arXiv subjects

Siqi Li

Publications and source records attributed to Siqi Li.

At least 73 records · Page 4Linked to original sources

arXiv2025

FMC: Formalization of Natural Language Mathematical Competition Problems

Efficient and accurate autoformalization methods, which leverage large-scale datasets of extensive natural language mathematical problems to construct formal language datasets, are key to advancing formal mathematical reasoning. In this paper, we propose an autoformalization pipeline based on large language models with error feedback, achieving a fully automatic and training-free formalization approach. Using this pipeline, we curate an Olympiad-level dataset aligning natural language problems with Lean formalizations. The dataset comprises $3,922$ mathematical problems in natural language and $9,787$ in Lean, of which $64.46\%$ were assessed as at least above-average quality, making it suitable as a benchmark for automated theorem provers. Additionally, we investigate the formalization and reasoning capabilities of various LLMs and empirically demonstrate that few-shot learning, error feedback, and increasing sampling numbers enhance the autoformalization process. Experiments of three automated theorem provers on the \dataset\ dataset also highlight its challenging nature and its value as a benchmark for formal reasoning tasks.

cs.CL↗

arXiv2025

GDI-Bench: A Benchmark for General Document Intelligence with Vision and Reasoning Decoupling

The rapid advancement of multimodal large language models (MLLMs) has profoundly impacted the document domain, creating a wide array of application scenarios. This progress highlights the need for a comprehensive benchmark to evaluate these models' capabilities across various document-specific tasks. However, existing benchmarks often fail to locate specific model weaknesses or guide systematic improvements. To bridge this gap, we introduce a General Document Intelligence Benchmark (GDI-Bench), featuring 2.3k images across 9 key scenarios and 19 document-specific tasks. By decoupling visual complexity and reasoning complexity, the GDI-Bench structures graded tasks that allow performance assessment by difficulty, aiding in model weakness identification and optimization guidance. We evaluate various open-source and closed-source models on GDI-Bench, conducting decoupled analyses in the visual and reasoning domains, revealing their strengths and weaknesses. To address the diverse tasks and domains in the GDI-Bench, we propose a GDI-Model that mitigates catastrophic forgetting during the supervised fine-tuning (SFT) process through an intelligence-preserving training strategy, thereby reinforcing the inherent weaknesses of the base model. Our model achieves state-of-the-art performance on previous benchmarks and the GDI-Bench. Both our benchmark and models are or will be open-sourced on https://huggingface.co/GDIBench.

cs.CL↗

arXiv2025

Toward Fair Federated Learning under Demographic Disparities and Data Imbalance

Ensuring fairness is critical when applying artificial intelligence to high-stakes domains such as healthcare, where predictive models trained on imbalanced and demographically skewed data risk exacerbating existing disparities. Federated learning (FL) enables privacy-preserving collaboration across institutions, but remains vulnerable to both algorithmic bias and subgroup imbalance - particularly when multiple sensitive attributes intersect. We propose FedIDA (Fed erated Learning for Imbalance and D isparity A wareness), a framework-agnostic method that combines fairness-aware regularization with group-conditional oversampling. FedIDA supports multiple sensitive attributes and heterogeneous data distributions without altering the convergence behavior of the underlying FL algorithm. We provide theoretical analysis establishing fairness improvement bounds using Lipschitz continuity and concentration inequalities, and show that FedIDA reduces the variance of fairness metrics across test sets. Empirical results on both benchmark and real-world clinical datasets confirm that FedIDA consistently improves fairness while maintaining competitive predictive performance, demonstrating its effectiveness for equitable and privacy-preserving modeling in healthcare. The source code is available on GitHub.

cs.CY↗

arXiv2025

Order within disorder: spectral key generation and distribution in random lasers

In secure communication, highly random entropy sources are essential for information security. Random lasers (RLs), which arise from multiple scattering in disordered structures, are potentially ideal entropy sources. Traditionally, RLs are viewed as disordered and unpredictable. However, in this work, we present novel evidence that orderly patterns exist beneath the seemingly disordered outputs of RLs. Utilizing deep learning techniques, a variety of advanced neural network models are used to analyze the spectral data in multiple dimensions. The results show that the time series of RLs spectra are unpredictable, but spectral wavelength component intensities can be recovered due to inter-modal correlations. This finding not only breaks through the traditional perception that RLs are unpredictable, but also reveals for the first time that RLs have the dual characteristics of both randomness and determinism. Based on this new characteristic, we further expand the application field of RLs and innovatively design a new type of key generation and distribution scheme. In this scheme, the disordered property of RLs is used for key generation to ensure high randomness, while their ordered property is used for key distribution to guarantee accuracy and reliability. The scheme provides a new strategy for secure communication.

physics.optics↗

arXiv2025

RealVVT: Towards Photorealistic Video Virtual Try-on via Spatio-Temporal Consistency

Virtual try-on has emerged as a pivotal task at the intersection of computer vision and fashion, aimed at digitally simulating how clothing items fit on the human body. Despite notable progress in single-image virtual try-on (VTO), current methodologies often struggle to preserve a consistent and authentic appearance of clothing across extended video sequences. This challenge arises from the complexities of capturing dynamic human pose and maintaining target clothing characteristics. We leverage pre-existing video foundation models to introduce RealVVT, a photoRealistic Video Virtual Try-on framework tailored to bolster stability and realism within dynamic video contexts. Our methodology encompasses a Clothing & Temporal Consistency strategy, an Agnostic-guided Attention Focus Loss mechanism to ensure spatial consistency, and a Pose-guided Long Video VTO technique adept at handling extended video sequences.Extensive experiments across various datasets confirms that our approach outperforms existing state-of-the-art models in both single-image and video VTO tasks, offering a viable solution for practical applications within the realms of fashion e-commerce and virtual fitting environments.

cs.CV↗

arXiv2025

Chameleon: Fast-slow Neuro-symbolic Lane Topology Extraction

Lane topology extraction involves detecting lanes and traffic elements and determining their relationships, a key perception task for mapless autonomous driving. This task requires complex reasoning, such as determining whether it is possible to turn left into a specific lane. To address this challenge, we introduce neuro-symbolic methods powered by vision-language foundation models (VLMs). Existing approaches have notable limitations: (1) Dense visual prompting with VLMs can achieve strong performance but is costly in terms of both financial resources and carbon footprint, making it impractical for robotics applications. (2) Neuro-symbolic reasoning methods for 3D scene understanding fail to integrate visual inputs when synthesizing programs, making them ineffective in handling complex corner cases. To this end, we propose a fast-slow neuro-symbolic lane topology extraction algorithm, named Chameleon, which alternates between a fast system that directly reasons over detected instances using synthesized programs and a slow system that utilizes a VLM with a chain-of-thought design to handle corner cases. Chameleon leverages the strengths of both approaches, providing an affordable solution while maintaining high performance. We evaluate the method on the OpenLane-V2 dataset, showing consistent improvements across various baseline detectors. Our code, data, and models are publicly available at https://github.com/XR-Lee/neural-symbolic

cs.CV↗

arXiv2025

ERetinex: Event Camera Meets Retinex Theory for Low-Light Image Enhancement

Low-light image enhancement aims to restore the under-exposure image captured in dark scenarios. Under such scenarios, traditional frame-based cameras may fail to capture the structure and color information due to the exposure time limitation. Event cameras are bio-inspired vision sensors that respond to pixel-wise brightness changes asynchronously. Event cameras' high dynamic range is pivotal for visual perception in extreme low-light scenarios, surpassing traditional cameras and enabling applications in challenging dark environments. In this paper, inspired by the success of the retinex theory for traditional frame-based low-light image restoration, we introduce the first methods that combine the retinex theory with event cameras and propose a novel retinex-based low-light image restoration framework named ERetinex. Among our contributions, the first is developing a new approach that leverages the high temporal resolution data from event cameras with traditional image information to estimate scene illumination accurately. This method outperforms traditional image-only techniques, especially in low-light environments, by providing more precise lighting information. Additionally, we propose an effective fusion strategy that combines the high dynamic range data from event cameras with the color information of traditional images to enhance image quality. Through this fusion, we can generate clearer and more detail-rich images, maintaining the integrity of visual information even under extreme lighting conditions. The experimental results indicate that our proposed method outperforms state-of-the-art (SOTA) methods, achieving a gain of 1.0613 dB in PSNR while reducing FLOPS by \textbf{84.28}\%.

cs.CV↗

arXiv2025

Self-Supervised Learning for Solar Radio Spectrum Classification

Solar radio observation is an important way to study the Sun. Solar radio bursts contain important information about solar activity. Therefore, real-time automatic detection and classification of solar radio bursts are of great value for subsequent solar physics research and space weather warnings. Traditional image classification methods based on deep learning often require consid-erable training data. To address insufficient solar radio spectrum images, transfer learning is generally used. However, the large difference between natural images and solar spectrum images has a large impact on the transfer learning effect. In this paper, we propose a self-supervised learning method for solar radio spectrum classification. Our method uses self-supervised training with a self-masking approach in natural language processing. Self-supervised learning is more conducive to learning the essential information about images compared with supervised methods, and it is more suitable for transfer learning. First, the method pre-trains using a large amount of other existing data. Then, the trained model is fine-tuned on the solar radio spectrum dataset. Experiments show that the method achieves a classification accuracy similar to that of convolutional neural networks and Transformer networks with supervised training.

astro-ph.IM↗

arXiv2025

Classification of Solar Radio Spectrum Based on Swin Transformer

Solar radio observation is a method used to study the Sun. It is very important for space weather early warning and solar physics research to automatically classify solar radio spectrums in real time and judge whether there is a solar radio burst. As the number of solar radio burst spectrums is small and uneven, this paper proposes a classification method for solar radio spectrums based on the Swin transformer. First, the method transfers the parameters of the pretrained model to the Swin transformer model. Then, the hidden layer weights of the Swin transformer are frozen, and the fully connected layer of the Swin transformer is trained on the target dataset. Finally, pa-rameter tuning is performed. The experimental results show that the method can achieve a true positive rate of 100%, which is more accurate than previous methods. Moreover, the number of our model parameters is only 20 million, which is 80% lower than that of the traditional VGG16 con-volutional neural network with more than 130 million parameters.

astro-ph.IM↗

arXiv2024

UME: Upcycling Mixture-of-Experts for Scalable and Efficient Automatic Speech Recognition

Recent advancements in scaling up models have significantly improved performance in Automatic Speech Recognition (ASR) tasks. However, training large ASR models from scratch remains costly. To address this issue, we introduce UME, a novel method that efficiently Upcycles pretrained dense ASR checkpoints into larger Mixture-of-Experts (MoE) architectures. Initially, feed-forward networks are converted into MoE layers. By reusing the pretrained weights, we establish a robust foundation for the expanded model, significantly reducing optimization time. Then, layer freezing and expert balancing strategies are employed to continue training the model, further enhancing performance. Experiments on a mixture of 170k-hour Mandarin and English datasets show that UME: 1) surpasses the pretrained baseline by a margin of 11.9% relative error rate reduction while maintaining comparable latency; 2) reduces training time by up to 86.7% and achieves superior accuracy compared to training models of the same size from scratch.

eess.AS↗

arXiv2024

Locate n' Rotate: Two-stage Openable Part Detection with Foundation Model Priors

Detecting the openable parts of articulated objects is crucial for downstream applications in intelligent robotics, such as pulling a drawer. This task poses a multitasking challenge due to the necessity of understanding object categories and motion. Most existing methods are either category-specific or trained on specific datasets, lacking generalization to unseen environments and objects. In this paper, we propose a Transformer-based Openable Part Detection (OPD) framework named Multi-feature Openable Part Detection (MOPD) that incorporates perceptual grouping and geometric priors, outperforming previous methods in performance. In the first stage of the framework, we introduce a perceptual grouping feature model that provides perceptual grouping feature priors for openable part detection, enhancing detection results through a cross-attention mechanism. In the second stage, a geometric understanding feature model offers geometric feature priors for predicting motion parameters. Compared to existing methods, our proposed approach shows better performance in both detection and motion parameter prediction. Codes and models are publicly available at https://github.com/lisiqi-zju/MOPD

cs.CV↗

arXiv2024

Feasibility of PET-enabled dual-energy CT imaging: First physical phantom and initial patient results

X-ray computed tomography (CT) in PET/CT is commonly operated with a single energy, resulting in a limitation of lacking tissue composition information. Dual-energy (DE) spectral CT enables material decomposition by using two different x-ray energies and may be combined with PET for improved multimodality imaging, but would either require hardware upgrade or increase radiation dose due to the added second x-ray CT scan. Recently proposed PET-enabled DECT method allows dual-energy spectral imaging using a conventional PET/CT scanner without the need for a second x-ray CT scan. A gamma-ray CT (gCT) image at 511 keV can be generated from the existing time-of-flight PET data with the maximum-likelihood attenuation and activity (MLAA) approach and is then combined with the low-energy x-ray CT image to form dual-energy spectral imaging. To improve the image quality of gCT, a kernel MLAA method was further proposed by incorporating x-ray CT as a priori information. The concept of this PET-enabled DECT has been validated using simulation studies, but not yet with 3D real data. In this work, we developed a general open-source implementation for gCT reconstruction from PET data and use this implementation for the first real data validation with both a physical phantom study and a human subject study on a uEXPLORER total-body PET/CT system. These results have demonstrated the feasibility of this method for spectral imaging and material decomposition.

physics.med-ph↗

arXiv2024

Attosecond Coherent Electron Motion in a Photoionized Aromatic Molecule

In molecular systems, the ultrafast motion of electrons initiates the process of chemical change. Tracking this electronic motion across molecules requires coupling attosecond time resolution to atomic-scale spatial sensitivity. In this work, we employ a pair of attosecond x-ray pulses from an x-ray free-electron laser to follow electron motion resulting from the sudden removal of an electron from a prototypical aromatic system, para-aminophenol. X-ray absorption enables tracking this motion with atomic-site specificity. Our measurements are compared with state-of-the-art computational modeling, reproducing the observed response across multiple timescales. Sub-femtosecond dynamics are assigned to states undergoing non-radiative decay, while few-femtosecond oscillatory motion is associated with electronic wavepacket motion in stable cation states, that will eventually couple to nuclear motion. Our work provides insight on the ultrafast charge motion preceding and initiating chemical transformations in moderately complex systems, and provides a powerful benchmark for computational models of ultrafast charge motion in matter.

physics.chem-ph↗

arXiv2024

FairFML: Fair Federated Machine Learning with a Case Study on Reducing Gender Disparities in Cardiac Arrest Outcome Prediction

Objective: Mitigating algorithmic disparities is a critical challenge in healthcare research, where ensuring equity and fairness is paramount. While large-scale healthcare data exist across multiple institutions, cross-institutional collaborations often face privacy constraints, highlighting the need for privacy-preserving solutions that also promote fairness. Materials and Methods: In this study, we present Fair Federated Machine Learning (FairFML), a model-agnostic solution designed to reduce algorithmic bias in cross-institutional healthcare collaborations while preserving patient privacy. As a proof of concept, we validated FairFML using a real-world clinical case study focused on reducing gender disparities in cardiac arrest outcome prediction. Results: We demonstrate that the proposed FairFML framework enhances fairness in federated learning (FL) models without compromising predictive performance. Our findings show that FairFML improves model fairness by up to 65% compared to the centralized model, while maintaining performance comparable to both local and centralized models, as measured by receiver operating characteristic analysis. Discussion and Conclusion: FairFML offers a promising and flexible solution for FL collaborations, with its adaptability allowing seamless integration with various FL frameworks and models, from traditional statistical methods to deep learning techniques. This makes FairFML a robust approach for developing fairer FL models across diverse clinical and biomedical applications.

cs.CY↗

arXiv2024

Optimizing Rare Word Accuracy in Direct Speech Translation with a Retrieval-and-Demonstration Approach

Direct speech translation (ST) models often struggle with rare words. Incorrect translation of these words can have severe consequences, impacting translation quality and user trust. While rare word translation is inherently challenging for neural models due to sparse learning signals, real-world scenarios often allow access to translations of past recordings on similar topics. To leverage these valuable resources, we propose a retrieval-and-demonstration approach to enhance rare word translation accuracy in direct ST models. First, we adapt existing ST models to incorporate retrieved examples for rare word translation, which allows the model to benefit from prepended examples, similar to in-context learning. We then develop a cross-modal (speech-to-speech, speech-to-text, text-to-text) retriever to locate suitable examples. We demonstrate that standard ST models can be effectively adapted to leverage examples for rare word translation, improving rare word translation accuracy over the baseline by 17.6% with gold examples and 8.5% with retrieved examples. Moreover, our speech-to-speech retrieval approach outperforms other modalities and exhibits higher robustness to unseen speakers. Our code is publicly available (https://github.com/SiqiLii/Retrieve-and-Demonstration-ST).

cs.CL↗

arXiv2024

A Stein characterisation of the distribution of the product of correlated normal random variables

We obtain a Stein characterisation of the distribution of the product of two correlated normal random variables with non-zero means, and more generally the distribution of the sum of independent copies of such random variables. Our Stein characterisation is shown to naturally generalise a number of other Stein characterisations in the literature. From our Stein characterisation we derive recursive formulas for the moments of the product of two correlated normal random variables, and more generally the sum of independent copies of such random variables, which allows for efficient computation of higher order moments.

math.PR↗

arXiv2024

A Hybrid RAG System with Comprehensive Enhancement on Complex Reasoning

Retrieval-augmented generation (RAG) is a framework enabling large language models (LLMs) to enhance their accuracy and reduce hallucinations by integrating external knowledge bases. In this paper, we introduce a hybrid RAG system enhanced through a comprehensive suite of optimizations that significantly improve retrieval quality, augment reasoning capabilities, and refine numerical computation ability. We refined the text chunks and tables in web pages, added attribute predictors to reduce hallucinations, conducted LLM Knowledge Extractor and Knowledge Graph Extractor, and finally built a reasoning strategy with all the references. We evaluated our system on the CRAG dataset through the Meta CRAG KDD Cup 2024 Competition. Both the local and online evaluations demonstrate that our system significantly enhances complex reasoning capabilities. In local evaluations, we have significantly improved accuracy and reduced error rates compared to the baseline model, achieving a notable increase in scores. In the meanwhile, we have attained outstanding results in online assessments, demonstrating the performance and generalization capabilities of the proposed system. The source code for our system is released in \url{https://gitlab.aicrowd.com/shizueyy/crag-new}.

cs.CL↗

arXiv2024

PreAfford: Universal Affordance-Based Pre-Grasping for Diverse Objects and Environments

Robotic manipulation with two-finger grippers is challenged by objects lacking distinct graspable features. Traditional pre-grasping methods, which typically involve repositioning objects or utilizing external aids like table edges, are limited in their adaptability across different object categories and environments. To overcome these limitations, we introduce PreAfford, a novel pre-grasping planning framework incorporating a point-level affordance representation and a relay training approach. Our method significantly improves adaptability, allowing effective manipulation across a wide range of environments and object types. When evaluated on the ShapeNet-v2 dataset, PreAfford not only enhances grasping success rates by 69% but also demonstrates its practicality through successful real-world experiments. These improvements highlight PreAfford's potential to redefine standards for robotic handling of complex manipulation tasks in diverse settings.

cs.RO↗