arXiv subjects

Sirui Zhao

Publications and source records attributed to Sirui Zhao.

At least 19 recordsLinked to original sources

arXiv2026

DynaForcing: Overcoming Dynamic Collapse in Self-Forcing Distillation for Streaming Avatar Generation

Audio-driven avatar generation requires realistic lip-sync, expressive motion, and real-time streaming. Recent work achieves the latter via self-forcing with Distribution Matching Distillation (DMD), but this paradigm suffers from a critical failure that has not been systematically characterized: dynamic collapse, where the student model converges to a near-static optimum with high perceptual quality but severely suppressed temporal dynamics. We trace this to two causes: the reverse KL objective in DMD, which biases toward low-motion modes, and unanchored self-conditioning, which creates a feedback loop that amplifies collapse. This is especially harmful for avatars, where even subtle motion loss breaks lip-sync and expression. To address this, we propose DynaForcing, a training framework with three complementary strategies applied at different levels. Specifically, Hybrid Forcing anchors rollouts to ground-truth dynamics at the data level to break the feedback loop. Dynamics-Aware Reward Regularization introduces explicit motion rewards via the RL interpretation of DMD to counteract the reverse KL bias at the loss level. Reference Perturbation perturbs reference images to decouple identity from static details, forcing the model to rely on audio for motion at the conditioning level. We further introduce computation graph pruning and gradient replay, reducing the GPU footprint of self-forcing by over an order of magnitude. Experiments show that DynaForcing recovers dynamics to teacher-comparable levels (Dyn-Deg: 0.31 -> 0.73, Sync-C: 7.03 -> 7.68) while improving visual quality, resolving the quality-dynamics trade-off throughout training without early stopping.

cs.CV↗

arXiv2026

Agentic Artificial Intelligence in Finance: A Comprehensive Survey

The emergence of agentic artificial intelligence (AI) represents a fundamental transformation in financial markets, characterized by autonomous systems capable of reasoning, planning, and adaptive decision-making with minimal human intervention. This comprehensive survey synthesizes recent advances in agentic AI across multiple dimensions of financial operations, including system architecture, market applications, regulatory frameworks, and systemic implications. We examine how agentic AI differs from traditional algorithmic trading and generative AI through its capacity for goal-oriented autonomy, continuous learning, and multi-agent coordination. Our analysis shows that while agentic AI offers substantial potential for enhanced market efficiency, liquidity provision, and risk management, it also introduces novel challenges related to market stability, regulatory compliance, interpretability, and systemic risk. Through a systematic review of foundational research, technical architectures, market applications, and governance frameworks, this survey provides scholars and practitioners with a structured understanding of how agentic AI is reshaping financial markets and identifies critical research directions for ensuring that these systems enhance both operational efficiency and market resilience.

econ.EM↗

arXiv2026

EviDep: Trustworthy Multimodal Depression Estimation via Disentangled Evidential Learning

Automated multimodal depression estimation in unconstrained environments is inherently challenged by naturalistic noise and complex behavioral variability. Prevailing deterministic methods, however, produce uncalibrated point estimates without quantifying predictive uncertainty, exposing decision-making to the risk of overconfident, untrustworthy estimates. To establish a reliable and trustworthy estimation paradigm, we propose EviDep, an evidential learning framework that jointly quantifies depression severity alongside aleatoric and epistemic uncertainties via a Normal-Inverse-Gamma distribution. To ensure the integrity of the extracted behavioral evidence and prevent artificial confidence inflation during multimodal fusion, EviDep introduces two tailored mechanisms. First, addressing the temporal-frequency heterogeneity of behavioral cues, a Frequency-aware Feature Extraction module leverages a wavelet-based Mixture-of-Experts to dynamically decouple stable macro-level affective baselines from transient micro-level behavioral bursts, effectively filtering out task-irrelevant artifacts. Second, a Disentangled Evidential Learning strategy enforces explicit decorrelation of features in these purified representations. By separating the cross-modal shared consensus from modality-specific behavioral nuances before Bayesian fusion, this rigorous disentanglement strictly prevents the model from double-counting overlapping information. Extensive experiments on the AVEC 2013, AVEC 2014, DAIC-WOZ, and E-DAIC datasets confirm that EviDep achieves state-of-the-art predictive accuracy and superior uncertainty calibration, thereby delivering a trustworthy, risk-aware decision-support tool for depression estimation.

cs.LG↗

arXiv2026

ActFER: Agentic Facial Expression Recognition via Active Tool-Augmented Visual Reasoning

Recent advances in Multimodal Large Language Models (MLLMs) have created new opportunities for facial expression recognition (FER), moving it beyond pure label prediction toward reasoning-based affect understanding. However, existing MLLM-based FER methods still follow a passive paradigm: they rely on externally prepared facial inputs and perform single-pass reasoning over fixed visual evidence, without the capability for active facial perception. To address this limitation, we propose ActFER, an agentic framework that reformulates FER as active visual evidence acquisition followed by multimodal reasoning. Specifically, ActFER dynamically invokes tools for face detection and alignment, selectively zooms into informative local regions, and reasons over facial Action Units (AUs) and emotions through a visual Chain-of-Thought. To realize such behavior, we further develop Utility-Calibrated GRPO (UC-GRPO), a reinforcement learning algorithm tailored to agentic FER. UC-GRPO uses AU-grounded multi-level verifiable rewards to densify supervision, query-conditional contrastive utility estimation to enable sample-aware dynamic credit assignment for local inspection, and emotion-aware EMA calibration to reduce noisy utility estimates while capturing emotion-wise inspection tendencies. This algorithm enables ActFER to learn both when local inspection is beneficial and how to reason over the acquired evidence. Comprehensive experiments show that ActFER trained with UC-GRPO consistently outperforms passive MLLM-based FER baselines and substantially improves AU prediction accuracy.

cs.CV↗

arXiv2026

Tango: Taming Visual Signals for Efficient Video Large Language Models

Token pruning has emerged as a mainstream approach for developing efficient Video Large Language Models (Video LLMs). This work revisits and advances the two predominant token-pruning paradigms: attention-based selection and similarity-based clustering. Our study reveals two critical limitations in existing methods: (1) conventional top-k selection strategies fail to fully account for the attention distribution, which is often spatially multi-modal and long-tailed in magnitude; and (2) direct similarity-based clustering frequently generates fragmented clusters, resulting in distorted representations after pooling. To address these bottlenecks, we propose Tango, a novel framework designed to optimize the utilization of visual signals. Tango integrates a diversity-driven strategy to enhance attention-based token selection, and introduces Spatio-temporal Rotary Position Embedding (ST-RoPE) to preserve geometric structure via locality priors. Comprehensive experiments across various Video LLMs and video understanding benchmarks demonstrate the effectiveness and generalizability of our approach. Notably, when retaining only 10% of the video tokens, Tango preserves 98.9% of the original performance on LLaVA-OV while delivering a 1.88$\times$ inference speedup.

cs.CV↗

arXiv2026

SyncBreaker:Stage-Aware Multimodal Adversarial Attacks on Audio-Driven Talking Head Generation

Diffusion-based audio-driven talking-head generation enables realistic portrait animation, but also introduces risks of misuse, such as fraud and misinformation. Existing protection methods are largely limited to a single modality, and neither image-only nor audio-only attacks can effectively suppress speech-driven facial dynamics. To address this gap, we propose SyncBreaker, a stage-aware multimodal protection framework that jointly perturbs portrait and audio inputs under modality-specific perceptual constraints. Our key contributions are twofold. First, for the image stream, we introduce nullifying supervision with Multi-Interval Sampling (MIS) across diffusion stages to steer the generation toward the static reference portrait by aggregating guidance from multiple denoising intervals. Second, for the audio stream, we propose Cross-Attention Fooling (CAF), which suppresses interval-specific audio-conditioned cross-attention responses. Both streams are optimized independently and combined at inference time to enable flexible deployment. We evaluate SyncBreaker in a white-box proactive protection setting. Extensive experiments demonstrate that SyncBreaker more effectively degrades lip synchronization and facial dynamics than strong single-modality baselines, while preserving input perceptual quality and remaining robust under purification. Code: https://github.com/kitty384/SyncBreaker.

cs.CV↗

arXiv2026

Dual-Path Learning based on Frequency Structural Decoupling and Regional-Aware Fusion for Low-Light Image Super-Resolution

Low-light image super-resolution (LLISR) is essential for restoring fine visual details and perceptual quality under insufficient illumination conditions with ubiquitous low-resolution devices. Although pioneer methods achieve high performance on single tasks, they solve both tasks in a serial manner, which inevitably leads to artifact amplification, texture suppression, and structural degradation. To address this, we propose Decoupling then Perceive (DTP), a novel frequency-aware framework that explicitly separates luminance and texture into semantically independent components, enabling specialized modeling and coherent reconstruction. Specifically, to adaptively separate the input into low-frequency luminance and high-frequency texture subspaces, we propose a Frequency-aware Structural Decoupling (FSD) mechanism, which lays a solid foundation for targeted representation learning and reconstruction. Based on the decoupled representation, a Semantics-specific Dual-path Representation (SDR) learning strategy that performs targeted enhancement and reconstruction for each frequency component is further designed, facilitating robust luminance adjustment and fine-grained texture recovery. To promote structural consistency and perceptual alignment in the reconstructed output, building upon this dual-path modeling, we further introduce a Cross-frequency Semantic Recomposition (CSR) module that selectively integrates the decoupled representations. Extensive experiments on the most widely used LLISR benchmarks demonstrate the superiority of our DTP framework, improving $+$1.6\% PSNR, $+$9.6\% SSIM, and $-$48\% LPIPS compared to the most state-of-the-art (SOTA) algorithm. Codes are released at https://github.com/JXVision/DTP.

cs.CV↗

arXiv2026

ReMA: A Training-Free Plug-and-Play Mixing Augmentation for Video Behavior Recognition

Video behavior recognition demands stable and discriminative representations under complex spatiotemporal variations. However, prevailing data augmentation strategies for videos remain largely perturbation-driven, often introducing uncontrolled variations that amplify non-discriminative factors, which finally weaken intra-class distributional structure and representation drift with inconsistent gains across temporal scales. To address these problems, we propose Representation-aware Mixing Augmentation (ReMA), a plug-and-play augmentation strategy that formulates mixing as a controlled replacement process to expand representations while preserving class-conditional stability. ReMA integrates two complementary mechanisms. Firstly, the Representation Alignment Mechanism (RAM) performs structured intra-class mixing under distributional alignment constraints, suppressing irrelevant intra-class drift while enhancing statistical reliability. Then, the Dynamic Selection Mechanism (DSM) generates motion-aware spatiotemporal masks to localize perturbations, guiding them away from discrimination-sensitive regions and promoting temporal coherence. By jointly controlling how and where mixing is applied, ReMA improves representation robustness without additional supervision or trainable parameters. Extensive experiments on diverse video behavior benchmarks demonstrate that ReMA consistently enhances generalization and robustness across different spatiotemporal granularities.

cs.CV↗

arXiv2025

DMP-TTS: Disentangled multi-modal Prompting for Controllable Text-to-Speech with Chained Guidance

Controllable text-to-speech (TTS) systems face significant challenges in achieving independent manipulation of speaker timbre and speaking style, often suffering from entanglement between these attributes. We present DMP-TTS, a latent Diffusion Transformer (DiT) framework with explicit disentanglement and multi-modal prompting. A CLAP-based style encoder (Style-CLAP) aligns cues from reference audio and descriptive text in a shared space and is trained with contrastive learning plus multi-task supervision on style attributes. For fine-grained control during inference, we introduce chained classifier-free guidance (cCFG) trained with hierarchical condition dropout, enabling independent adjustment of content, timbre, and style guidance strengths. Additionally, we employ Representation Alignment (REPA) to distill acoustic-semantic features from a pretrained Whisper model into intermediate DiT representations, stabilizing training and accelerating convergence. Experiments show that DMP-TTS delivers stronger style controllability than open-source baselines while maintaining competitive intelligibility and naturalness. Code and demos will be available at https://y61329697.github.io/DMP-TTS/.

cs.SD↗

arXiv2025

Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length

Audio-driven avatar interaction demands real-time, streaming, and infinite-length generation -- capabilities fundamentally at odds with the sequential denoising and long-horizon drift of current diffusion models. We present Live Avatar, an algorithm-system co-designed framework that addresses both challenges for a 14-billion-parameter diffusion model. On the algorithm side, a two-stage pipeline distills a pretrained bidirectional model into a causal, few-step streaming one, while a set of complementary long-horizon strategies eliminate identity drift and visual artifacts, enabling stable autoregressive generation exceeding 10000 seconds. On the system side, Timestep-forcing Pipeline Parallelism (TPP) assigns each GPU a fixed denoising timestep, converting the sequential diffusion chain into an asynchronous spatial pipeline that simultaneously boosts throughput and improves temporal consistency. Live Avatar achieves 45 FPS with a TTFF of 1.21\,s on 5 H800 GPUs, and to our knowledge is the first to enable practical real-time streaming of a 14B diffusion model for infinite-length avatar generation. We further introduce GenBench, a standardized long-form benchmark, to facilitate reproducible evaluation. Our project page is at https://liveavatar.github.io/.

cs.CV↗

arXiv2025

LIHE: Linguistic Instance-Split Hyperbolic-Euclidean Framework for Generalized Weakly-Supervised Referring Expression Comprehension

Existing Weakly-Supervised Referring Expression Comprehension (WREC) methods, while effective, are fundamentally limited by a one-to-one mapping assumption, hindering their ability to handle expressions corresponding to zero or multiple targets in realistic scenarios. To bridge this gap, we introduce the Weakly-Supervised Generalized Referring Expression Comprehension task (WGREC), a more practical paradigm that handles expressions with variable numbers of referents. However, extending WREC to WGREC presents two fundamental challenges: supervisory signal ambiguity, where weak image-level supervision is insufficient for training a model to infer the correct number and identity of referents, and semantic representation collapse, where standard Euclidean similarity forces hierarchically-related concepts into non-discriminative clusters, blurring categorical boundaries. To tackle these challenges, we propose a novel WGREC framework named Linguistic Instance-Split Hyperbolic-Euclidean (LIHE), which operates in two stages. The first stage, Referential Decoupling, predicts the number of target objects and decomposes the complex expression into simpler sub-expressions. The second stage, Referent Grounding, then localizes these sub-expressions using HEMix, our innovative hybrid similarity module that synergistically combines the precise alignment capabilities of Euclidean proximity with the hierarchical modeling strengths of hyperbolic geometry. This hybrid approach effectively prevents semantic collapse while preserving fine-grained distinctions between related concepts. Extensive experiments demonstrate LIHE establishes the first effective weakly supervised WGREC baseline on gRefCOCO and Ref-ZOM, while HEMix achieves consistent improvements on standard REC benchmarks, improving IoU@0.5 by up to 2.5\%. The code is available at https://anonymous.4open.science/r/LIHE.

cs.CV↗

arXiv2025

Robust Low-Rank Sparse Framework for Video-Based Affective Computing

Video-based Affective Computing (VAC), vital for emotion analysis and human-computer interaction, suffers from model instability and representational degradation due to complex emotional dynamics. Since the meaning of different emotional fluctuations may differ under different emotional contexts, the core limitation is the lack of a hierarchical structural mechanism to disentangle distinct affective components, i.e., emotional bases (the long-term emotional tone), and transient fluctuations (the short-term emotional fluctuations). To address this, we propose the Low-Rank Sparse Emotion Understanding Framework (LSEF), a unified model grounded in the Low-Rank Sparse Principle, which theoretically reframes affective dynamics as a hierarchical low-rank sparse compositional process. LSEF employs three plug-and-play modules, i.e., the Stability Encoding Module (SEM) captures low-rank emotional bases; the Dynamic Decoupling Module (DDM) isolates sparse transient signals; and the Consistency Integration Module (CIM) reconstructs multi-scale stability and reactivity coherence. This framework is optimized by a Rank Aware Optimization (RAO) strategy that adaptively balances gradient smoothness and sensitivity. Extensive experiments across multiple datasets confirm that LSEF significantly enhances robustness and dynamic discrimination, which further validates the effectiveness and generality of hierarchical low-rank sparse modeling for understanding affective dynamics.

cs.CV↗

arXiv2025

Bind-Your-Avatar: Multi-Talking-Character Video Generation with Dynamic 3D-mask-based Embedding Router

Recent years have witnessed remarkable advances in audio-driven talking head generation. However, existing approaches predominantly focus on single-character scenarios. While some methods can create separate conversation videos between two individuals, the critical challenge of generating unified conversation videos with multiple physically co-present characters sharing the same spatial environment remains largely unaddressed. This setting presents two key challenges: audio-to-character correspondence control and the lack of suitable datasets featuring multi-character talking videos within the same scene. To address these challenges, we introduce Bind-Your-Avatar, an MM-DiT-based model specifically designed for multi-talking-character video generation in the same scene. Specifically, we propose (1) A novel framework incorporating a fine-grained Embedding Router that binds `who' and `speak what' together to address the audio-to-character correspondence control. (2) Two methods for implementing a 3D-mask embedding router that enables frame-wise, fine-grained control of individual characters, with distinct loss functions based on observed geometric priors and a mask refinement strategy to enhance the accuracy and temporal smoothness of the predicted masks. (3) The first dataset, to the best of our knowledge, specifically constructed for multi-talking-character video generation, and accompanied by an open-source data processing pipeline, and (4) A benchmark for the dual-talking-characters video generation, with extensive experiments demonstrating superior performance over multiple state-of-the-art methods.

cs.CV↗

arXiv2025

MERba: Multi-Receptive Field MambaVision for Micro-Expression Recognition

Micro-expressions (MEs) are brief, involuntary facial movements that reveal genuine emotions, offering valuable insights for psychological assessment and criminal investigations. Despite significant progress in automatic ME recognition (MER), existing methods still struggle to simultaneously capture localized muscle activations and global facial dependencies, both essential for decoding subtle emotional cues. To address this challenge, we propose MERba, a hierarchical multi-receptive field architecture specially designed for MER, which incorporates a series of Local-Global Feature Integration stages. Within each stage, detailed intra-window motion patterns are captured using MERba Local Extractors, which integrate MambaVision Mixers with a tailored asymmetric multi-scanning strategy to enhance local spatial sensitivity. These localized features are then aggregated through lightweight self-attention layers that explicitly model inter-window relationships, enabling effective global context construction. Furthermore, to mitigate the challenge of high inter-class similarity among negative MEs, we introduce a Dual-Granularity Classification Module that decomposes the recognition task into a coarse-to-fine paradigm. Extensive experiments on three benchmark datasets demonstrate that MERba consistently outperforms existing methods, with ablation studies confirming the effectiveness of each proposed component.

cs.HC↗

arXiv2025

MELLM: A Flow-Guided Large Language Model for Micro-Expression Understanding

Micro-expressions (MEs), brief and low-intensity facial movements revealing concealed emotions, are crucial for affective computing. Despite notable progress in ME recognition, existing methods are largely confined to discrete emotion classification, lacking the capacity for comprehensive ME Understanding (MEU), particularly in interpreting subtle facial dynamics and underlying emotional cues. While Multimodal Large Language Models (MLLMs) offer potential for MEU with their advanced reasoning abilities, they still struggle to perceive such subtle facial affective behaviors. To bridge this gap, we propose a ME Large Language Model (MELLM) that integrates optical flow-based sensitivity to subtle facial motions with the powerful inference ability of LLMs. Specifically, an iterative, warping-based optical-flow estimator, named MEFlowNet, is introduced to precisely capture facial micro-movements. For its training and evaluation, we construct MEFlowDataset, a large-scale optical-flow dataset with 54,611 onset-apex image pairs spanning diverse identities and subtle facial motions. Subsequently, we design a Flow-Guided Micro-Expression Understanding paradigm. Under this framework, the optical flow signals extracted by MEFlowNet are leveraged to build MEU-Instruct, an instruction-tuning dataset for MEU. MELLM is then fine-tuned on MEU-Instruct, enabling it to translate subtle motion patterns into human-readable descriptions and generate corresponding emotional inferences. Experiments demonstrate that MEFlowNet significantly outperforms existing optical flow methods in facial and ME-flow estimation, while MELLM achieves state-of-the-art accuracy and generalization across multiple ME benchmarks. To the best of our knowledge, this work presents two key contributions: MEFlowNet as the first dedicated ME flow estimator, and MELLM as the first LLM tailored for MEU.

cs.CV↗

arXiv2025

MER-CLIP: AU-Guided Vision-Language Alignment for Micro-Expression Recognition

As a critical psychological stress response, micro-expressions (MEs) are fleeting and subtle facial movements revealing genuine emotions. Automatic ME recognition (MER) holds valuable applications in fields such as criminal investigation and psychological diagnosis. The Facial Action Coding System (FACS) encodes expressions by identifying activations of specific facial action units (AUs), serving as a key reference for ME analysis. However, current MER methods typically limit AU utilization to defining regions of interest (ROIs) or relying on specific prior knowledge, often resulting in limited performance and poor generalization. To address this, we integrate the CLIP model's powerful cross-modal semantic alignment capability into MER and propose a novel approach namely MER-CLIP. Specifically, we convert AU labels into detailed textual descriptions of facial muscle movements, guiding fine-grained spatiotemporal ME learning by aligning visual dynamics and textual AU-based representations. Additionally, we introduce an Emotion Inference Module to capture the nuanced relationships between ME patterns and emotions with higher-level semantic understanding. To mitigate overfitting caused by the scarcity of ME data, we put forward LocalStaticFaceMix, an effective data augmentation strategy blending facial images to enhance facial diversity while preserving critical ME features. Finally, comprehensive experiments on four benchmark ME datasets confirm the superiority of MER-CLIP. Notably, UF1 scores on CAS(ME)3 reach 0.7832, 0.6544, and 0.4997 for 3-, 4-, and 7-class classification tasks, significantly outperforming previous methods.

cs.HC↗

arXiv2024

Sparrow: Data-Efficient Video-LLM with Text-to-Image Augmentation

Recent years have seen the success of Multimodal Large Language Models (MLLMs) in the domain of vision understanding. The success of these models can largely be attributed to the dominant scaling law, which states that larger parameter sizes and data volumes contribute to better performance. Notably, data scaling has been primarily driven by automatic data pipelines, which focus on the self-instruction of LLMs. The paradigm has been taken for granted for quite some time, but the study of the effectiveness of scaling with these data has been neglected for a long time. In this context, this work revisits scaling with synthetic data and focuses on developing video-LLMs from a data-centric perspective. Our primary study approach involves fine-tuning pre-trained image-LLMs with video data and examining learning efficiency through data scaling. Results from our preliminary experiments reveal a low learning efficiency phenomenon when simply scaling up video data samples, which, through our probing, can be ascribed to a lack of instruction diversity. Aiming at this issue, we propose a data augmentation method called Sparrow, which synthesizes video-like samples from pure text instruction data. Mixing these synthetic samples with the video data enables a more efficient training scheme. Through comprehensive experiments, we demonstrate that our proposed method achieves performance comparable to or even superior to that of baselines trained with significantly more samples. Meanwhile, we find that incorporating these synthetic samples can enhance the performance of long video understanding without requiring training on long video data. The code and data examples are available at https://github.com/VITA-MLLM/Sparrow.

cs.CV↗

arXiv2024

MME-Survey: A Comprehensive Survey on Evaluation of Multimodal LLMs

As a prominent direction of Artificial General Intelligence (AGI), Multimodal Large Language Models (MLLMs) have garnered increased attention from both industry and academia. Building upon pre-trained LLMs, this family of models further develops multimodal perception and reasoning capabilities that are impressive, such as writing code given a flow chart or creating stories based on an image. In the development process, evaluation is critical since it provides intuitive feedback and guidance on improving models. Distinct from the traditional train-eval-test paradigm that only favors a single task like image classification, the versatility of MLLMs has spurred the rise of various new benchmarks and evaluation methods. In this paper, we aim to present a comprehensive survey of MLLM evaluation, discussing four key aspects: 1) the summarised benchmarks types divided by the evaluation capabilities, including foundation capabilities, model self-analysis, and extented applications; 2) the typical process of benchmark counstruction, consisting of data collection, annotation, and precautions; 3) the systematic evaluation manner composed of judge, metric, and toolkit; 4) the outlook for the next benchmark. This work aims to offer researchers an easy grasp of how to effectively evaluate MLLMs according to different needs and to inspire better evaluation methods, thereby driving the progress of MLLM research.

cs.CV↗