not fit into available memory is to first screen the dictionary to remove unneeded features. Prior research has shown that sequential screening methods offer the greatest promise in this endeavor. Most existing work on sequential screening targets the context of tuning parameter selection, where one screens and solves a sequence of $N$ lasso problems with a fixed grid of geometrically spaced regularization parameters. In contrast, we focus on the scenario where a target regularization parameter has already been chosen via cross-validated model selection, and we then need to solve many lasso instances using this fixed value. In this context, we propose and explore a feedback controlled sequential screening scheme. Feedback is used at each iteration to select the next problem to be solved. This allows the sequence of problems to be adapted to the instance presented and the number of intermediate problems to be automatically selected. We demonstrate our feedback scheme using several datasets including a dictionary of approximate size 100,000 by 300,000.

Chen Xu, Yongquan Zhang, Runze Li

In modern scientific research, massive datasets with huge numbers of observations are frequently encountered. To facilitate the computational process, a divide-and-conquer scheme is often used for the analysis of big data. In such a strategy, a full dataset is first split into several manageable segments; the final output is then averaged from the individual outputs of the segments. Despite its popularity in practice, it remains largely unknown that whether such a distributive strategy provides valid theoretical inferences to the original data. In this paper, we address this fundamental issue for the distributed kernel regression (DKR), where the algorithmic feasibility is measured by the generalization performance of the resulting estimator. To justify DKR, a uniform convergence rate is needed for bounding the generalization error over the individual outputs, which brings new and challenging issues in the big data setup. Under mild conditions, we show that, with a proper number of segments, DKR leads to an estimator that is generalization consistent to the unknown regression function. The obtained results justify the method of DKR and shed light on the feasibility of using other distributed algorithms for processing big data. The promising preference of the method is supported by both simulation and real data examples.

Shaobo Lin, Chen Xu, Jingshan Zeng et al.

$l^q$-regularization has been demonstrated to be an attractive technique in machine learning and statistical modeling. It attempts to improve the generalization (prediction) capability of a machine (model) through appropriately shrinking its coefficients. The shape of a $l^q$ estimator differs in varying choices of the regularization order $q$. In particular, $l^1$ leads to the LASSO estimate, while $l^{2}$ corresponds to the smooth ridge regression. This makes the order $q$ a potential tuning parameter in applications. To facilitate the use of $l^{q}$-regularization, we intend to seek for a modeling strategy where an elaborative selection on $q$ is avoidable. In this spirit, we place our investigation within a general framework of $l^{q}$-regularized kernel learning under a sample dependent hypothesis space (SDHS). For a designated class of kernel functions, we show that all $l^{q}$ estimators for $0< q < \infty$ attain similar generalization error bounds. These estimated bounds are almost optimal in the sense that up to a logarithmic factor, the upper and lower bounds are asymptotically identical. This finding tentatively reveals that, in some modeling contexts, the choice of $q$ might not have a strong impact in terms of the generalization capability. From this perspective, $q$ can be arbitrarily specified, or specified merely by other no generalization criteria like smoothness, computational complexity, sparsity, etc..