arXiv subjects

Yihan Li

Publications and source records attributed to Yihan Li.

36 records · Page 2Linked to original sources

arXiv2026

PsychEval: A Multi-Session and Multi-Therapy Benchmark for High-Realism AI Psychological Counselor

To develop a reliable AI for psychological assessment, we introduce \texttt{PsychEval}, a multi-session, multi-therapy, and highly realistic benchmark designed to address three key challenges: \textbf{1) Can we train a highly realistic AI counselor?} Realistic counseling is a longitudinal task requiring sustained memory and dynamic goal tracking. We propose a multi-session benchmark (spanning 6-10 sessions across three distinct stages) that demands critical capabilities such as memory continuity, adaptive reasoning, and longitudinal planning. The dataset is annotated with extensive professional skills, comprising over 677 meta-skills and 4577 atomic skills. \textbf{2) How to train a multi-therapy AI counselor?} While existing models often focus on a single therapy, complex cases frequently require flexible strategies among various therapies. We construct a diverse dataset covering five therapeutic modalities (Psychodynamic, Behaviorism, CBT, Humanistic Existentialist, and Postmodernist) alongside an integrative therapy with a unified three-stage clinical framework across six core psychological topics. \textbf{3) How to systematically evaluate an AI counselor?} We establish a holistic evaluation framework with 18 therapy-specific and therapy-shared metrics across Client-Level and Counselor-Level dimensions. To support this, we also construct over 2,000 diverse client profiles. Extensive experimental analysis fully validates the superior quality and clinical fidelity of our dataset. Crucially, \texttt{PsychEval} transcends static benchmarking to serve as a high-fidelity reinforcement learning environment that enables the self-evolutionary training of clinically responsible and adaptive AI counselors.

cs.AI↗

arXiv2025

Graph Neural Networks for Surgical Scene Segmentation

Purpose: Accurate identification of hepatocystic anatomy is critical to preventing surgical complications during laparoscopic cholecystectomy. Deep learning models often struggle with occlusions, long-range dependencies, and capturing the fine-scale geometry of rare structures. This work addresses these challenges by introducing graph-based segmentation approaches that enhance spatial and semantic understanding in surgical scene analyses. Methods: We propose two segmentation models integrating Vision Transformer (ViT) feature encoders with Graph Neural Networks (GNNs) to explicitly model spatial relationships between anatomical regions. (1) A static k Nearest Neighbours (k-NN) graph with a Graph Convolutional Network with Initial Residual and Identity Mapping (GCNII) enables stable long-range information propagation. (2) A dynamic Differentiable Graph Generator (DGG) with a Graph Attention Network (GAT) supports adaptive topology learning. Both models are evaluated on the Endoscapes-Seg50 and CholecSeg8k benchmarks. Results: The proposed approaches achieve up to 7-8% improvement in Mean Intersection over Union (mIoU) and 6% improvement in Mean Dice (mDice) scores over state-of-the-art baselines. It produces anatomically coherent predictions, particularly on thin, rare and safety-critical structures. Conclusion: The proposed graph-based segmentation methods enhance both performance and anatomical consistency in surgical scene segmentation. By combining ViT-based global context with graph-based relational reasoning, the models improve interpretability and reliability, paving the way for safer laparoscopic and robot-assisted surgery through a precise identification of critical anatomical features.

cs.CV↗

arXiv2025

Extreme breakdown of the Einstein relation in liquid water under centrifugation

We present evidence that the Einstein relation (ER) breaks down completely in pure water and dilute aqueous solutions under strong centrifugation fields at 40 oC. Isotopologues (e.g., H2O-18) and solutes migrate at a speed of only 5% of that predicted based on the ER. The ER is restored with the addition of solutes above a transition concentration (ct). We further discovered a new scaling law between the solute's partial molar density, the centrifugal acceleration, and ct, which can be quantitatively described by a two-phase model in analog to the Avrami model for phase transformation. The breakdown may stem from long-range dipole interactions or the hydrogen bond network in water, which are disrupted by the presence of solutes. This report shows that studying transport under centrifugation can be a new strategy to understand fundamental transport properties and complex interactions in liquids.

cond-mat.soft↗

arXiv2025

A Middle Path for On-Premises LLM Deployment: Preserving Privacy Without Sacrificing Model Confidentiality

Privacy-sensitive users require deploying large language models (LLMs) within their own infrastructure (on-premises) to safeguard private data and enable customization. However, vulnerabilities in local environments can lead to unauthorized access and potential model theft. To address this, prior research on small models has explored securing only the output layer within hardware-secured devices to balance model confidentiality and customization. Yet this approach fails to protect LLMs effectively. In this paper, we discover that (1) query-based distillation attacks targeting the secured top layer can produce a functionally equivalent replica of the victim model; (2) securing the same number of layers, bottom layers before a transition layer provide stronger protection against distillation attacks than top layers, with comparable effects on customization performance; and (3) the number of secured layers creates a trade-off between protection and customization flexibility. Based on these insights, we propose SOLID, a novel deployment framework that secures a few bottom layers in a secure environment and introduces an efficient metric to optimize the trade-off by determining the ideal number of hidden layers. Extensive experiments on five models (1.3B to 70B parameters) demonstrate that SOLID outperforms baselines, achieving a better balance between protection and downstream customization.

cs.LG↗

arXiv2025

SKYLENAGE Technical Report: Mathematical Reasoning and Contest-Innovation Benchmarks for Multi-Level Math Evaluation

Large language models (LLMs) now perform strongly on many public math suites, yet frontier separation within mathematics increasingly suffers from ceiling effects. We present two complementary benchmarks: SKYLENAGE-ReasoningMATH, a 100-item, structure-aware diagnostic set with per-item metadata on length, numeric density, and symbolic complexity; and SKYLENAGE-MATH, a 150-item contest-style suite spanning four stages from high school to doctoral under a seven-subject taxonomy. We evaluate fifteen contemporary LLM variants under a single setup and analyze subject x model and grade x model performance. On the contest suite, the strongest model reaches 44% while the runner-up reaches 37%; accuracy declines from high school to doctoral, and top systems exhibit a doctoral-to-high-school retention near 79%. On the reasoning set, the best model attains 81% overall, and hardest-slice results reveal clear robustness gaps between leaders and the mid-tier. In summary, we release SKYLENAGE-ReasoningMATH and report aggregate results for SKYLENAGE-MATH; together, SKYLENAGE provides a hard, reasoning-centered and broadly covering math benchmark with calibrated difficulty and rich metadata, serving as a reference benchmark for future evaluations of mathematical reasoning.

cs.CL↗

arXiv2025

VLMgineer: Vision Language Models as Robotic Toolsmiths

Tool design and use reflect the ability to understand and manipulate the physical world through creativity, planning, and foresight. As such, these capabilities are often regarded as measurable indicators of intelligence across biological species. While much of today's research on robotic intelligence focuses on generating better controllers, inventing smarter tools offers a complementary form of physical intelligence: shifting the onus of problem-solving onto the tool's design. Given the vast and impressive common-sense, reasoning, and creative capabilities of today's foundation models, we investigate whether these models can provide useful priors to automatically design and effectively wield such tools? We present VLMgineer, a framework that harnesses the code generation abilities of vision language models (VLMs) together with evolutionary search to iteratively co-design physical tools and the action plans that operate them to perform a task. We evaluate VLMgineer on a diverse new benchmark of everyday manipulation scenarios that demand creative tool design and use. Across this suite, VLMgineer consistently discovers tools and policies that solve tasks more effectively and innovatively, transforming challenging robotics problems into straightforward executions. It also outperforms VLM-generated designs from human specifications and existing human-crafted tools for everyday tasks. To facilitate future research on automated tool invention, we will release our benchmark and code.

cs.RO↗

arXiv2025

IteraOptiRacing: A Unified Planning-Control Framework for Real-time Autonomous Racing for Iterative Optimal Performance

This paper presents a unified planning-control strategy for competing with other racing cars called IteraOptiRacing in autonomous racing environments. This unified strategy is proposed based on Iterative Linear Quadratic Regulator for Iterative Tasks (i2LQR), which can improve lap time performance in the presence of surrounding racing obstacles. By iteratively using the ego car's historical data, both obstacle avoidance for multiple moving cars and time cost optimization are considered in this unified strategy, resulting in collision-free and time-optimal generated trajectories. The algorithm's constant low computation burden and suitability for parallel computing enable real-time operation in competitive racing scenarios. To validate its performance, simulations in a high-fidelity simulator are conducted with multiple randomly generated dynamic agents on the track. Results show that the proposed strategy outperforms existing methods across all randomly generated autonomous racing scenarios, enabling enhanced maneuvering for the ego racing car.

cs.RO↗

arXiv2025

RAPID Hand: A Robust, Affordable, Perception-Integrated, Dexterous Manipulation Platform for Generalist Robot Autonomy

This paper addresses the scarcity of low-cost but high-dexterity platforms for collecting real-world multi-fingered robot manipulation data towards generalist robot autonomy. To achieve it, we propose the RAPID Hand, a co-optimized hardware and software platform where the compact 20-DoF hand, robust whole-hand perception, and high-DoF teleoperation interface are jointly designed. Specifically, RAPID Hand adopts a compact and practical hand ontology and a hardware-level perception framework that stably integrates wrist-mounted vision, fingertip tactile sensing, and proprioception with sub-7 ms latency and spatial alignment. Collecting high-quality demonstrations on high-DoF hands is challenging, as existing teleoperation methods struggle with precision and stability on complex multi-fingered systems. We address this by co-optimizing hand design, perception integration, and teleoperation interface through a universal actuation scheme, custom perception electronics, and two retargeting constraints. We evaluate the platform's hardware, perception, and teleoperation interface. Training a diffusion policy on collected data shows superior performance over prior works, validating the system's capability for reliable, high-quality data collection. The platform is constructed from low-cost and off-the-shelf components and will be made public to ensure reproducibility and ease of adoption.

cs.RO↗

arXiv2025

Origin Tracer: A Method for Detecting LoRA Fine-Tuning Origins in LLMs

As large language models (LLMs) continue to advance, their deployment often involves fine-tuning to enhance performance on specific downstream tasks. However, this customization is sometimes accompanied by misleading claims about the origins, raising significant concerns about transparency and trust within the open-source community. Existing model verification techniques typically assess functional, representational, and weight similarities. However, these approaches often struggle against obfuscation techniques, such as permutations and scaling transformations. To address this limitation, we propose a novel detection method Origin-Tracer that rigorously determines whether a model has been fine-tuned from a specified base model. This method includes the ability to extract the LoRA rank utilized during the fine-tuning process, providing a more robust verification framework. This framework is the first to provide a formalized approach specifically aimed at pinpointing the sources of model fine-tuning. We empirically validated our method on thirty-one diverse open-source models under conditions that simulate real-world obfuscation scenarios. We empirically analyze the effectiveness of our framework and finally, discuss its limitations. The results demonstrate the effectiveness of our approach and indicate its potential to establish new benchmarks for model verification.

cs.AI↗

arXiv2025

Global-Local Distillation Network-Based Audio-Visual Speaker Tracking with Incomplete Modalities

In speaker tracking research, integrating and complementing multi-modal data is a crucial strategy for improving the accuracy and robustness of tracking systems. However, tracking with incomplete modalities remains a challenging issue due to noisy observations caused by occlusion, acoustic noise, and sensor failures. Especially when there is missing data in multiple modalities, the performance of existing multi-modal fusion methods tends to decrease. To this end, we propose a Global-Local Distillation-based Tracker (GLDTracker) for robust audio-visual speaker tracking. GLDTracker is driven by a teacher-student distillation model, enabling the flexible fusion of incomplete information from each modality. The teacher network processes global signals captured by camera and microphone arrays, and the student network handles local information subject to visual occlusion and missing audio channels. By transferring knowledge from teacher to student, the student network can better adapt to complex dynamic scenes with incomplete observations. In the student network, a global feature reconstruction module based on the generative adversarial network is constructed to reconstruct global features from feature embedding with missing local information. Furthermore, a multi-modal multi-level fusion attention is introduced to integrate the incomplete feature and the reconstructed feature, leveraging the complementarity and consistency of audio-visual and global-local features. Experimental results on the AV16.3 dataset demonstrate that the proposed GLDTracker outperforms existing state-of-the-art audio-visual trackers and achieves leading performance on both standard and incomplete modalities datasets, highlighting its superiority and robustness in complex conditions. The code and models will be available.

cs.CV↗

arXiv2023

i2LQR: Iterative LQR for Iterative Tasks in Dynamic Environments

This work introduces a novel control strategy called Iterative Linear Quadratic Regulator for Iterative Tasks (i2LQR), which aims to improve closed-loop performance with local trajectory optimization for iterative tasks in a dynamic environment. The proposed algorithm is reference-free and utilizes historical data from previous iterations to enhance the performance of the autonomous system. Unlike existing algorithms, the i2LQR computes the optimal solution in an iterative manner at each timestamp, rendering it well-suited for iterative tasks with changing constraints at different iterations. To evaluate the performance of the proposed algorithm, we conduct numerical simulations for an iterative task aimed at minimizing completion time. The results show that i2LQR achieves an optimized performance with respect to learning-based MPC (LMPC) as the benchmark in static environments, and outperforms LMPC in dynamic environments with both static and dynamics obstacles.

eess.SY↗

arXiv2023

Spectral Flow, Llarull's Rigidity Theorem in Odd Dimensions and its Generalization

For a compact spin Riemannian manifold $(M,g^{TM})$ of dimension $n$ such that the associated scalar curvature $k^{TM}$ verifies that $k^{TM}\geqslant n(n-1)$, Llarull's rigidity theorem says that any area-decreasing smooth map $f$ from $M$ to the unit sphere $\mathbb{S}^{n}$ of nonzero degree is an isometry. We present in this paper a new proof for Llarull's rigidity theorem in odd dimensions via a spectral flow argument. This approach also works for a generalization of Llarrull's theorem when the sphere $\mathbb{S}^{n}$ is replaced by an arbitrary smooth strictly convex closed hypersurface in $\mathbb{R}^{n+1}$. The results answer two questions by Gromov.

math.DG↗

arXiv2022

Asymptotic Spectral Flow

In this paper we study the asymptotic behavior of the spectral flow of a one-parameter family $\{D_s\}$ of Dirac operators acting on the spinor bunldle $S$ twisted by a vector bundle $E$ of rank $k$, with the parameter $s\in [0,r]$ when $r$ gets sufficiently large. Our method uses the variation of eta invariant and local index theory technique. The key is a uniform estimate of the eta invariant $\barη(D_r)$ which is established via local index theory technique and heat kernel estimate.

math.DG↗

arXiv2022

Learning Invariable Semantical Representation from Language for Extensible Policy Generalization

Recently, incorporating natural language instructions into reinforcement learning (RL) to learn semantically meaningful representations and foster generalization has caught many concerns. However, the semantical information in language instructions is usually entangled with task-specific state information, which hampers the learning of semantically invariant and reusable representations. In this paper, we propose a method to learn such representations called element randomization, which extracts task-relevant but environment-agnostic semantics from instructions using a set of environments with randomized elements, e.g., topological structures or textures, yet the same language instruction. We theoretically prove the feasibility of learning semantically invariant representations through randomization. In practice, we accordingly develop a hierarchy of policies, where a high-level policy is designed to modulate the behavior of a goal-conditioned low-level policy by proposing subgoals as semantically invariant representations. Experiments on challenging long-horizon tasks show that (1) our low-level policy reliably generalizes to tasks against environment changes; (2) our hierarchical policy exhibits extensible generalization in unseen new tasks that can be decomposed into several solvable sub-tasks; and (3) by storing and replaying language trajectories as succinct policy representations, the agent can complete tasks in a one-shot fashion, i.e., once one successful trajectory has been attained.

cs.CL↗

arXiv2018

End-to-end Deep Learning from Raw Sensor Data: Atrial Fibrillation Detection using Wearables

We present a convolutional-recurrent neural network architecture with long short-term memory for real-time processing and classification of digital sensor data. The network implicitly performs typical signal processing tasks such as filtering and peak detection, and learns time-resolved embeddings of the input signal. We use a prototype multi-sensor wearable device to collect over 180h of photoplethysmography (PPG) data sampled at 20Hz, of which 36h are during atrial fibrillation (AFib). We use end-to-end learning to achieve state-of-the-art results in detecting AFib from raw PPG data. For classification labels output every 0.8s, we demonstrate an area under ROC curve of 0.9999, with false positive and false negative rates both below $2\times 10^{-3}$. This constitutes a significant improvement on previous results utilising domain-specific feature engineering, such as heart rate extraction, and brings large-scale atrial fibrillation screenings within imminent reach.

stat.ML↗

arXiv2015

Recent Advances in Programmable Photonic-Assisted Ultrabroadband Radio-Frequency Arbitrary Waveform Generation

This paper reviews recent advances in photonic-assisted radio-frequency arbitrary waveform generation (RF-AWG), with emphasis on programmable ultrabroadband microwave and millimeter-wave waveforms. The key enabling components in these techniques are programmable optical pulse shaping, frequency-to-time mapping via dispersive propagation, and high-speed photodetection. The main advantages and challenges of several different photonic RF-AWG schemes are discussed. We further review some proof-of-concept demonstrations of ultrabroadband RF-AWG applications, including high-resolution ranging and ultrabroadband non-line-of-sight channel compensation. Finally, we present recent progress toward RF-AWG with increased time aperture and time-bandwidth product.

physics.optics↗

arXiv2011

Switching of +/-360deg domain wall states in a nanoring by an azimuthal Oersted field

We demonstrate magnetic switching between two $360^\circ$ domain wall vortex states in cobalt nanorings, which are candidate magnetic states for robust and low power MRAM devices. These $360^\circ$ domain wall (DW) or "twisted onion" states can have clockwise or counterclockwise circulation, the two states for data storage. Reliable switching between the states is necessary for any realistic device. We accomplish this switching by applying a circular Oersted field created by passing current through a metal atomic force microscope tip placed at the center of the ring. After initializing in an onion state, we rotate the DWs to one side of the ring by passing a current through the center, and can switch between the two twisted states by reversing the current, causing the DWs to split and meet again on the opposite side of the ring. A larger current will annihilate the DWs and create a perfect vortex state in the rings.

cond-mat.mes-hall↗

arXiv2011

Downlink Power Allocation for Stored Variable-Bit-Rate Videos

In this paper, we study the problem of power allocation for streaming multiple variable-bit-rate (VBR) videos in the downlink of a cellular network. We consider a deterministic model for VBR video traffic and finite playout buffer at the mobile users. The objective is to derive the optimal downlink power allocation for the VBR video sessions, such that the video data can be delivered in a timely fashion without causing playout buffer overflow and underflow. The formulated problem is a nonlinear nonconvex optimization problem. We analyze the convexity conditions for the formulated problem and propose a two-step greedy approach to solve the problem. We also develop a distributed algorithm based on the dual decomposition technique. The performance of the proposed algorithms are validated with simulations using VBR video traces under realistic scenarios.

cs.IT↗