This article investigates the optimal control problem of a stochastic nonlinear strict-feedback system subjected to both Wiener and Poisson noises. Since the strict-feedback system does not satisfy the matching condition and involves unknown nonlinear terms and unmeasurable stochastic noises in modeling, an optimized backstepping (OB) technique based on the reinforcement learning (RL) strategy is adopted to design the controller within the identifier–critic–actor architecture. However, the OB technique needs to construct all the virtual and actual controllers as the optimal solutions of their respective subsystems, which inevitably increases the complexity of the algorithm. To alleviate this situation, a novel RL method is proposed, so that the optimized control is unaffected by the disturbances induced by both Wiener and Poisson noises. In addition, an adaptive neural network identifier is incorporated into the RL framework to ensure that the proposed control scheme can be smoothly applied to the unknown nonlinear dynamic system. Finally, a vehicle tracking control example is presented to demonstrate the effectiveness of the proposed method.

In sports analytics, tactical visualization is widely used to convey valuable insights. However, due to the complex domain knowledge and contextual information involved in tactical visualizations, it is challenging for users to connect high-level tactical insights to corresponding visual patterns. This requires users to engage in a reasoning process to interpret insights within game contexts, which remains insufficiently supported in existing visual-text linking studies. In this work, we propose InsightChaser, a novel approach to bridge tactical insights and soccer visualizations through visual-text linking and visual reasoning enhancement. InsightChaser constructs knowledge graphs to represent both visual elements and contextual game information. Integrating large language models (LLMs), our approach retrieves relevant visual elements and establishes explicit links with insights. Moreover, InsightChaser utilizes LLMs to enhance these visual-text links by providing reasoning explanations and visual effects. We further develop an interactive visualization system that supports navigation and explanation of enhanced visual-text links. Users can explore linked tactical insights interactively and reason through enhanced visual explanations. We conduct two case studies using real-world soccer data and a user study to demonstrate the effectiveness of our approach.

Large Language Models (LLMs) have achieved remarkable success across various domains.However, a fundamental question remains: Can LLMs effectively utilize causal knowledge for prediction and generation?Through empirical studies, we find that LLMs trained directly on large-scale data often capture spurious correlations rather than true causal relationships, leading to suboptimal performance, especially in out-of-distribution (OOD) scenarios.To address this challenge, we propose Causal Attention Tuning (CAT), a novel approach that injects fine-grained causal knowledge into the attention mechanism.We propose an automated pipeline that leverages human priors to automatically generate token-level causal signals and introduce the Re-Attention mechanism to guide training, helping the model focus on causal structures while mitigating noise and biases in attention scores.Experimental results on our proposed Spurious Token Game (STG) benchmark and multiple downstream tasks demonstrate that our approach effectively leverages causal knowledge for prediction and remains robust in OOD scenarios.The CAT achieves an average improvement of 5.76% on the STG dataset and 1.56% on downstream tasks.Notably, the OOD performance of the Llama-3.1-8Bmodel on STG_M increased from 64.5% to 90.5%, and Qwen's OOD performance on the STG_H dataset improved from 25.4% to 55.9%.Implementation details can be found here.

Recently, large reasoning models (LRMs) have demonstrated state-of-the-art performance across a wide range of benchmarks.However, a common challenge for these models is the "overthinking" problem, which leads to excessive reasoning steps and significant computational overhead.Furthermore, the issues with long Chain-of-Thought (CoT) are especially pronounced in smaller models (≤ 3B parameters).Aside from producing excessively verbose "reflection words", they often exhibit repetition and get trapped in unproductive generation loops.Existing solutions typically involve either using flexible reasoning chains as training data or leveraging the model's latent space to bypass intermediate reasoning steps, but none of these methods have considered directly optimizing reasoning trajectories during the sampling phase of training.In our work, we introduce the Multi-Turn Intervention Sampling Framework (MuTIS).Our framework leverages multi-turn interventions to produce concise reasoning chains.It fine-tunes reasoning models through reinforcement learning, demonstrably breaking the accuracy-efficiency trade-off.It also demonstrates strong scalability, exhibiting excellent performance on 7B models.

Tactics play an important role in team sports by guiding how players interact on the field. Both sports fans and experts have a demand for analyzing sports tactics. Existing approaches allow users to visually perceive the multivariate tactical effects. However, these approaches require users to experience a complex reasoning process to connect the multiple interactions within each tactic to the final tactical effect. In this work, we collaborate with basketball experts and propose a progressive approach to help users gain a deeper understanding of how each tactic works and customize tactics on demand. Users can progressively sketch on a tactic board, and a coach agent will simulate the possible actions in each step and present the simulation to users with facet visualizations. We develop an extensible framework that integrates large language models (LLMs) and visualizations to help users communicate with the coach agent with multimodal inputs. Based on the framework, we design and develop Smartboard, an agent-based interactive visualization system for fine-grained tactical analysis, especially for play design. Smartboard provides users with a structured process of setup, simulation, and evolution, allowing for iterative exploration of tactics based on specific personalized scenarios. We conduct case studies based on real-world basketball datasets to demonstrate the effectiveness and usefulness of our system.

Deep hashing has been widely applied in large-scale image retrieval due to its powerful computational efficiency. Nevertheless, the vulnerability of deep hashing to adversarial examples has been revealed, particularly to targeted attacks with stronger manipulability. Existing targeted attack methods for deep hashing default to selecting target labels from random images, usually encompassing multiple classes for attack in multi-label datasets. However, they exhabit poor performance when facing a preciser single target label selection. In this work, we propose a novel Precise Target-Oriented Attack dubbed PTA, to enhance the precision of such targeted attacks. Specifically, we further categorize the general target label into preciser single target label for attack. By relaxing the non-differentiable indicator function, we directly adopt Average Precision (AP) as optimization objective to guide the generation of adversarial examples on a small subset of the entire database, thus achieving stronger precision. Extensive experiments demonstrate that the proposed PTA achieves state-of-the-art performance in both general and single target label selection, with superior transferability and universality.