arXiv subjects

Xiaolin Li

Publications and source records attributed to Xiaolin Li.

At least 19 recordsLinked to original sources

arXiv2026

Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines

Industrial LLM teams often ship behavior updates by repeatedly DPO-training a base model on sequences of related preference-data campaigns. The dominant failure mode in this regime is not always classical catastrophic forgetting: a pipeline may preserve previously learned behaviors while still failing to accumulate reusable methodological knowledge about how to train the next campaign. We call this failure mode scientific amnesia. This paper turns that practitioner intuition into a measurable industrial problem. We contribute: (i) a diagnostic suite for amnesia, (ii) a Program-based pipeline that chains FSDP-sharded DPO checkpoints across Qwen2.5-7B-Instruct runs, (iii) a 30-campaign HumanEval subdomain benchmark, and (iv) a comparative diagnostic study of five strategy proposers: random memory, rule-based scheduling, retrieval-only memory, warm-start Bayesian optimization, and MSCL, a meta-scientific memory and reasoner candidate. Across a single-seed 5-condition * 3-step real-LM chain, 4 of 5 candidates degrade in step-level peak pass@1, including MSCL; only the deliberately conservative rule-based schedule improves. Follow-up pilots qualify rather than overturn this finding: in a heterogeneous chain, MSCL is the only completed candidate that improves, whereas in a small multi-seed homogeneous sweep, retrieval-only has the best mean Delta and no pairwise candidate gap is statistically distinguishable. The contribution is therefore diagnostic, not a claim that MSCL solves the problem: scientific amnesia is observable in a production-like continual-DPO pipeline, and conclusions about interventions depend sharply on chain regime, evaluator design, and seed coverage.

cs.AI↗

arXiv2026

Your Privacy My Cloak: Backdoor Attacks on Differentially Private Federated Learning

Prior research suggests that differential privacy (DP) inherently enhances the robustness of federated learning (FL) against backdoor attacks. In this paper, we challenge this assumption. Through an empirical analysis of two baseline attack strategies, we uncover a fundamental tension in DP-FL: while bypassing DP allows state-of-the-art defenses to detect and filter malicious updates, complying with DP inadvertently masks their distinguishing statistical characteristics. Consequently, existing defenses become ineffective as DP reduces the raw backdoor signal. Building on this masking effect, we propose RING, a novel attack that explicitly exploits DP to conceal malicious contributions while maximizing attack impact. By collaboratively crafting adversarial perturbations, compromised clients reconstruct a strong backdoor signal during aggregation without triggering anomaly detection. RING operates as a perturbation layer that is agnostic to the underlying backdoor technique, making it broadly applicable and composable with existing attacks -- a property that significantly amplifies the threat it poses to DP-FL. Extensive evaluations across four image and text datasets under non-iid distributions show that RING achieves an average attack success rate of 90.3% against six state-of-the-art defenses under a moderate privacy budget, an improvement of up to 26.08x over baseline strategies. Finally, we evaluate potential countermeasures and find that mitigating this threat incurs significant utility trade-offs, exposing a fundamental security gap in the deployment of differentially private FL.

cs.LG↗

arXiv2026

Dissipation-induced Nonlinear Topological Gear Switching

Nonlinear interaction enables topological phenomena impossible in linear systems. A paradigm is nonlinear Thouless pump, where the transport of solitons can be topologically quantized even when band occupation is nonuniform. Such nonlinear quantization traditionally requires a time-periodic Hamiltonian with static nonlinearity and, much as in the linear case, is inherently independent of pumping speed. Instead, we demonstrate a dissipation-induced topological gear switching, where quantized soliton transport can be switched on and off via the adiabatic pumping speed itself. This phenomenon has no counterpart in prior conservative nonlinear pumps, nor in linear non-Hermitian pumps. Crucially, quantization here no longer requires a time-periodic nonlinear Hamiltonian; it stems from a genuinely non-equilibrium mechanism captured by an effective conservative model whose \textit{nonlinearity varies aperiodically in time}. Remarkably, a quantized nonlinear transport can be induced even when this nonlinear aperiodic driving is such that the system is pumped from the linear to nonlinear regimes. Our results open a route toward nonequilibrium nonlinear topological matter, where topological effect is dynamically reconfigurable via time-varying nonlinearities, with experimental implications for photonic, atomic, or superconducting platforms and beyond.

cond-mat.quant-gas↗

arXiv2025

Physics-Informed ML Exploration of Structure-Transport Relationships in Hard Carbon

Sodium-ion batteries are a cost-effective and sustainable alternative to lithium-ion systems for large-scale energy storage. Hard carbon (HC) anodes, composed of disordered graphitic and amorphous domains, offer high capacity but exhibit complex, poorly understood ion transport behavior. In particular, the relationship between local microstructure and sodium mobility remains unresolved, hindering rational performance optimization. Here, we introduce a data-driven framework that combines machine-learned interatomic potentials with molecular dynamics simulations to systematically investigate sodium diffusion across a broad range of carbon densities and sodium loadings. By computing per-ion structural descriptors, we identify the microscopic factors that govern ion transport. Unsupervised learning uncovers distinct diffusion modes, including hopping, clustering, and void trapping, while supervised analysis highlights tortuosity and NaNa coordination as primary determinants of mobility. Correlation mapping further connects these transport regimes to processing variables such as bulk density and sodium content. This physics-informed approach establishes quantitative structure-transport relationships that capture the heterogeneity of disordered carbon. Our findings deliver mechanistic insights into sodium-ion dynamics and provide actionable design principles for engineering high-performance HC anodes in next-generation battery systems.

cond-mat.mtrl-sci↗

arXiv2025

Frequency Prior Guided Matching: A Data Augmentation Approach for Generalizable Semi-Supervised Polyp Segmentation

Automated polyp segmentation is essential for early diagnosis of colorectal cancer, yet developing robust models remains challenging due to limited annotated data and significant performance degradation under domain shift. Although semi-supervised learning (SSL) reduces annotation requirements, existing methods rely on generic augmentations that ignore polyp-specific structural properties, resulting in poor generalization to new imaging centers and devices. To address this, we introduce Frequency Prior Guided Matching (FPGM), a novel augmentation framework built on a key discovery: polyp edges exhibit a remarkably consistent frequency signature across diverse datasets. FPGM leverages this intrinsic regularity in a two-stage process. It first learns a domain-invariant frequency prior from the edge regions of labeled polyps. Then, it performs principled spectral perturbations on unlabeled images, aligning their amplitude spectra with this learned prior while preserving phase information to maintain structural integrity. This targeted alignment normalizes domain-specific textural variations, thereby compelling the model to learn the underlying, generalizable anatomical structure. Validated on six public datasets, FPGM establishes a new state-of-the-art against ten competing methods. It demonstrates exceptional zero-shot generalization capabilities, achieving over 10% absolute gain in Dice score in data-scarce scenarios. By significantly enhancing cross-domain robustness, FPGM presents a powerful solution for clinically deployable polyp segmentation under limited supervision.

cs.CV↗

arXiv2025

Mitigating Data Poisoning Attacks to Local Differential Privacy

The distributed nature of local differential privacy (LDP) invites data poisoning attacks and poses unforeseen threats to the underlying LDP-supported applications. In this paper, we propose a comprehensive mitigation framework for popular frequency estimation, which contains a suite of novel defenses, including malicious user detection, attack pattern recognition, and damaged utility recovery. In addition to existing attacks, we explore new adaptive adversarial activities for our mitigation design. For detection, we present a new method to precisely identify bogus reports and thus LDP aggregation can be performed over the ``clean'' data. When the attack behavior becomes stealthy and direct filtering out malicious users is difficult, we further propose a detection that can effectively recognize hidden adversarial patterns, thus facilitating the decision-making of service providers. These detection methods require no additional data and attack information and incur minimal computational cost. Our experiment demonstrates their excellent performance and substantial improvement over previous work in various settings. In addition, we conduct an empirical analysis of LDP post-processing for corrupted data recovery and propose a new post-processing method, through which we reveal new insights into protocol recommendations in practice and key design principles for future research.

cs.CR↗

arXiv2025

DCentNet: Decentralized Multistage Biomedical Signal Classification using Early Exits

DCentNet is a novel decentralized multistage signal classification approach designed for biomedical data from IoT wearable sensors, integrating early exit points (EEP) to enhance energy efficiency and processing speed. Unlike traditional centralized processing methods, which result in high energy consumption and latency, DCentNet partitions a single CNN model into multiple sub-networks using EEPs. By introducing encoder-decoder pairs at EEPs, the system compresses large feature maps before transmission, significantly reducing wireless data transfer and power usage. If an input is confidently classified at an EEP, processing stops early, optimizing efficiency. Initial sub-networks can be deployed on fog or edge devices to further minimize energy consumption. A genetic algorithm is used to optimize EEP placement, balancing performance and complexity. Experimental results on ECG classification show that with one EEP, DCentNet reduces wireless data transmission by 94.54% and complexity by 21%, while maintaining original accuracy and sensitivity. With two EEPs, sensitivity reaches 98.36%, accuracy 97.74%, wireless data transmission decreases by 91.86%, and complexity is reduced by 22%. Implemented on an ARM Cortex-M4 MCU, DCentNet achieves an average power saving of 73.6% compared to continuous wireless ECG transmission.

eess.SP↗

arXiv2025

EquiBoost: An Equivariant Boosting Approach to Molecular Conformation Generation

Molecular conformation generation plays key roles in computational drug design. Recently developed deep learning methods, particularly diffusion models have reached competitive performance over traditional cheminformatical approaches. However, these methods are often time-consuming or require extra support from traditional methods. We propose EquiBoost, a boosting model that stacks several equivariant graph transformers as weak learners, to iteratively refine 3D conformations of molecules. Without relying on diffusion techniques, EquiBoost balances accuracy and efficiency more effectively than diffusion-based methods. Notably, compared to the previous state-of-the-art diffusion method, EquiBoost improves generation quality and preserves diversity, achieving considerably better precision of Average Minimum RMSD (AMR) on the GEOM datasets. This work rejuvenates boosting and sheds light on its potential to be a robust alternative to diffusion models in certain scenarios.

cs.LG↗

arXiv2024

EquiFlow: Equivariant Conditional Flow Matching with Optimal Transport for 3D Molecular Conformation Prediction

Molecular 3D conformations play a key role in determining how molecules interact with other molecules or protein surfaces. Recent deep learning advancements have improved conformation prediction, but slow training speeds and difficulties in utilizing high-degree features limit performance. We propose EquiFlow, an equivariant conditional flow matching model with optimal transport. EquiFlow uniquely applies conditional flow matching in molecular 3D conformation prediction, leveraging simulation-free training to address slow training speeds. It uses a modified Equiformer model to encode Cartesian molecular conformations along with their atomic and bond properties into higher-degree embeddings. Additionally, EquiFlow employs an ODE solver, providing faster inference speeds compared to diffusion models with SDEs. Experiments on the QM9 dataset show that EquiFlow predicts small molecule conformations more accurately than current state-of-the-art models.

cs.LG↗

arXiv2024

Solving bound-state equations in $\text{QCD}_2$ with bosonic and fermionic quarks

We investigate the bound-state equations (BSEs) in two-dimensional QCD in the $N_c\to \infty$ limit, viewed from both the infinite momentum frame (IMF) and the finite momentum frame (FMF). The BSE of a meson in the original 't Hooft model, viz., spinor $\text{QCD}_2$ containing only fermionc quarks, has been extensively studied in literature. In this work, we focus on the BSEs pertaining to two types of "exotic" hadrons, a "tetraquark" which is composed of a bosonic quark and bosonic antiquark, and a "baryon" which is composed of a bosonic antiquark and a fermionic quark. Utilizing the Hamiltonian approach, we derive the corresponding BSEs for both types of "exotic" hadrons, from the perspectives of the light-front and equal-time quantization, and confirm the known results. The recently available BSEs for "tetraquark" in FMF has also been recovered with the aid of the diagrammatic approach. For the first time we also present the BSEs of a "baryon" in FMF in the extended 't Hooft model. By solving various BSEs numerically, we obtain the mass spectra pertaining to "tetraquark" and "baryon" and the corresponding bound-state wave functions of the lowest-lying states. It is numerically demonstrated that, when a "tetraquark" or "baryon" is continuously boosted, the forward-moving component of the bound-state wave function approaches the corresponding light-cone wave function, while the backward-moving component fades away.

hep-ph↗

arXiv2024

M4: Multi-Proxy Multi-Gate Mixture of Experts Network for Multiple Instance Learning in Histopathology Image Analysis

Multiple instance learning (MIL) has been successfully applied for whole slide images (WSIs) analysis in computational pathology, enabling a wide range of prediction tasks from tumor subtyping to inferring genetic mutations and multi-omics biomarkers. However, existing MIL methods predominantly focus on single-task learning, resulting in not only overall low efficiency but also the overlook of inter-task relatedness. To address these issues, we proposed an adapted architecture of Multi-gate Mixture-of-experts with Multi-proxy for Multiple instance learning (M4), and applied this framework for simultaneous prediction of multiple genetic mutations from WSIs. The proposed M4 model has two main innovations: (1) utilizing a mixture of experts with multiple gating strategies for multi-genetic mutation prediction on a single pathological slide; (2) constructing multi-proxy expert network and gate network for comprehensive and effective modeling of pathological image information. Our model achieved significant improvements across five tested TCGA datasets in comparison to current state-of-the-art single-task methods. The code is available at:https://github.com/Bigyehahaha/M4.

cs.CV↗

arXiv2024

Scientific Large Language Models: A Survey on Biological & Chemical Domains

Large Language Models (LLMs) have emerged as a transformative power in enhancing natural language comprehension, representing a significant stride toward artificial general intelligence. The application of LLMs extends beyond conventional linguistic boundaries, encompassing specialized linguistic systems developed within various scientific disciplines. This growing interest has led to the advent of scientific LLMs, a novel subclass specifically engineered for facilitating scientific discovery. As a burgeoning area in the community of AI for Science, scientific LLMs warrant comprehensive exploration. However, a systematic and up-to-date survey introducing them is currently lacking. In this paper, we endeavor to methodically delineate the concept of "scientific language", whilst providing a thorough review of the latest advancements in scientific LLMs. Given the expansive realm of scientific disciplines, our analysis adopts a focused lens, concentrating on the biological and chemical domains. This includes an in-depth examination of LLMs for textual knowledge, small molecules, macromolecular proteins, genomic sequences, and their combinations, analyzing them in terms of model architectures, capabilities, datasets, and evaluation. Finally, we critically examine the prevailing challenges and point out promising research directions along with the advances of LLMs. By offering a comprehensive overview of technical developments in this field, this survey aspires to be an invaluable resource for researchers navigating the intricate landscape of scientific LLMs.

cs.CL↗

arXiv2023

BatmanNet: Bi-branch Masked Graph Transformer Autoencoder for Molecular Representation

Although substantial efforts have been made using graph neural networks (GNNs) for AI-driven drug discovery (AIDD), effective molecular representation learning remains an open challenge, especially in the case of insufficient labeled molecules. Recent studies suggest that big GNN models pre-trained by self-supervised learning on unlabeled datasets enable better transfer performance in downstream molecular property prediction tasks. However, the approaches in these studies require multiple complex self-supervised tasks and large-scale datasets, which are time-consuming, computationally expensive, and difficult to pre-train end-to-end. Here, we design a simple yet effective self-supervised strategy to simultaneously learn local and global information about molecules, and further propose a novel bi-branch masked graph transformer autoencoder (BatmanNet) to learn molecular representations. BatmanNet features two tailored complementary and asymmetric graph autoencoders to reconstruct the missing nodes and edges, respectively, from a masked molecular graph. With this design, BatmanNet can effectively capture the underlying structure and semantic information of molecules, thus improving the performance of molecular representation. BatmanNet achieves state-of-the-art results for multiple drug discovery tasks, including molecular properties prediction, drug-drug interaction, and drug-target interaction, on 13 benchmark datasets, demonstrating its great potential and superiority in molecular representation learning.

cs.LG↗

arXiv2023

High-Sensitive Microwave Electrometry with Enhanced Instantaneous Bandwidth

Rydberg microwave (MW) sensors are superior to conventional antenna-based techniques because of their wide operating frequency range and outstanding potential sensitivity. Here, we demonstrate a Rydberg microwave receiver with a high sensitivity of $62\,\mathrm{nV} \mathrm{cm}^{-1} \mathrm{Hz}^{-1/2}$ and broad instantaneous bandwidth of up to $10.2\,\mathrm{MHz}$. Such excellent performance was achieved by the amplification of one generated sideband wave induced by the strong coupling field in the six-wave mixing process of the Rydberg superheterodyne receiver, which was well predicted by our theory. Our system, which possesses a uniquely enhanced instantaneous bandwidth and high-sensitivity features that can be improved further, will promote the application of Rydberg microwave electrometry in radar and communication.

quant-ph↗

arXiv2023

Direct and in situ examination of Li+ transport kinetics in isotope labelled solid electrolyte interphase

Here, using unique in-situ liquid secondary ion mass spectroscopy on isotope-labelled solid-electrolyte-interphase (SEI), assisted by cryogenic transmission electron microscopy and constrained ab initio molecular dynamics simulation, for the first time we answer the question regarding Li+ transport mechanism across SEI, and quantitatively determine the Li+-mobility therein. We unequivocally unveil that Li+ transport in SEI follows a mechanism of successive displacement, rather than "direct-hopping". We further reveal, in accordance with spatial-dependence of SEI structure across the thickness, the apparent Li+ self-diffusivity varies from 6.7*10-19 m2/s to 1.0*10-20 m2/s, setting a quantitative gauging of ionic transport behavior of SEI layer against the underlining electrode as well as the rate limiting step of battery operation. This direct study on Li+ kinetics in SEI fills part of the decade-long knowledge gap about the most important component in advanced batteries and provides more precise guidelines to the tailoring of interphasial chemistries for future battery chemistries.

cond-mat.mtrl-sci↗

arXiv2023

Generation and modulation of multiple 2D bulk photovoltaic effects in space-time reversal asymmetric 2H-FeCl2

The two-dimensional (2D) bulk photovoltaic effect (BPVE) is a cornerstone for future highly efficient 2D solar cells and optoelectronics. The ferromagnetic semiconductor 2H-FeCl2 is shown to realize a new type of BPVE in which spatial inversion (P), time reversal (T), and space-time reversal (PT) symmetries are broken (PT-broken). Using density functional theory and perturbation theory, we show that 2H-FeCl2 exhibits giant photocurrents, photo-spin-currents, and photo-orbital-currents under illumination by linearly polarized light. The injection-like and shift-like photocurrents coexist and propagate in different directions. The material also demonstrates substantial photoconductance, photo-spin-conductance, and photo-orbital-conductance, with magnitudes up to 4650 (nm{\cdot}μA/V2), 4620 (nm{\cdot}μA/V2 {\hbar}/2e), and 6450 (nm{\cdot}μA/V2 {\hbar}/e), respectively. Furthermore, the injection-currents, shift-spin-currents, and shift-orbital-currents can be readily switched via rotating the magnetizations of 2H-FeCl2. These results demonstrate the superior performance and intriguing control of a new type of BPVE in 2H-FeCl2.

cond-mat.mtrl-sci↗

arXiv2022

Group-wise Reinforcement Feature Generation for Optimal and Explainable Representation Space Reconstruction

Representation (feature) space is an environment where data points are vectorized, distances are computed, patterns are characterized, and geometric structures are embedded. Extracting a good representation space is critical to address the curse of dimensionality, improve model generalization, overcome data sparsity, and increase the availability of classic models. Existing literature, such as feature engineering and representation learning, is limited in achieving full automation (e.g., over heavy reliance on intensive labor and empirical experiences), explainable explicitness (e.g., traceable reconstruction process and explainable new features), and flexible optimal (e.g., optimal feature space reconstruction is not embedded into downstream tasks). Can we simultaneously address the automation, explicitness, and optimal challenges in representation space reconstruction for a machine learning task? To answer this question, we propose a group-wise reinforcement generation perspective. We reformulate representation space reconstruction into an interactive process of nested feature generation and selection, where feature generation is to generate new meaningful and explicit features, and feature selection is to eliminate redundant features to control feature sizes. We develop a cascading reinforcement learning method that leverages three cascading Markov Decision Processes to learn optimal generation policies to automate the selection of features and operations and the feature crossing. We design a group-wise generation strategy to cross a feature group, an operation, and another feature group to generate new features and find the strategy that can enhance exploration efficiency and augment reward signals of cascading agents. Finally, we present extensive experiments to demonstrate the effectiveness, efficiency, traceability, and explicitness of our system.

cs.LG↗

arXiv2022

Semi-supervised Drifted Stream Learning with Short Lookback

In many scenarios, 1) data streams are generated in real time; 2) labeled data are expensive and only limited labels are available in the beginning; 3) real-world data is not always i.i.d. and data drift over time gradually; 4) the storage of historical streams is limited and model updating can only be achieved based on a very short lookback window. This learning setting limits the applicability and availability of many Machine Learning (ML) algorithms. We generalize the learning task under such setting as a semi-supervised drifted stream learning with short lookback problem (SDSL). SDSL imposes two under-addressed challenges on existing methods in semi-supervised learning, continuous learning, and domain adaptation: 1) robust pseudo-labeling under gradual shifts and 2) anti-forgetting adaptation with short lookback. To tackle these challenges, we propose a principled and generic generation-replay framework to solve SDSL. The framework is able to accomplish: 1) robust pseudo-labeling in the generation step; 2) anti-forgetting adaption in the replay step. To achieve robust pseudo-labeling, we develop a novel pseudo-label classification model to leverage supervised knowledge of previously labeled data, unsupervised knowledge of new data, and, structure knowledge of invariant label semantics. To achieve adaptive anti-forgetting model replay, we propose to view the anti-forgetting adaptation task as a flat region search problem. We propose a novel minimax game-based replay objective function to solve the flat region search problem and develop an effective optimization solver. Finally, we present extensive experiments to demonstrate our framework can effectively address the task of anti-forgetting learning in drifted streams with short lookback.

cs.LG↗