arXiv subjects

Siwei Liu

Publications and source records attributed to Siwei Liu.

At least 19 recordsLinked to original sources

arXiv2026

SSE-Bio: A Structured Self-Evolving Agent with Agentic Retrieval Policy for Multi-Hop Biomedical Reasoning

Biomedical multi-hop question answering (QA) requires models to connect evidence across intermediate entities such as diseases, drugs, proteins, and phenotypes. Existing agents typically rely on static retrieval workflows or coarse-grained prompt rewriting, which can lead to instruction drift when reasoning procedures need to be updated. We propose SSE-Bio, a structured self-evolving agent with an agentic retrieval policy for multi-hop biomedical reasoning. Instead of globally rewriting agent instructions, SSE-Bio maintains a structured state, selectively retrieves knowledge triplets and prior templates through a trainable proxy policy, and improves its reasoning memory through fine-grained template editing. To optimise retrieval decisions, we introduce a proxy-training strategy based on group relative policy optimization, where the proxy is improved through decision-contrastive groups over alternative retrieval choices. Experiments on three biomedical multi-hop QA benchmarks show that SSE-Bio consistently outperforms existing baselines, achieving an improvement of 6.56 absolute points over the strongest self-evolving baseline on BioHopR.

cs.CL↗

arXiv2026

A Multi-Camera Optical Tag Neuronavigation and AR Augmentation Framework for Non-Invasive Brain Stimulation

Accurate neuronavigation is essential for generating the intended effect with transcranial magnetic stimulation (TMS). Precise coil placement also directly influences stimulation efficacy. Traditional neuronavigation systems often rely on costly and still hard to use and error-prone tracking systems. To solve these limitations, we present a computer-vision-based neuronavigation system for real-time tracking of patient and TMS instrumentation. The system can feed the necessary data for a digital twin to track TMS stimulation targets. We integrate a self-coordinating optical tracking system with multiple consumer-grade cameras and visible tags with a dynamic 3D brain model in Unity. This model updates in real time to represent the current stimulation coil position and the estimated stimulation point to intuitively visualize neural targets for clinicians. We incorporate an augmented reality (AR) module to bridge the gap between the visualization of the digital twin and the real world and project the brain model in real-time onto the head of a patient. AR headsets or mobile AR devices allow clinicians to interactively view and adjust the placement of the stimulation transducer intuitively instead of guidance through abstract numbers and 6D cross hairs on an external screen. The proposed technique provides improved spatial precision as well as accuracy. A case study with ten participants with a medical background also demonstrates that the system has high usability.

cs.HC↗

arXiv2026

Optical Tag-Based Neuronavigation and Augmentation System for Non-Invasive Brain Stimulation

Accurate neuronavigation is critical for effective transcranial magnetic stimulation (TMS), as stimulation outcomes depend directly on precise coil placement. Existing neuronavigation systems are often costly, complex, and prone to tracking errors. To address these limitations, we present a computer vision based neuronavigation system that enables real time tracking of the patient and TMS instrumentation. The system integrates a multi camera optical tracking setup with consumer grade hardware and visible markers to drive a digital twin of the stimulation process. A dynamic 3D brain model in Unity updates in real time to visualize coil position and estimated stimulation targets. Augmented reality (AR) is further incorporated to project this model directly onto the patient's head, enabling intuitive, in situ coil adjustment without reliance on abstract numerical displays. Overall, the proposed approach improves spatial precision and accuracy while enhancing usability.

cs.HC↗

arXiv2025

ProtoMol: Enhancing Molecular Property Prediction via Prototype-Guided Multimodal Learning

Multimodal molecular representation learning, which jointly models molecular graphs and their textual descriptions, enhances predictive accuracy and interpretability by enabling more robust and reliable predictions of drug toxicity, bioactivity, and physicochemical properties through the integration of structural and semantic information. However, existing multimodal methods suffer from two key limitations: (1) they typically perform cross-modal interaction only at the final encoder layer, thus overlooking hierarchical semantic dependencies; (2) they lack a unified prototype space for robust alignment between modalities. To address these limitations, we propose ProtoMol, a prototype-guided multimodal framework that enables fine-grained integration and consistent semantic alignment between molecular graphs and textual descriptions. ProtoMol incorporates dual-branch hierarchical encoders, utilizing Graph Neural Networks to process structured molecular graphs and Transformers to encode unstructured texts, resulting in comprehensive layer-wise representations. Then, ProtoMol introduces a layer-wise bidirectional cross-modal attention mechanism that progressively aligns semantic features across layers. Furthermore, a shared prototype space with learnable, class-specific anchors is constructed to guide both modalities toward coherent and discriminative representations. Extensive experiments on multiple benchmark datasets demonstrate that ProtoMol consistently outperforms state-of-the-art baselines across a variety of molecular property prediction tasks.

cs.LG↗

arXiv2025

GraphMed-LT: Patient-Specific Graph Memory with Latent Clinical Thought Refinement for Multi-Turn Medical Conversations

Multi-turn medical question answering (QA) aims to model realistic clinical diagnosis, where a doctor gathers patient information across multiple turns of conversation. Existing multi-turn medical conversation systems have shown promising progress, but they often rely on accumulated conversation histories as memory, leaving clinical evidence fragmented across turns. We propose GraphMed-LT, a patient-specific graph memory approach with latent clinical thought refinement for multi-turn medical conversations. GraphMed-LT extracts patient-specific clinical triplets from patient responses, retrieves relevant knowledge triplets, and organises them into an incrementally updated graph memory. The graph memory is projected into graph-conditioned evidence tokens and refined inside a trainable doctor agent through hidden-state feedback, enabling the agent to update its internal clinical context before asking follow-up questions or producing the final answer. Experiments on three multi-turn medical QA benchmarks show that GraphMed-LT consistently outperforms existing multi-turn medical conversation baselines across multiple LLM backbones, achieving up to a 6.3 percentage-point absolute improvement over the strongest baseline. Further analyses show that GraphMed-LT asks more answerable follow-up questions and provides consistent gains across medical specialties.

cs.CL↗

arXiv2025

An Iterative Bayesian Robbins--Monro Sequence

This study introduces an iterative Bayesian Robbins--Monro (IBRM) sequence, which unites the classical Robbins--Monro sequence with statistical estimation for faster root-finding under noisy observations. Although the standard Robbins--Monro method iteratively approaches solutions, its convergence speed is limited by noisy measurements and naivety to any prior information about the objective function. The proposed Bayesian sequence dynamically updates the prior distribution with newly obtained observations to accelerate convergence rates and robustness. The paper demonstrates almost sure convergence of the sequence and analyses its convergence rates for both one-dimensional and multi-dimensional problems. We evaluate the method in a practical application that suffers from large variability and allows only a few function evaluations, specifically estimating thresholds in noninvasive brain stimulation, where the method is more robust and accurate than conventional alternatives. Simulations involving 25,000 virtual subjects illustrate reduced error margins and decreased outlier frequency with direct impact on clinical use.

math.OC↗

arXiv2025

A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems

Recent advances in large language models have sparked growing interest in AI agents capable of solving complex, real-world tasks. However, most existing agent systems rely on manually crafted configurations that remain static after deployment, limiting their ability to adapt to dynamic and evolving environments. To this end, recent research has explored agent evolution techniques that aim to automatically enhance agent systems based on interaction data and environmental feedback. This emerging direction lays the foundation for self-evolving AI agents, which bridge the static capabilities of foundation models with the continuous adaptability required by lifelong agentic systems. In this survey, we provide a comprehensive review of existing techniques for self-evolving agentic systems. Specifically, we first introduce a unified conceptual framework that abstracts the feedback loop underlying the design of self-evolving agentic systems. The framework highlights four key components: System Inputs, Agent System, Environment, and Optimisers, serving as a foundation for understanding and comparing different strategies. Based on this framework, we systematically review a wide range of self-evolving techniques that target different components of the agent system. We also investigate domain-specific evolution strategies developed for specialised fields such as biomedicine, programming, and finance, where optimisation objectives are tightly coupled with domain constraints. In addition, we provide a dedicated discussion on the evaluation, safety, and ethical considerations for self-evolving agentic systems, which are critical to ensuring their effectiveness and reliability. This survey aims to provide researchers and practitioners with a systematic understanding of self-evolving AI agents, laying the foundation for the development of more adaptive, autonomous, and lifelong agentic systems.

cs.AI↗

arXiv2025

Large Model Driven Solar Activity AI Forecaster: A Scalable Dual Data-Model Framework

Solar activity drives space weather, affecting Earth's magnetosphere and technological infrastructure, which makes accurate solar flare forecasting critical. Current space weather models under-utilize multi-modal solar data, lack iterative enhancement via expert knowledge, and rely heavily on human forecasters under the Observation-Orientation-Decision-Action (OODA) paradigm. Here we present the "Solar Activity AI Forecaster", a scalable dual data-model driven framework built on foundational models, integrating expert knowledge to autonomously replicate human forecasting tasks with quantifiable outputs. It is implemented in the OODA paradigm and comprises three modules: a Situational Perception Module that generates daily solar situation awareness maps by integrating multi-modal observations; In-Depth Analysis Tools that characterize key solar features (active regions, coronal holes, filaments); and a Flare Prediction Module that forecasts strong flares for the full solar disk and active regions. Executed within a few minutes, the model outperforms or matches human forecasters in generalization across multi-source data, forecast accuracy, and operational efficiency. This work establishes a new paradigm for AI-based space weather forecasting, demonstrating AI's potential to enhance forecast accuracy and efficiency, and paving the way for autonomous operational forecasting systems.

astro-ph.SR↗

arXiv2025

EvoAgentX: An Automated Framework for Evolving Agentic Workflows

Multi-agent systems (MAS) have emerged as a powerful paradigm for orchestrating large language models (LLMs) and specialized tools to collaboratively address complex tasks. However, existing MAS frameworks often require manual workflow configuration and lack native support for dynamic evolution and performance optimization. In addition, many MAS optimization algorithms are not integrated into a unified framework. In this paper, we present EvoAgentX, an open-source platform that automates the generation, execution, and evolutionary optimization of multi-agent workflows. EvoAgentX employs a modular architecture consisting of five core layers: the basic components, agent, workflow, evolving, and evaluation layers. Specifically, within the evolving layer, EvoAgentX integrates three MAS optimization algorithms, TextGrad, AFlow, and MIPRO, to iteratively refine agent prompts, tool configurations, and workflow topologies. We evaluate EvoAgentX on HotPotQA, MBPP, and MATH for multi-hop reasoning, code generation, and mathematical problem solving, respectively, and further assess it on real-world tasks using GAIA. Experimental results show that EvoAgentX consistently achieves significant performance improvements, including a 7.44% increase in HotPotQA F1, a 10.00% improvement in MBPP pass@1, a 10.00% gain in MATH solve accuracy, and an overall accuracy improvement of up to 20.00% on GAIA. The source code is available at: https://github.com/EvoAgentX/EvoAgentX

cs.AI↗

arXiv2025

SEW: Self-Evolving Agentic Workflows for Automated Code Generation

Large Language Models (LLMs) have demonstrated effectiveness in code generation tasks. To enable LLMs to address more complex coding challenges, existing research has focused on crafting multi-agent systems with agentic workflows, where complex coding tasks are decomposed into sub-tasks, assigned to specialized agents. Despite their effectiveness, current approaches heavily rely on hand-crafted agentic workflows, with both agent topologies and prompts manually designed, which limits their ability to automatically adapt to different types of coding problems. To address these limitations and enable automated workflow design, we propose \textbf{S}elf-\textbf{E}volving \textbf{W}orkflow (\textbf{SEW}), a novel self-evolving framework that automatically generates and optimises multi-agent workflows. Extensive experiments on three coding benchmark datasets, including the challenging LiveCodeBench, demonstrate that our SEW can automatically design agentic workflows and optimise them through self-evolution, bringing up to 12\% improvement on LiveCodeBench compared to using the backbone LLM only. Furthermore, by investigating different representation schemes of workflow, we provide insights into the optimal way to encode workflow information with text.

cs.SE↗

arXiv2025

REOBench: Benchmarking Robustness of Earth Observation Foundation Models

Earth observation foundation models have shown strong generalization across multiple Earth observation tasks, but their robustness under real-world perturbations remains underexplored. To bridge this gap, we introduce REOBench, the first comprehensive benchmark for evaluating the robustness of Earth observation foundation models across six tasks and twelve types of image corruptions, including both appearance-based and geometric perturbations. To ensure realistic and fine-grained evaluation, our benchmark focuses on high-resolution optical remote sensing images, which are widely used in critical applications such as urban planning and disaster response. We conduct a systematic evaluation of a broad range of models trained using masked image modeling, contrastive learning, and vision-language pre-training paradigms. Our results reveal that (1) existing Earth observation foundation models experience significant performance degradation when exposed to input corruptions. (2) The severity of degradation varies across tasks, model architectures, backbone sizes, and types of corruption, with performance drop varying from less than 1% to over 20%. (3) Vision-language models show enhanced robustness, particularly in multimodal tasks. REOBench underscores the vulnerability of current Earth observation foundation models to real-world corruptions and provides actionable insights for developing more robust and reliable models. Code and data are publicly available at https://github.com/lx709/REOBench.

cs.CV↗

arXiv2024

Three mechanistically different variability and noise sources in the trial-to-trial fluctuations of responses to brain stimulation

Motor-evoked potentials (MEPs) are among the few directly observable responses to external brain stimulation and serve a variety of applications, often in the form of input-output (IO) curves. Previous statistical models with two variability sources inherently consider the small MEPs at the low-side plateau as part of the neural recruitment properties. However, recent studies demonstrated that small MEP responses under resting conditions are contaminated and over-shadowed by background noise of mostly technical quality, e.g., caused by the amplifier, and suggested that the neural recruitment curve should continue below this noise level. This work intends to separate physiological variability from background noise and improve the description of recruitment behaviour. We developed a triple-variability-source model around a logarithmic logistic function without a lower plateau and incorporated an additional source for background noise. Compared to models with two or fewer variability sources, our approach better described IO characteristics, evidenced by lower Bayesian Information Criterion scores across all subjects and pulse shapes. The model independently extracted hidden variability information across the stimulated neural system and isolated it from background noise, which led to an accurate estimation of the IO curve parameters. This new model offers a robust tool to analyse brain stimulation IO curves in clinical and experimental neuroscience and reduces the risk of spurious results from inappropriate statistical methods. The presented model together with the corresponding calibration method provides a more accurate representation of MEP responses and variability sources, advances our understanding of cortical excitability, and may improve the assessment of neuromodulation effects.

q-bio.NC↗

arXiv2024

SGAC: A Graph Neural Network Framework for Imbalanced and Structure-Aware AMP Classification

Classifying Antimicrobial Peptides (AMPs) from the vast collection of peptides derived from metagenomic sequencing offers a promising avenue for combating antibiotic resistance. However, most existing AMP classification methods rely primarily on sequence-based representations and fail to capture the spatial structural information critical for accurate identification. Although recent graph-based approaches attempt to incorporate structural information, they typically construct residue- or atom-level graphs that introduce redundant atomic details and increase structural complexity. Furthermore, the class imbalance between the small number of known AMPs and the abundant non-AMPs significantly hinders predictive performance. To address these challenges, we employ lightweight OmegaFold to predict the three-dimensional structures of peptides and construct peptide graphs using C {\alpha} atoms to capture their backbone geometry and spatial topology. Building on this representation, we propose the Spatial GNN-based AMP Classifier (SGAC), a novel framework that leverages Graph Neural Networks (GNNs) to extract structural features and generate discriminative graph representations. To handle class imbalance, SGAC incorporates Weight-enhanced Contrastive Learning to cluster structurally similar peptides and separate dissimilar ones through adaptive weighting, and applies Weight-enhanced Pseudo-label Distillation to generate high-confidence pseudo labels for unlabeled samples, achieving balanced and consistent representation learning. Experiments on publicly available AMP and non-AMP datasets demonstrate that SGAC significantly achieves state-of-the-art performance compared to baselines.

q-bio.QM↗

arXiv2024

On the Structural Memory of LLM Agents

Memory plays a pivotal role in enabling large language model~(LLM)-based agents to engage in complex and long-term interactions, such as question answering (QA) and dialogue systems. While various memory modules have been proposed for these tasks, the impact of different memory structures across tasks remains insufficiently explored. This paper investigates how memory structures and memory retrieval methods affect the performance of LLM-based agents. Specifically, we evaluate four types of memory structures, including chunks, knowledge triples, atomic facts, and summaries, along with mixed memory that combines these components. In addition, we evaluate three widely used memory retrieval methods: single-step retrieval, reranking, and iterative retrieval. Extensive experiments conducted across four tasks and six datasets yield the following key insights: (1) Different memory structures offer distinct advantages, enabling them to be tailored to specific tasks; (2) Mixed memory structures demonstrate remarkable resilience in noisy environments; (3) Iterative retrieval consistently outperforms other methods across various scenarios. Our investigation aims to inspire further research into the design of memory systems for LLM-based agents.

cs.CL↗

arXiv2024

A Decade of Deep Learning: A Survey on The Magnificent Seven

Deep learning has fundamentally reshaped the landscape of artificial intelligence over the past decade, enabling remarkable achievements across diverse domains. At the heart of these developments lie multi-layered neural network architectures that excel at automatic feature extraction, leading to significant improvements in machine learning tasks. To demystify these advances and offer accessible guidance, we present a comprehensive overview of the most influential deep learning algorithms selected through a broad-based survey of the field. Our discussion centers on pivotal architectures, including Residual Networks, Transformers, Generative Adversarial Networks, Variational Autoencoders, Graph Neural Networks, Contrastive Language-Image Pre-training, and Diffusion models. We detail their historical context, highlight their mathematical foundations and algorithmic principles, and examine subsequent variants, extensions, and practical considerations such as training methodologies, normalization techniques, and learning rate schedules. Beyond historical and technical insights, we also address their applications, challenges, and potential research directions. This survey aims to serve as a practical manual for both newcomers seeking an entry point into cutting-edge deep learning methods and experienced researchers transitioning into this rapidly evolving domain.

cs.LG↗

arXiv2024

DuSEGO: Dual Second-order Equivariant Graph Ordinary Differential Equation

Graph Neural Networks (GNNs) with equivariant properties have achieved significant success in modeling complex dynamic systems and molecular properties. However, their expressiveness ability is limited by: (1) Existing methods often overlook the over-smoothing issue caused by traditional GNN models, as well as the gradient explosion or vanishing problems in deep GNNs. (2) Most models operate on first-order information, neglecting that the real world often consists of second-order systems, which further limits the model's representation capabilities. To address these issues, we propose the \textbf{Du}al \textbf{S}econd-order \textbf{E}quivariant \textbf{G}raph \textbf{O}rdinary Differential Equation (\method{}) for equivariant representation. Specifically, \method{} apply the dual second-order equivariant graph ordinary differential equations (Graph ODEs) on graph embeddings and node coordinates, simultaneously. Theoretically, we first prove that \method{} maintains the equivariant property. Furthermore, we provide theoretical insights showing that \method{} effectively alleviates the over-smoothing problem in both feature representation and coordinate update. Additionally, we demonstrate that the proposed \method{} mitigates the exploding and vanishing gradients problem, facilitating the training of deep multi-layer GNNs. Extensive experiments on benchmark datasets validate the superiority of the proposed \method{} compared to baselines.

cs.LG↗

arXiv2024

Revolutionizing Disease Diagnosis with simultaneous functional PET/MR and Deeply Integrated Brain Metabolic, Hemodynamic, and Perfusion Networks

Simultaneous functional PET/MR (sf-PET/MR) presents a cutting-edge multimodal neuroimaging technique. It provides an unprecedented opportunity for concurrently monitoring and integrating multifaceted brain networks built by spatiotemporally covaried metabolic activity, neural activity, and cerebral blood flow (perfusion). Albeit high scientific/clinical values, short in hardware accessibility of PET/MR hinders its applications, let alone modern AI-based PET/MR fusion models. Our objective is to develop a clinically feasible AI-based disease diagnosis model trained on comprehensive sf-PET/MR data with the power of, during inferencing, allowing single modality input (e.g., PET only) as well as enforcing multimodal-based accuracy. To this end, we propose MX-ARM, a multimodal MiXture-of-experts Alignment and Reconstruction Model. It is modality detachable and exchangeable, allocating different multi-layer perceptrons dynamically ("mixture of experts") through learnable weights to learn respective representations from different modalities. Such design will not sacrifice model performance in uni-modal situation. To fully exploit the inherent complex and nonlinear relation among modalities while producing fine-grained representations for uni-modal inference, we subsequently add a modal alignment module to line up a dominant modality (e.g., PET) with representations of auxiliary modalities (MR). We further adopt multimodal reconstruction to promote the quality of learned features. Experiments on precious multimodal sf-PET/MR data for Mild Cognitive Impairment diagnosis showcase the efficacy of our model toward clinically feasible precision medicine.

eess.IV↗

arXiv2024

PASCL: Supervised Contrastive Learning with Perturbative Augmentation for Particle Decay Reconstruction

In high-energy physics, particles produced in collision events decay in a format of a hierarchical tree structure, where only the final decay products can be observed using detectors. However, the large combinatorial space of possible tree structures makes it challenging to recover the actual decay process given a set of final particles. To better analyse the hierarchical tree structure, we propose a graph-based deep learning model to infer the tree structure to reconstruct collision events. In particular, we use a compact matrix representation termed as lowest common ancestor generations (LCAG) matrix, to encode the particle decay tree structure. Then, we introduce a perturbative augmentation technique applied to node features, aiming to mimic experimental uncertainties and increase data diversity. We further propose a supervised graph contrastive learning algorithm to utilize the information of inter-particle relations from multiple decay processes. Extensive experiments show that our proposed supervised graph contrastive learning with perturbative augmentation (PASCL) method outperforms state-of-the-art baseline models on an existing physics-based dataset, significantly improving the reconstruction accuracy. This method provides a more effective training strategy for models with the same parameters and makes way for more accurate and efficient high-energy particle physics data analysis.

hep-ph↗