Data plays a pivotal role in the groundbreaking advancements in artificial intelligence. The quantitative analysis of data significantly contributes to model training, enhancing both the efficiency and quality of data utilization. However, existing data analysis tools often lag in accuracy. For instance, many of these tools even assume that the loss function of neural networks is convex. These limitations make it challenging to implement current methods effectively. In this paper, we introduce a new formulation to approximate a sample's influence by accumulating the differences in influence between consecutive learning steps, which we term Diff-In. Specifically, we formulate the sample-wise influence as the cumulative sum of its changes/differences across successive training iterations. By employing second-order approximations, we approximate these difference terms with high accuracy while eliminating the need for model convexity required by existing methods. Despite being a second-order method, Diff-In maintains computational complexity comparable to that of first-order methods and remains scalable. This efficiency is achieved by computing the product of the Hessian and gradient, which can be efficiently approximated using finite differences of first-order gradients. We assess the approximation accuracy of Diff-In both theoretically and empirically. Our theoretical analysis demonstrates that Diff-In achieves significantly lower approximation error compared to existing influence estimators. Extensive experiments further confirm its superior performance across multiple benchmark datasets in three data-centric tasks: data cleaning, data deletion, and coreset selection. Notably, our experiments on data pruning for large-scale vision-language pre-training show that Diff-In can scale to millions of data points and outperforms strong baselines.

Reinforcement fine-tuning (RFT) is essential for enhancing the reasoning and generalization capabilities of large language models, but its success heavily relies on the quality of the training data. While data selection has been extensively studied in supervised learning, its role in reinforcement learning, particularly during the RFT stage, remains largely underexplored. In this work, we introduce RFT-Inf, the first influence estimator designed for data in reinforcement learning. RFT-Inf quantifies the importance of each training example by measuring how its removal affects the final training reward, offering a direct estimate of its contribution to model learning. To ensure scalability, we propose a first-order approximation of the RFT-Inf score by backtracking through the optimization process and applying temporal differentiation to the sample-wise influence term, along with a first-order Taylor approximation to adjacent time steps. This yields a lightweight, gradient-based estimator that evaluates the alignment between an individual sample’s gradient and the average gradient direction of all training samples, where a higher degree of alignment implies greater training utility. Extensive experiments demonstrate that RFT-Inf consistently improves reward performance and accelerates convergence in reinforcement fine-tuning.

We present a novel framework named PortraitNeRF to generate high-fidelity talking portrait videos for performing faithful identity-preserving reenactment of source videos. This is a challenging task because the generated results should be natural and match the speaker’s head movement, expression, eye blinks and speech audio. To acquire sufficient guidance from source video, the proposed PortraitNeRF exploits not only speech audio but also detailed motion information derived from visual data, including facial expressions, head pose and head position information. By adopting only a single neural radiance field, PortraitNeRF is able to generate complete and coordinated portrait video without bells and whistles. The completeness is ensured by the single-NeRF structure, and the superior head-torso coordination ability comes from using head pose and position information as its conditional input. Moreover, a simple yet effective mouth region emphasis strategy that fits well with the NeRF mechanism helps improving the accuracy of mouth shape. Experimental results and ablation studies demonstrate the superiority and effectiveness of PortraitNeRF.

In this paper, we present an implicit surface reconstruction method with 3D Gaussian Splatting (3DGS), namely 3DGSR, that allows for accurate 3D reconstruction with intricate details while inheriting the high efficiency and rendering quality of 3DGS. The key insight is to incorporate an implicit signed distance field (SDF) within 3D Gaussians for surface modeling, and to enable the alignment and joint optimization of both SDF and 3D Gaussians. To achieve this, we design coupling strategies that align and associate the SDF with 3D Gaussians, allowing for unified optimization and enforcing surface constraints on the 3D Gaussians. With alignment, optimizing the 3D Gaussians provides supervisory signals for SDF learning, enabling the reconstruction of intricate details. However, this only offers sparse supervisory signals to the SDF at locations occupied by Gaussians, which is insufficient for learning a continuous SDF. Then, to address this limitation, we incorporate volumetric rendering and align the rendered geometric attributes (depth, normal) with that derived from 3DGS. In sum, these two designs allow SDF and 3DGS to be aligned, jointly optimized, and mutually boosted. Our extensive experimental results demonstrate that our 3DGSR enables high-quality 3D surface reconstruction while preserving the efficiency and rendering quality of 3DGS. Besides, our method competes favorably with leading surface reconstruction techniques while offering a more efficient learning process and much better rendering qualities.

Synthesizing realistic videos according to a given speech is still an open challenge. Previous works have been plagued by issues such as inaccurate lip shape generation and poor image quality. The key reason is that only motions and appearances on limited facial areas (e.g., lip area) are mainly driven by the input speech. Therefore, directly learning a mapping function from speech to the entire head image is prone to ambiguity, particularly when using a short video for training. We thus propose a decomposition-synthesis-composition framework named Speech to Lip (Speech2Lip) that disentangles speech-sensitive and speech-insensitive motion/appearance to facilitate effective learning from limited training data, resulting in the generation of natural-looking videos. First, given a fixed head pose (i.e., canonical space), we present a speech-driven implicit model for lip image generation which concentrates on learning speech-sensitive motion and appearance. Next, to model the major speech-insensitive motion (i.e., head movement), we introduce a geometry-aware mutual explicit mapping (GAMEM) module that establishes geometric mappings between different head poses. This allows us to paste generated lip images at the canonical space onto head images with arbitrary poses and synthesize talking videos with natural head movements. In addition, a Blend-Net and a contrastive sync loss are introduced to enhance the overall synthesis performance. Quantitative and qualitative results on three benchmarks demonstrate that our model can be trained by a video of just a few minutes in length and achieve state-of-the-art performance in both visual quality and speechvisual synchronization. Code: https://github.com/CVMILab/Speech2Lip.

Existing methods for adapting Neural Radiance Fields (NeRFs) to scene changes require extensive data capture and model retraining, which is both time-consuming and labor-intensive. In this paper, we tackle the challenge of efficiently adapting NeRFs to real-world scene changes over time using a few new images while retaining the memory of unaltered areas, focusing on the continual learning aspect of NeRFs. To this end, we propose CL-NeRF, which consists of two key components: a lightweight expert adaptor for adapting to new changes and evolving scene representations and a conflict-aware knowledge distillation learning objective for memorizing unchanged parts. We also present a new benchmark for evaluating Continual Learning of NeRFs with comprehensive metrics. Our extensive experiments demonstrate that CL-NeRF can synthesize high-quality novel views of both changed and unchanged regions with high training efficiency, surpassing existing methods in terms of reducing forgetting and adapting to changes. Code and benchmark will be made available.

High Efficiency Video Coding (HEVC) is the most recent video coding standard aiming to further reduce the bitrate over 50% as compared to the state-of-the-art H.264/Advanced Video Coding under the same visual quality. In order to achieve this, a number of advanced coding techniques have been adopted in HEVC, including the quadtree structure of Coding Unit (CU), Prediction Unit (PU) and Transform Unit (TU), etc. However, these coding techniques lead to a tremendous increase in HEVC encoding computations. In order to reduce the HEVC encoding computational complexity, the optimal stopping theory is employed herein to design an efficient algorithm to optimize the decision making process when choosing the best coding parameters of CU, PU and TU. Extensive comparative experimental results are performed by the proposed algorithm and another two recent works, which demonstrate that the proposed algorithm is very efficient and better in reducing the HEVC encoding computations while keeping the video quality and compression efficiency almost intact.

The latest video coding standard high efficiency video coding (HEVC) has made a significant progress in compression efficiency than previous standard H.264/advanced video coding (AVC) while it has led to a tremendous increase in encoding computations. Recently, a Bayesian model based transform unit (TU) depth decision approach has been designed to accelerate TU depth decision, which requires numerous variance computations. In this work, a novel relevant feature based Bayesian model is proposed for fast TU depth decision. Experimental results demonstrate that the best performance is achieved while the depths of upper TU, left TU and co-located TU are all taken into considerations. Moreover, as compared with previous research, the proposed algorithm reduces much more encoding computations while keeping the video quality and compression efficiency more or less intact.