Weicheng Wang

Zhicheng Zhang, Weicheng Wang, Yongjie Zhu et al.

Understanding and predicting emotion from videos has gathered significant attention in recent studies, driven by advancements in video large language models (VideoLLMs). While advanced methods have made progress in video emotion analysis, the intrinsic nature of emotions poses significant challenges. Emotions are characterized by dynamic and cues-dependent properties, making it difficult to understand complex and evolving emotional states with reasonable rationale. To tackle these challenges, we propose a novel affective cues-guided reasoning framework that unifies fundamental attribute perception, expression analysis, and high-level emotional understanding in a stage-wise manner. At the core of our approach is a family of video emotion foundation models (VidEmo), specifically designed for emotion reasoning and instruction-following. These models undergo a two-stage tuning process: first, curriculum emotion learning for injecting emotion knowledge, followed by affective-tree reinforcement learning for emotion reasoning. Moreover, we establish a foundational data infrastructure and introduce a emotion-centric fine-grained dataset (Emo-CFG) consisting of 2.1M diverse instruction-based samples. Emo-CFG includes explainable emotional question-answering, fine-grained captions, and associated rationales, providing essential resources for advancing emotion understanding tasks. Experimental results demonstrate that our approach achieves competitive performance, setting a new milestone across 15 face perception tasks.

Weicheng Wang, Zhicheng Zhang, Zhongqi Zhang et al.

Video editing aims to modify input videos according to user intent. Recently, end-to-end training methods have garnered widespread attention, constructing paired video editing data through video generation or editing models. However, compared to image editing, the high annotation costs of video data severely constrain the scale, quality, and task diversity of video editing datasets when relying on video generative models or manual annotation. To bridge this gap, we propose LIVE, a joint training framework that leverages large-scale, high-quality image editing data alongside video datasets to bolster editing capabilities. To mitigate the domain discrepancy between static images and dynamic videos, we introduce a frame-wise token noise strategy, which treats the latents of specific frames as reasoning tokens, leveraging large pretrained video generative models to create plausible temporal transformations. Moreover, through cleaning public datasets and constructing an automated data pipeline, we adopt a two-stage training strategy to anneal video editing capabilities. Furthermore, we curate a comprehensive evaluation benchmark encompassing over 60 challenging tasks that are prevalent in image editing but scarce in existing video datasets. Extensive comparative and ablation experiments demonstrate that our method achieves state-of-the-art performance. The source code will be publicly available.

Qi Cheng, Yang Sun, Weidong Yu et al.

The query optimizer in a Database Management Systems (DBMS), translates declarative queries into efficient execution plans. Conventional bottom-up optimization consists of two main stages: Query Rewrite (QRW) and Cost-Based Optimization (CBO). However, applying a rewrite rule during QRW may not always be beneficial; the best choice may depend on the (estimated) execution cost of the original and rewritten expressions. Fully exploiting such cost-dependent rules necessitates interleaving QRW with frequent CBO invocations, thereby incurring substantial overhead and often impractical optimization times. To mitigate this inefficiency, we introduce a novel cost-based rewrite framework for bottom-up optimizers. The core of our approach is a multi-level caching mechanism for intermediate CBO results aimed at eliminating redundant computation. Furthermore, we establish and exploit upper cost bounds to intelligently prune the search space during optimization. We also contribute methodological solutions for caching and reusing intermediate plan results within a bottom-up optimizer architecture. The framework has been implemented in the GaussDB optimizer. Experiments show that it significantly reduces overall optimization time, demonstrating the effectiveness of our approach.

Zhicheng Zhang, Wentao Gu, Weicheng Wang et al.

Omnimodal understanding entails a massive, highly redundant search space of cross-modal interactions, demanding focused and deliberative reasoning. Current reasoning paradigms rely on either sequential step-by-step generation or parallel sample-by-sample rollouts, leading to isolated reasoning trajectories. This inability to share promising intermediate paths severely limits exploration efficiency and causes compounding errors in complex audio-visual tasks. To break this bottleneck, we introduce Omni-o3, a novel framework driven by a deep nested deduction policy. By formulating reasoning as a dynamic recursive search, Omni-o3 inherently shares reasoning prefixes across branches, enabling the iterative execution of four atomic cognitive actions: expansion, selection, simulation, and backpropagation. To empower this framework, we propose a robust two-stage training paradigm: (1) cold-start supervised fine-tuning on 101K high-quality, long-chain trajectories distilled from 3.5M diverse omnimodal samples, enabling necessary recursive search patterns; and (2) nested group rollout-driven exploratory reinforcement learning on 18K complex multi-turn samples, explicitly guided by a novel multi-step reward model to stimulate deep nested reasoning. Extensive experiments demonstrate that Omni-o3 achieves competitive performance across 11 benchmarks, unlocking advanced capabilities in comprehensive audio-visual, visual-centric, and audio-centric reasoning tasks.

Victor Junqiu Wei, Weicheng Wang, Di Jiang et al.

Automatic speech Recognition (ASR) is a fundamental and important task in the field of speech and natural language processing. It is an inherent building block in many applications such as voice assistant, speech translation, etc. Despite the advancement of ASR technologies in recent years, it is still inevitable for modern ASR systems to have a substantial number of erroneous recognition due to environmental noise, ambiguity, etc. Therefore, the error correction in ASR is crucial. Motivated by this, this paper studies ASR error correction in the Chinese language, which is one of the most popular languages and enjoys a large number of users in the world. We first create a benchmark dataset named \emph{ASR-EC} that contains a wide spectrum of ASR errors generated by industry-grade ASR systems. To the best of our knowledge, it is the first Chinese ASR error correction benchmark. Then, inspired by the recent advances in \emph{large language models (LLMs)}, we investigate how to harness the power of LLMs to correct ASR errors. We apply LLMs to ASR error correction in three paradigms. The first paradigm is prompting, which is further categorized as zero-shot, few-shot, and multi-step. The second paradigm is finetuning, which finetunes LLMs with ASR error correction data. The third paradigm is multi-modal augmentation, which collectively utilizes the audio and ASR transcripts for error correction. Extensive experiments reveal that prompting is not effective for ASR error correction. Finetuning is effective only for a portion of LLMs. Multi-modal augmentation is the most effective method for error correction and achieves state-of-the-art performance.

Shixuan Zhao, Weicheng Wang, Ninghui Li et al.

Protecting sensitive information in data-driven collaborations, such as AI training, while meeting the diverse requirements of multiple mutually distrusted stakeholders, is both crucial and challenging. This paper presents Styx, a novel framework to address this challenge by integrating sticky policies with Trusted Execution Environments (TEEs). At a high level, Styx employs a hardware-TEE-protected middleware with a programming language runtime to form a sandboxed environment for both the data processing and policy enforcement. We carefully designed a data processing workflow and pipelines to enable a strong yet flexible data-specific policy enforcement throughout the entire data lifecycle and data derivation to achieve data-in-use protection, data lifecycle protection and dynamic collaboration. We implemented Styx and demonstrated its ability to make collaborative computing, such as joint AI training, more secure, privacy-preserving, and policy-compliant. Our evaluation shows the performance overheads imposed by Styx are reasonable on single-node computation with the capability to scale to a large distributed multi-node deployment.

Victor Junqiu Wei, Weicheng Wang, Di Jiang et al.

Due to the rising awareness of privacy protection and the voluminous scale of speech data, it is becoming infeasible for Automatic Speech Recognition (ASR) system developers to train the acoustic model with complete data as before. For example, the data may be owned by different curators, and it is not allowed to share with others. In this paper, we propose a novel paradigm to solve salient problems plaguing the ASR field. In the first stage, multiple acoustic models are trained based upon different subsets of the complete speech data, while in the second phase, two novel algorithms are utilized to generate a high-quality acoustic model based upon those trained on data subsets. We first propose the Genetic Merge Algorithm (GMA), which is a highly specialized algorithm for optimizing acoustic models but suffers from low efficiency. We further propose the SGD-Based Optimizational Merge Algorithm (SOMA), which effectively alleviates the efficiency bottleneck of GMA and maintains superior model accuracy. Extensive experiments on public data show that the proposed methods can significantly outperform the state-of-the-art. Furthermore, we introduce Shapley Value to estimate the contribution score of the trained models, which is useful for evaluating the effectiveness of the data and providing fair incentives to their curators.

Shaofeng H. -C. Jiang, Yaonan Jin, Jianing Lou et al.

We study Euclidean $k$-Means under the Massively Parallel Computation (MPC) model, focusing on the \emph{fully-scalable} setting. Our main result is a fully-scalable $O((\log n/\log\log n)^2)$-approximation in $O(1)$ rounds. Previously, fully-scalable algorithms for $k$-Means either run in super-constant $O(\log\log n \cdot \log\log\log n)$ rounds, albeit with a better $O(1)$-approximation [Cohen-Addad et al., SODA'26], or suffer from bicriteria guarantees [Bhaskara and Wijewardena, ICML'18; Czumaj et al., ICALP'24]. Our algorithm also gives an $O(\log n/\log\log n)$-approximation for $k$-Median, which improves a recent $O(\log n)$-approximation [Goranci et al., SODA'26], and this $o(\log n)$ ratio breaks the fundamental barrier of tree embedding methods used therein. Our main technical contribution is a new variant of the MP algorithm [Mettu and Plaxton, SICOMP'03] that works for general metrics, whose new guarantee is the Lagrangian Multiplier Preserving (LMP) property, which, importantly, holds even under arbitrary distance distortions. Allowing distance distortion is crucial for efficient MPC implementations and useful for efficient algorithm design in general, whereas preserving the LMP property under distance distortion is known to be a significant technical challenge. As a byproduct of our techniques, we also obtain an $O(1)$-approximation to the optimal \emph{value} in $O(1)$ rounds, which conceptually suggests that achieving a true $O(1)$-approximation (for the solution) in $O(1)$ rounds may be a sensible goal for future study.