THTianyu Hu

Tianyu Hu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Tianyu Hu has a total of 48 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Discovery of Multicomponent Invar Materials With Low Thermal Expansion, High Curie Temperature, and Superior Phase Stability Assisted by Active Learning

    by Song Yang, Jinghan Li, Wanda Yang, Mingyi Li, Tianyu Hu, Kenichi Kato, Ke An, Yan Chen, Dunji Yu, Jun Miao, et al.

    Zero thermal expansion (ZTE) materials provide exceptional dimensional stability under temperature fluctuations, making them indispensable for high‐precision instrumentation and extreme environments. However, their natural scarcity, combined with the vast compositional space in multicomponent systems, renders traditional trial‐and‐error approaches both time‐consuming and cost‐prohibitive, posing an emergent challenge for modern high‐tech applications. Here, we establish a task‐specific active‐learning framework with empirical fine‐tuning to accelerate ZTE materials' design. By mining sparse experimental datasets, we identified a high‐potential compositional region and uncovered several novel low‐expansion alloys. Among them, Cr 1.4 Co 8.9 Ni 30.5 Fe 59 exhibits a low coefficient of thermal expansion of , and a high Curie temperature reaching 580 . Real‐time in situ neutron and synchrotron x‐ray diffraction confirm its single‐phase face‐centered cubic structure, excellent ductility, and robust phase stability against thermal and mechanical stimuli. Crucially, machine learning analysis pinpointed six key descriptors highly correlated with low‐expansion performance, providing data‐driven insights into the magnetovolume origins of this behavior. This work not only yields a high‐performance dimensionally stable alloy, but also demonstrates how integrating physical insights with data‐driven design can accelerate advanced materials development.

  • Open Access

    Autonomous driving system based on dual process theory and deliberate practice theory

    by Xiao Zhang, Tianyu Hu, Juntao Lyu, Qiankun Liu, Qianchuan Zhao, Mingyi Li, Kang Li, Songde Han, Jiansheng Chen, Huimin Ma

    Autonomous driving, despite significant progress, is still not widely applied in open, unconstrained environments, primarily owing to deficiencies in hazard perception, few-shot generalization, corner case generalization, and evaluation metrics, resulting in reliability concerns. To address these challenges, we propose CogniDrive, a framework based on dual-process and deliberate practice theories, leveraging contextual reasoning of the Large Language Model (LLM) to enhance driving systems robustness and generalization. Inspired by dual-process theory, CogniDrive comprises two cognition modes: InstinctNav for rapid, intuitive decision-making and ReflectPlan for reflective reasoning. Enhanced by a thought model and experience embedding for LLM, InstinctNav combines behavioral cloning and retrieval augmented generation to enhance few-shot learning efficiency based on deliberate practice theory. ReflectPlan processes and internalizes reward signals embedded in language tokens within the prompt, derived from a self-reflection mechanism, to enable continuous improvement and generalization. To detect hazards in corner cases precluded by limited training data distribution, a vision language model is integrated for comprehensive environmental understanding through multimodal self-reflection. We further propose an evaluation framework that complements traditional metrics by emphasizing safety, comfort, and energy efficiency, and demonstrate state-of-the-art performance through extensive open-loop and closed-loop experiments.


  • Spatio-Temporal Probabilistic Forecasting of Wind Speed Using Transformer-Based Diffusion Models

    by Hao Liu, Junqi Liu, Tianyu Hu, Huimin Ma

    Spatio-temporal wind speed forecasting plays a crucial role in enhancing energy conversion efficiency and optimizing resource allocation, forming a cornerstone of sustainable development. However, existing methods for spatio-temporal wind speed forecasting face challenges in capturing intricate spatio-temporal dependencies and adapting to dynamic variations in wind speed data. To address these limitations, we propose the Probabilistic Spatio-Temporal Diffusion Transformer (PSTDT), a novel model that combines the generative power of denoising diffusion probabilistic models with the robust spatio-temporal modeling capabilities of Transformers. This approach introduces a dual-spatial attention module to model static positional relationships and dynamic dependencies from historical data, thereby capturing evolving spatial correlations. Additionally, PSTDT integrates a dual-phase temporal module for modeling cross-period temporal dependencies alongside auto-regressive temporal features, enhancing its capacity to address the complexities of time-series forecasting. Furthermore, a temporal-adaptive layer normalization is incorporated to dynamically adjust normalization parameters, improving the model’s forecast accuracy and stability. Extensive experiments demonstrate that PSTDT outperforms state-of-the-art methods across multiple datasets, reducing continuous ranked probability score by 8% –20% and mean absolute error by 7% –19% .


  • Open Access

    Efficient Knowledge Transfer in Multi-Task Learning through Task-Adaptive Low-Rank Representation

    by Xiao Zhang, Kangsheng Wang, Tianyu Hu, Huimin Ma

    Pre-trained language models (PLMs) demonstrate remarkable intelligence but struggle with emerging tasks unseen during training in real-world applications. Training separate models for each new task is usually impractical. Multi-task learning (MTL) addresses this challenge by transferring shared knowledge from source tasks to target tasks. As an dominant parameter-efficient fine-tuning method, prompt tuning (PT) enhances MTL by introducing an adaptable vector that captures task-specific knowledge, which acts as a prefix to the original prompt that preserves shared knowledge, while keeping PLM parameters frozen. However, PT struggles to effectively capture the heterogeneity of task-specific knowledge due to its limited representational capacity. To address this challenge, we propose Task-Adaptive Low-Rank Representation (TA-LoRA), an MTL method built on PT, employing the low-rank representation to model task heterogeneity and a fast-slow weights mechanism where the slow weight encodes shared knowledge, while the fast weight captures task-specific nuances, avoiding the mixing of shared and task-specific knowledge, caused by training lowrank representations from scratch. Moreover, a zero-initialized attention mechanism is introduced to minimize the disruption of immature low-rank components on original prompts during warm-up epochs. Experiments on 16 tasks demonstrate that TALoRA achieves state-of-the-art performance in full-data and fewshot settings while maintaining superior parameter efficiency.


  • Open Access

    RRT-MVS: Recurrent Regularization Transformer for Multi-View Stereo

    by Jianfei Jiang, Liyong Wang, Haochen Yu, Tianyu Hu, Jiansheng Chen, Huimin Ma

    Learning-based multi-view stereo methods aim to predict depth maps for reconstructing dense point clouds. These methods rely on regularization to reduce redundancy in the cost volume. However, existing methods have limitations: CNN-based regularization is restricted to local receptive fields, while Transformer-based regularization struggles with handling depth discontinuities. These limitations often result in inaccurate depth maps with significant noise, particularly noticeable in the boundary and background regions. In this paper, we propose a Recurrent Regularization Transformer for Multi-View Stereo (RRT-MVS), which addresses these limitations by regularizing the cost volume separately for depth and spatial dimensions. Specifically, we introduce Recurrent Self-Attention (R-SA) to aggregate global matching costs within and across the cost maps and filter out noisy feature correlations. Additionally, we present Depth Residual Attention (DRA) to aggregate depth correlations within the cost volume and a Positional Adapter (PA) to enhance 3D positional awareness in each 2D cost map, further augmenting the effectiveness of R-SA. Experimental results demonstrate that RRT-MVS achieves state-of-the-art performance on the DTU and Tanks-and-Temples datasets. Notably, RRT-MVS ranks first on both the Tanks-and-Temples intermediate and advanced benchmarks among all published methods.


  • Open Access

    CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency

    by Kangsheng Wang, Xiao Zhang, Juntao Lyu, Tianyu Hu, Huimin Ma

    Chain-based reasoning methods like chain of thought (CoT) play a rising role in solving reasoning tasks for large language models (LLMs). However, the causal hallucinations between a step of reasoning and corresponding state transitions are becoming a significant obstacle to advancing LLMs' reasoning capabilities, especially in long-range reasoning tasks. This paper proposes a non-chain-based reasoning framework for simultaneous consideration of causal significance and consistency, i.e., the Causal Significance and Consistency Enhancer (CSCE). We customize LLM's loss function utilizing treatment effect assessments to enhance its reasoning ability from two aspects: causal significance and consistency. This ensures that the model captures essential causal relationships and maintains robust and consistent performance across various scenarios. Additionally, we transform the reasoning process from the cascading multiple one-step reasoning commonly used in Chain-Based methods, like CoT, to a causal-enhanced method that outputs the entire reasoning process in one go, further improving the model's reasoning efficiency. Extensive experiments show that our method improves both the reasoning success rate and speed. These improvements further demonstrate that non-chain-based methods can also aid LLMs in completing reasoning tasks.


  • Open Access

    Enhancing Autonomous Driving through Dual-Process Learning with Behavior and Reflection Integration

    by Xiao Zhang, Kangsheng Wang, Tianyu Hu, Huimin Ma

    Contemporary autonomous driving (AD) methodologies, which predominantly convert visual features into control directives, face long-tail challenges due to constraints imposed by limited data distribution. Conversely, human drivers exhibit proficiency in such conditions, underscoring the significance of emulating human cognition in AD systems. Therefore, we introduce Dual-Process Learning (D-PL) approach for cognitiveenhanced decision-making. Inspired by dual-process theory, the D-PL method combines Behavior Pattern Learning (BPL) and Self-Reflective Learning (SRL) to integrate quick, intuitive decisions with deliberate, analytical reasoning, constructing a hierarchical decision model for sophisticated trajectory planning. Our approach improves decision-making, enhances adaptability, and tackles the crucial open-world generalization challenge encountered by current AD methods. Comprehensive evaluations on the nuScenes dataset validate the robustness of our method, demonstrating its superior performance in navigating the intricacies of real-world contrasting with conventional models.

  • Open Access

    Autonomous Driving System based on Dual Process Theory and Deliberate Practice Theory

    by Tianyu Hu, Xiao Zhang, Huimin Ma, Qiankun Liu, Kangsheng Wang, Jiansheng Chen

    Autonomous driving (AD), despite significant progress, is still not widely applied in open, unconstrained environments, primarily owing to deficiencies in hazard perception, few-shot generalization, corner case generalization, and evaluation metrics, resulting in reliability concerns. Therefore, we propose CogniDrive, an innovative framework based on dual-process and deliberate practice theories, leveraging contextual reasoning of the large language model (LLM) to enhance AD robustness and generalization. Inspired by dual-process theory, CogniDrive comprises two cognition modes: InstinctNav for rapid, intuitive decision-making and ReflectPlan for reflective reasoning. Enhanced by a thought model and experience embedding for LLM, InstinctNav combines behavioral cloning and retrieval augmented generation to enhance few-shot learning efficiency based on deliberate practice theory. ReflectPlan processes and internalizes reward signals embedded in language tokens within the prompt, derived from a self-reflection mechanism, to enable continuous improvement and generalization. To detect hazards in corner cases precluded by limited training data distribution, a vision language model is integrated for comprehensive environmental understanding through multimodal self-reflection. Furthermore, we propose an evaluation system that addresses the incompleteness of traditional metrics and emphasizes safety, comfort, and energy efficiency. Experiments on nuScenes and CODA datasets using the proposed evaluation system demonstrate CogniDrive's superior robustness and adaptability.