arXiv subjects

Yunfan Gao

Publications and source records attributed to Yunfan Gao.

At least 19 recordsLinked to original sources

arXiv2026

J-Miner: Recovering the Decision Logic of Fine-Tuned LLM Classifiers as Compact Rules

Task-fine-tuned large language model (LLM) classifiers acquire task-specific decision knowledge, but this knowledge remains implicit in distributed internal computations, making their decision logic difficult to interpret. We introduce the Executable Decision Compression (EDC) framework and propose J-Miner, which mines vocabulary-named variables from internal readouts and learns rules shared across inputs to produce executable explanations. Analysis reveals that a small set of these variables captures much of the classifier's decision behavior, holding for both varying parameter scales within a family and distinct families. Across six binary tasks, a rule using just one variable reproduces 76.7% of source-classifier decisions on average, rising to 88.8% with 16 variables. Most of the decision information retained by these variables comes from internal activations beyond literal surface matching. A lightweight text reader predicts the variable states, allowing the same fixed rules to execute independently of the source classifier.

cs.LG↗

arXiv2026

Riccati-ZORO: An efficient algorithm for heuristic online optimization of internal feedback laws in robust and stochastic model predictive control

We present Riccati-ZORO, an algorithm for tube-based optimal control problems (OCP). Tube OCPs predict a tube of trajectories to capture predictive uncertainty. The tube induces a constraint tightening via additional backoff terms. This backoff can significantly affect the performance, and thus implicitly defines a cost of uncertainty. Optimizing the feedback law used to predict the tube can significantly reduce the backoffs, but its online computation is challenging. Riccati-ZORO jointly optimizes the nominal trajectory and uncertainty tube based on a heuristic uncertainty cost design. The algorithm alternates between two subproblems: (i) a nominal OCP with fixed backoffs, (ii) an unconstrained tube OCP, which optimizes the feedback gains for a fixed nominal trajectory. For the tube optimization, we propose a cost function informed by the proximity of the nominal trajectory to constraints, prioritizing reduction of the corresponding backoffs. These ideas are developed for ellipsoidal tubes under linear state feedback. In this case, the decomposition into the two subproblems substantially reduces the computational complexity with respect to the state dimension from $\mathcal{O}(n_x^6)$ to $\mathcal{O}(n_x^3)$, i.e., the complexity of a nominal OCP. We investigate the algorithm in numerical experiments, and provide two open-source implementations: a prototyping version in CasADi and a high-performance implementation integrated into the acados OCP solver.

math.OC↗

arXiv2026

Hierarchical Textual Knowledge for Enhanced Image Clustering

Image clustering aims to group images in an unsupervised fashion. Traditional methods focus on knowledge from visual space, making it difficult to distinguish between visually similar but semantically different classes. Recent advances in vision-language models enable the use of textual knowledge to enhance image clustering. However, most existing methods rely on coarse class labels or simple nouns, overlooking the rich conceptual and attribute-level semantics embedded in textual space. In this paper, we propose a knowledge-enhanced clustering (KEC) method that constructs a hierarchical concept-attribute structured knowledge with the help of large language models (LLMs) to guide clustering. Specifically, we first condense redundant textual labels into abstract concepts and then automatically extract discriminative attributes for each single concept and similar concept pairs, via structured prompts to LLMs. This knowledge is instantiated for each input image to achieve the knowledge-enhanced features. The knowledge-enhanced features with original visual features are adapted to various downstream clustering algorithms. We evaluate KEC on 20 diverse datasets, showing consistent improvements across existing methods using additional textual knowledge. KEC without training outperforms zero-shot CLIP on 14 out of 20 datasets. Furthermore, the naive use of textual knowledge may harm clustering performance, while KEC provides both accuracy and robustness.

cs.CV↗

arXiv2026

HingeMem: Boundary Guided Long-Term Memory with Query Adaptive Retrieval for Scalable Dialogues

Long-term memory is critical for dialogue systems that support continuous, sustainable, and personalized interactions. However, existing methods rely on continuous summarization or OpenIE-based graph construction paired with fixed Top-\textit{k} retrieval, leading to limited adaptability across query categories and high computational overhead. In this paper, we propose HingeMem, a boundary-guided long-term memory that operationalizes event segmentation theory to build an interpretable indexing interface via boundary-triggered hyperedges over four elements: person, time, location, and topic. When any such element changes, HingeMem draws a boundary and writes the current segment, thereby reducing redundant operations and preserving salient context. To enable robust and efficient retrieval under diverse information needs, HingeMem introduces query-adaptive retrieval mechanisms that jointly decide (a) \textit{what to retrieve}: determine the query-conditioned routing over the element-indexed memory; (b) \textit{how much to retrieve}: control the retrieval depth based on the estimated query type. Extensive experiments across LLM scales (from 0.6B to production-tier models; \textit{e.g.}, Qwen3-0.6B to Qwen-Flash) on LOCOMO show that HingeMem achieves approximately $20\%$ relative improvement over strong baselines without query categories specification, while reducing computational cost (68\%$\downarrow$ question answering token cost compared to HippoRAG2). Beyond advancing memory modeling, HingeMem's adaptive retrieval makes it a strong fit for web applications requiring efficient and trustworthy memory over extended interactions.

cs.CL↗

arXiv2026

Semi-Infinite Programming for Collision-Avoidance in Optimal and Model Predictive Control

This paper presents a novel approach for collision avoidance in optimal and model predictive control, in which the environment is represented by a large number of points and the robot as a union of padded polygons. The conditions that none of the points shall collide with the robot can be written in terms of an infinite number of constraints per obstacle point. We show that the resulting semi-infinite programming (SIP) optimal control problem (OCP) can be efficiently tackled through a combination of two methods: local reduction and an external active-set method. Specifically, this involves iteratively identifying the closest point obstacles, determining the lower-level distance minimizer among all feasible robot shape parameters, and solving the upper-level finitely-constrained subproblems. In addition, this paper addresses robust collision avoidance in the presence of ellipsoidal state uncertainties. Enforcing constraint satisfaction over all possible uncertainty realizations extends the dimension of constraint infiniteness. The infinitely many constraints arising from translational uncertainty are handled by local reduction together with the robot shape parameterization, while rotational uncertainty is addressed via a backoff reformulation. A controller implemented based on the proposed method is demonstrated on a real-world robot running at 20Hz, enabling fast and collision-free navigation in tight spaces. An application to 3D collision avoidance is also demonstrated in simulation.

cs.RO↗

arXiv2026

Embodied Science: Closing the Discovery Loop with Agentic Embodied AI

Artificial intelligence has demonstrated remarkable capability in predicting scientific properties, yet scientific discovery remains an inherently physical, long-horizon pursuit governed by experimental cycles. Most current computational approaches are misaligned with this reality, framing discovery as isolated, task-specific predictions rather than continuous interaction with the physical world. Here, we argue for embodied science, a paradigm that reframes scientific discovery as a closed loop tightly coupling agentic reasoning with physical execution. We propose a unified Perception-Language-Action-Discovery (PLAD) framework, wherein embodied agents perceive experimental environments, reason over scientific knowledge, execute physical interventions, and internalize outcomes to drive subsequent exploration. By grounding computational reasoning in robust physical feedback, this approach bridges the gap between digital prediction and empirical validation, offering a roadmap for autonomous discovery systems in the life and chemical sciences.

cs.AI↗

arXiv2026

Lyra: A Hardware-Accelerated RISC-V Verification Framework with Generative Model-Based Processor Fuzzing

As processor designs grow more complex, verification remains bottlenecked by slow software simulation and low-quality random test stimuli. Recent research has applied software fuzzers to hardware verification, but these rely on semantically blind random mutations that may generate shallow, low-quality stimuli unable to explore complex behaviors. These limitations result in slow coverage convergence and prohibitively high verification costs. In this paper, we present Lyra, a heterogeneous RISC-V verification framework that addresses both challenges by pairing hardware-accelerated verification with an ISA-aware generative model. Lyra executes the DUT and reference model concurrently on an FPGA SoC, enabling high-throughput differential checking and hardware-level coverage collection. Instead of creating verification stimuli randomly or through simple mutations, we train a domain-specialized generative model, LyraGen, with inherent semantic awareness to generate high-quality, semantically rich instruction sequences. Empirical results show Lyra achieves up to $1.27\times$ higher coverage and accelerates end-to-end verification by up to $107\times$ to $3343\times$ compared to state-of-the-art software fuzzers, while consistently demonstrating lower convergence difficulty.

cs.AR↗

arXiv2025

CitySeeker: How Do VLMS Explore Embodied Urban Navigation With Implicit Human Needs?

Vision-Language Models (VLMs) have made significant progress in explicit instruction-based navigation; however, their ability to interpret implicit human needs (e.g., "I am thirsty") in dynamic urban environments remains underexplored. This paper introduces CitySeeker, a novel benchmark designed to assess VLMs' spatial reasoning and decision-making capabilities for exploring embodied urban navigation to address implicit needs. CitySeeker includes 6,440 trajectories across 8 cities, capturing diverse visual characteristics and implicit needs in 7 goal-driven scenarios. Extensive experiments reveal that even top-performing models (e.g., Qwen2.5-VL-32B-Instruct) achieve only 21.1% task completion. We find key bottlenecks in error accumulation in long-horizon reasoning, inadequate spatial cognition, and deficient experiential recall. To further analyze them, we investigate a series of exploratory strategies-Backtracking Mechanisms, Enriching Spatial Cognition, and Memory-Based Retrieval (BCR), inspired by human cognitive mapping's emphasis on iterative observation-reasoning cycles and adaptive path optimization. Our analysis provides actionable insights for developing VLMs with robust spatial intelligence required for tackling "last-mile" navigation challenges.

cs.AI↗

arXiv2025

Real-Time-Feasible Collision-Free Motion Planning For Ellipsoidal Objects

Online planning of collision-free trajectories is a fundamental task for robotics and self-driving car applications. This paper revisits collision avoidance between ellipsoidal objects using differentiable constraints. Two ellipsoids do not overlap if and only if the endpoint of the vector between the center points of the ellipsoids does not lie in the interior of the Minkowski sum of the ellipsoids. This condition is formulated using a parametric over-approximation of the Minkowski sum, which can be made tight in any given direction. The resulting collision avoidance constraint is included in an optimal control problem (OCP) and evaluated in comparison to the separating-hyperplane approach. Not only do we observe that the Minkowski-sum formulation is computationally more efficient in our experiments, but also that using pre-determined over-approximation parameters based on warm-start trajectories leads to a very limited increase in suboptimality. This gives rise to a novel real-time scheme for collision-free motion planning with model predictive control (MPC). Both the real-time feasibility and the effectiveness of the constraint formulation are demonstrated in challenging real-world experiments.

cs.RO↗

arXiv2025

Synergizing RAG and Reasoning: A Systematic Review

Recent breakthroughs in large language models (LLMs), particularly in reasoning capabilities, have propelled Retrieval-Augmented Generation (RAG) to unprecedented levels. By synergizing retrieval mechanisms with advanced reasoning, LLMs can now tackle increasingly complex problems. This paper presents a systematic review of the collaborative interplay between RAG and reasoning, clearly defining "reasoning" within the RAG context. It construct a comprehensive taxonomy encompassing multi-dimensional collaborative objectives, representative paradigms, and technical implementations, and analyze the bidirectional synergy methods. Additionally, we critically evaluate current limitations in RAG assessment, including the absence of intermediate supervision for multi-step reasoning and practical challenges related to cost-risk trade-offs. To bridge theory and practice, we provide practical guidelines tailored to diverse real-world applications. Finally, we identify promising research directions, such as graph-based knowledge integration, hybrid model collaboration, and RL-driven optimization. Overall, this work presents a theoretical framework and practical foundation to advance RAG systems in academia and industry, fostering the next generation of RAG solutions.

cs.IR↗

arXiv2025

StePO-Rec: Towards Personalized Outfit Styling Assistant via Knowledge-Guided Multi-Step Reasoning

Advancements in Generative AI offers new opportunities for FashionAI, surpassing traditional recommendation systems that often lack transparency and struggle to integrate expert knowledge, leaving the potential for personalized fashion styling remain untapped. To address these challenges, we present PAFA (Principle-Aware Fashion), a multi-granular knowledge base that organizes professional styling expertise into three levels of metadata, domain principles, and semantic relationships. Using PAFA, we develop StePO-Rec, a knowledge-guided method for multi-step outfit recommendation. StePO-Rec provides structured suggestions using a scenario-dimension-attribute framework, employing recursive tree construction to align recommendations with both professional principles and individual preferences. A preference-trend re-ranking system further adapts to fashion trends while maintaining the consistency of the user's original style. Experiments on the widely used personalized outfit dataset IQON show a 28% increase in Recall@1 and 32.8% in MAP. Furthermore, case studies highlight improved explainability, traceability, result reliability, and the seamless integration of expertise and personalization.

cs.IR↗

arXiv2025

U-NIAH: Unified RAG and LLM Evaluation for Long Context Needle-In-A-Haystack

Recent advancements in Large Language Models (LLMs) have expanded their context windows to unprecedented lengths, sparking debates about the necessity of Retrieval-Augmented Generation (RAG). To address the fragmented evaluation paradigms and limited cases in existing Needle-in-a-Haystack (NIAH), this paper introduces U-NIAH, a unified framework that systematically compares LLMs and RAG methods in controlled long context settings. Our framework extends beyond traditional NIAH by incorporating multi-needle, long-needle, and needle-in-needle configurations, along with different retrieval settings, while leveraging the synthetic Starlight Academy dataset-a fictional magical universe-to eliminate biases from pre-trained knowledge. Through extensive experiments, we investigate three research questions: (1) performance trade-offs between LLMs and RAG, (2) error patterns in RAG, and (3) RAG's limitations in complex settings. Our findings show that RAG significantly enhances smaller LLMs by mitigating the "lost-in-the-middle" effect and improving robustness, achieving an 82.58% win-rate over LLMs. However, we observe that retrieval noise and reverse chunk ordering degrade performance, while surprisingly, advanced reasoning LLMs exhibit reduced RAG compatibility due to sensitivity to semantic distractors. We identify typical error patterns including omission due to noise, hallucination under high noise critical condition, and self-doubt behaviors. Our work not only highlights the complementary roles of RAG and LLMs, but also provides actionable insights for optimizing deployments. Code: https://github.com/Tongji-KGLLM/U-NIAH.

cs.CL↗

arXiv2024

Decoding Urban Industrial Complexity: Enhancing Knowledge-Driven Insights via IndustryScopeGPT

Industrial parks are critical to urban economic growth. Yet, their development often encounters challenges stemming from imbalances between industrial requirements and urban services, underscoring the need for strategic planning and operations. This paper introduces IndustryScopeKG, a pioneering large-scale multi-modal, multi-level industrial park knowledge graph, which integrates diverse urban data including street views, corporate, socio-economic, and geospatial information, capturing the complex relationships and semantics within industrial parks. Alongside this, we present the IndustryScopeGPT framework, which leverages Large Language Models (LLMs) with Monte Carlo Tree Search to enhance tool-augmented reasoning and decision-making in Industrial Park Planning and Operation (IPPO). Our work significantly improves site recommendation and functional planning, demonstrating the potential of combining LLMs with structured datasets to advance industrial park management. This approach sets a new benchmark for intelligent IPPO research and lays a robust foundation for advancing urban industrial development. The dataset and related code are available at https://github.com/Tongji-KGLLM/IndustryScope.

cs.AI↗

arXiv2024

Modular RAG: Transforming RAG Systems into LEGO-like Reconfigurable Frameworks

Retrieval-augmented Generation (RAG) has markedly enhanced the capabilities of Large Language Models (LLMs) in tackling knowledge-intensive tasks. The increasing demands of application scenarios have driven the evolution of RAG, leading to the integration of advanced retrievers, LLMs and other complementary technologies, which in turn has amplified the intricacy of RAG systems. However, the rapid advancements are outpacing the foundational RAG paradigm, with many methods struggling to be unified under the process of "retrieve-then-generate". In this context, this paper examines the limitations of the existing RAG paradigm and introduces the modular RAG framework. By decomposing complex RAG systems into independent modules and specialized operators, it facilitates a highly reconfigurable framework. Modular RAG transcends the traditional linear architecture, embracing a more advanced design that integrates routing, scheduling, and fusion mechanisms. Drawing on extensive research, this paper further identifies prevalent RAG patterns-linear, conditional, branching, and looping-and offers a comprehensive analysis of their respective implementation nuances. Modular RAG presents innovative opportunities for the conceptualization and deployment of RAG systems. Finally, the paper explores the potential emergence of new operators and paradigms, establishing a solid theoretical foundation and a practical roadmap for the continued evolution and practical deployment of RAG technologies.

cs.CL↗

arXiv2024

Stochastic Model Predictive Control with Optimal Linear Feedback for Mobile Robots in Dynamic Environments

Robot navigation around humans can be a challenging problem since human movements are hard to predict. Stochastic model predictive control (MPC) can account for such uncertainties and approximately bound the probability of a collision to take place. In this paper, to counteract the rapidly growing human motion uncertainty over time, we incorporate state feedback in the stochastic MPC. This allows the robot to more closely track reference trajectories. To this end the feedback policy is left as a degree of freedom in the optimal control problem. The stochastic MPC with feedback is validated in simulation experiments and is compared against nominal MPC and stochastic MPC without feedback. The added computation time can be limited by reducing the number of additional variables for the feedback law with a small compromise in control performance.

cs.RO↗

arXiv2024

Retrieval-Augmented Generation for Large Language Models: A Survey

Large Language Models (LLMs) showcase impressive capabilities but encounter challenges like hallucination, outdated knowledge, and non-transparent, untraceable reasoning processes. Retrieval-Augmented Generation (RAG) has emerged as a promising solution by incorporating knowledge from external databases. This enhances the accuracy and credibility of the generation, particularly for knowledge-intensive tasks, and allows for continuous knowledge updates and integration of domain-specific information. RAG synergistically merges LLMs' intrinsic knowledge with the vast, dynamic repositories of external databases. This comprehensive review paper offers a detailed examination of the progression of RAG paradigms, encompassing the Naive RAG, the Advanced RAG, and the Modular RAG. It meticulously scrutinizes the tripartite foundation of RAG frameworks, which includes the retrieval, the generation and the augmentation techniques. The paper highlights the state-of-the-art technologies embedded in each of these critical components, providing a profound understanding of the advancements in RAG systems. Furthermore, this paper introduces up-to-date evaluation framework and benchmark. At the end, this article delineates the challenges currently faced and points out prospective avenues for research and development.

cs.CL↗

arXiv2023

Efficient Zero-Order Robust Optimization for Real-Time Model Predictive Control with acados

Robust and stochastic optimal control problem (OCP) formulations allow a systematic treatment of uncertainty, but are typically associated with a high computational cost. The recently proposed zero-order robust optimization (zoRO) algorithm mitigates the computational cost of uncertainty-aware MPC by propagating the uncertainties outside of the MPC problem. This paper details the combination of zoRO with the real-time iteration (RTI) scheme and presents an efficient open-source implementation in acados, utilizing BLASFEO for the linear algebra operations. In addition to the scaling advantages posed by the zoRO algorithm, the efficient implementation drastically reduces the computational overhead, and, combined with an RTI scheme, enables the use of tube-based MPC for a wider range of applications. The flexibility, usability and effectiveness of the proposed implementation is demonstrated on two examples. On the practical example of a differential drive robot, the proposed implementation results in a tenfold reduction of computation time with respect to the previously available zoRO implementation.

math.OC↗

arXiv2023

Collision-free Motion Planning for Mobile Robots by Zero-order Robust Optimization-based MPC

This paper presents an implementation of robust model predictive control (MPC) for collision-free reference trajectory tracking for mobile robots. The presented approach considers the robot motion to be subject to process noise bounded by ellipsoidal sets. In order to efficiently handle the evolution of the disturbance ellipsoids within the MPC, the zero-order robust optimization (zoRO) scheme is applied. The idea is to fix the disturbance ellipsoids within one optimization iteration and solve the problem repeatedly with updated disturbance ellipsoid trajectories. The zero-order approach is suboptimal in general. However, we show that it does not impair convergence to the reference trajectory in the absence of obstacles. The experiments on an industrial mobile robot prototype demonstrate the performance of the controller.

eess.SY↗