Xun Wang

Intelligent financial service platforms are increasingly required to deliver continuous, reliable, and interpretable stock prediction services under volatile market conditions. In such service scenarios, stock prediction is not merely a standalone forecasting task, but a reliability-sensitive service process that must integrate heterogeneous market evidence, arbitrate conflicting signals, and adapt its prediction logic as market regimes evolve. To address these issues, this paper proposes TextGrad-Trader, a graph-enhanced multimodal reasoning framework for adaptive stock prediction services. TextGrad-Trader first decomposes technical indicators, corporate fundamentals, news semantics, and relational momentum into role-specific evidence through a graph-inductive expert pool, preserving the traceability of heterogeneous signals before fusion. It then introduces Bayesian contextual gating to estimate signal credibility by jointly considering historical scenario priors and current reasoning coherence, thereby enabling reliability-aware signal arbitration. Based on the conflict-resolved evidence, an adversarial Pareto-reasoning tree evaluates candidate prediction responses under joint utility and downside-risk constraints, improving the robustness of service delivery in uncertain market states. Finally, textual gradient descent converts historical prediction feedback into reusable reasoning guidance, allowing the prompt-parameterized service policy to evolve with non-stationary markets. Experiments on real-world stock data show that TextGrad-Trader improves risk-aware prediction performance, and ablation analyses verify the contribution of its major service components. The source code is publicly available on GitHub.

Collaborative Filtering (CF) is widely used for rating prediction in Recommender Systems (RSs). Traditional CF methods are highly sensitive to data sparsity, while recent Deep Learning (DL) based CF models enhance representation capacity by capturing complex non-linear user–item interactions. However, most DL based CF methods rely on standard Supervised Learning (SL) and optimize prediction errors independently for each interaction. Under sparse data, gradient signals become noisy, which can lead to unstable optimization dynamics and suboptimal convergence when training deep models. To address this issue, we propose a Supervision-Guided Reinforcement Learning framework for rating prediction in Collaborative Filtering, termed SGRLCF. Instead of only minimizing per-interaction regression errors, SGRLCF leverages ground-truth labels to construct reinforcement rewards that guide the learning process and regulate optimization behavior during training. Moreover, we design a dual-level reward scheme that jointly considers immediate prediction deviation and overall training improvement, providing structured feedback to stabilize learning and enhance generalization. Extensive experiments on four real-world datasets show that SGRLCF outperforms state-of-the-art baselines in terms of RMSE and MAE in both single-domain and cross-domain scenarios.

Following the success of deep neural networks in 2D pose estimation, reconstruction-based approaches have significantly advanced multi-person 3D pose estimation from sparse multi-view images. These methods typically detect 2D poses independently in each view and then associate them for 3D reconstruction. However, despite strong progress, recent state-of-the-art methods still face critical limitations: 1) They often depend on global optimization over a large and complex set of multi-view 2D joints to jointly infer 3D poses for all individuals, making the process highly complex and prone to suboptimal solutions; 2) Their tight coupling with the bottom-up detector OpenPose hinders the use of more advanced top-down or single-stage 2D pose estimators and restricts the integration of richer instance-level cues learned by these models. To address these limitations, we propose TwinPose, a novel framework that alleviates the complexity of global pose inference by optimizing within person-specific 3D pose subspaces, while fully supporting diverse 2D pose detectors and effectively leveraging pose-instance cues. The key idea is to introduce a twin pose — a 3D counterpart of each 2D pose — that inherits its instance representation and aggregates geometrically consistent 2D joints from other views. All twin poses are unified in a common 3D space, where those belonging to the same individual naturally share a number of bones. This structural property enables association by counting shared bones, forming person-specific subspaces from which each individual's 3D pose can be inferred independently in an efficient and robust manner. Extensive experiments demonstrate that TwinPose achieves state-of-the-art performance in both accuracy and efficiency across multiple public and proprietary datasets. Importantly, it is fully detector-agnostic, allowing seamless integration with current and future advances in 2D pose estimation while remaining highly robust to noisy or imperfect 2D predictions. Project page with code and additional resources: https://github.com/zgspose/TwinPose

目的现有纹理滤波算法难以在保持小尺度或弱梯度结构的同时平滑大尺度或强梯度纹理，而且用于算法训练的合成图像和真实场景图像存在领域偏移。为实现尺度自适应的纹理滤波，提出了一种动态侧窗口滤波核预测网络。为提高算法在真实场景图像上的泛化性能，制作了一个混合合成纹理滤波数据集。方法本文算法分为两个阶段，首先基于Transformer和卷积设计编码—解码模块，生成结构区域分割图和过平滑图，然后基于引导滤波和侧窗口滤波设计了滤波核预测模块，在结构和纹理信息的引导下，预测8组动态侧窗口滤波核的采样点和权重值，分别对过平滑图采样和滤波，最后线性融合，获得了最终滤波结果。本文数据集混合了在分割区域填充多种纹理的填充子数据集和将结构背景融合单种纹理的融合子数据集两部分。结果实验在6个数据集上与18种算法进行比较。相比于性能第2的算法，本文算法在峰值信噪比和结构相似性上均有提升，在视觉效果中增强了细窄结构并减少了纹理残留和颜色不均现象。本文数据集降低了算法混淆真实结构和纹理的概率。结论本文算法综合Transformer的全局依赖捕捉和卷积的低级特征提取优势，借助侧窗口滤波核增强结构保持能力，引入动态采样点和权重值机制实现尺度自适应感知，利用引导信息平衡了对大尺度纹理的平滑和对小尺度结构的重建。本文数据集模拟真实场景图像中的结构和纹理模式，提升了算法对真实结构和纹理的识别以及滤波效果。数据集已开源至https://www.kaggle.com/datasets/klase01/hstfdataset。

Recently, transformer-based Human Mesh Recovery (HMR) from monocular images has achieved remarkable progress by effectively modeling long-range dependencies among body parts. However, the large-scale transformer architectures that drive these state-of-the-art results impose prohibitive computational and storage demands, severely limiting their deployment in real-time or resource-constrained scenarios. Existing knowledge distillation methods offer a potential solution but typically rely on a single distribution, either final outputs or intermediate features, thus failing to capture the diverse and complementary representations inherent in transformer-based HMR models, such as attention patterns. To address these limitations, we propose Multi-dIstribution kNowledge Distillation (MIND), a framework tailored for transformer-based HMR that transfers knowledge from multiple complementary distributions: final outputs for high-level task knowledge, intermediate features for visual representation learning, and internal attention heatmaps to preserve spatial focus patterns. Furthermore, motivated by the inherent conceptual similarity between Generative Adversarial Networks (GANs) and knowledge distillation, we introduce a Wasserstein distillation strategy that leverages the Wasserstein-GAN framework to robustly align these distributions between the teacher and student models. Extensive experiments on Human3.6M, 3DPW, and COCO demonstrate that MIND not only surpasses existing distillation baselines, reducing PA-MPJPE by 3.1mm on 3DPW, but also matches or exceeds the state-of-the-art teacher model HMR2.0 [1] while using only 17.8% of its parameters and 13.7% of its computation cost.

In current text-to-video retrieval (T2VR), videos to be retrieved have been properly trimmed so that a correspondence between the videos and ad-hoc textual queries naturally exists. Note in practice that videos circulated on the Internet and social media platforms, while being relatively short, are typically rich in their content. Often, multiple scenes / actions / events are shown in a single video, leading to a more challenging T2VR setting wherein only part of the video content is relevant w.r.t. a given query. This paper presents a first study on this setting which we term Partially Relevant Video Retrieval (PRVR). Considering that a video typically consists of multiple moments, a video is regarded as partially relevant w.r.t. to a given query if it contains a query-related moment. We formulate the PRVR task as a multiple instance learning problem, and propose a Multi-Scale Similarity Learning (MS-SL++) network that jointly learns both clip-scale and frame-scale similarities to determine the partial relevance between video-query pairs. Extensive experiments on three diverse video-text datasets (TVshow Retrieval, ActivityNet-Captions and Charades-STA) demonstrate the viability of the proposed method.