arXiv subjects
Publications and source records attributed to Peiyao Li.
In this paper, we characterize compactness in terms of the Berezin transform and obtain an essential norm estimate for operators in the Toeplitz algebra on Bergman spaces of the Siegel upper half-space.
The increasing demand for deep learning in seismic interpretation has highlighted significant challenges, particularly the reliance on massive, labeled datasets and the inefficiency of training isolated models for individual tasks. To address these limitations, we introduce a unified, prompt-guided flow-matching framework (SeisDiff-intp) capable of executing multiple seismic interpretation tasks within a single model. By conditioning on varying prompts, the model dynamically switches between interpretation objectives without requiring structural modifications. Furthermore, to overcome the scarcity of labeled data for complex subsurface features, we propose an integrated generative augmentation strategy. By employing the flow matching setting, the framework can synthesize diverse and geologically realistic training pairs, specifically targeting structurally complex. Experimental results demonstrate that the proposed approach, coupled with generative augmentation, delivers high-quality, task-specific interpretations with stable and reproducible inference behavior. Ultimately, this approach provides a scalable, flexible, and robust alternative to single-task deep learning based seismic interpretation models.
We introduce SeisDiff-denoNIA, a diffusion-based seismic denoising framework that trains directly on noise extracted from field data, eliminating the dependence on synthetic datasets. Unlike conventional denoising methods that require clean signal labels, our approach uses field noise recorded as training targets, enabling the diffusion model to explicitly learn the true noise distribution. We further demonstrate the framework on a field DAS-VSP survey, where the model effectively suppresses multiple types of noise, including production, instrument, and environment related noise, while preserving key seismic events. By denoising shot gathers with SeisDiff-denoNIA prior to migration, the resulting images exhibit improved event continuity and enhanced reflection visibility without artificial artifacts. In addition, synthetic experiments show that the method substantially outperforms traditional signal-based diffusion models under low-SNR conditions. These results suggest that explicitly modeling noise is not only viable but advantageous for a broad class of seismic denoising tasks, particularly in challenging field environments.
We present NoLBERT, a lightweight, timestamped foundational language model for empirical research -- particularly for forecasting in economics, finance, and the social sciences. By pretraining exclusively on text from 1976 to 1995, NoLBERT avoids both lookback and lookahead biases (information leakage) that can undermine econometric inference. It exceeds domain-specific baselines on NLP benchmarks while maintaining temporal consistency. Applied to patent texts, NoLBERT enables the construction of firm-level innovation networks and shows that gains in innovation centrality predict higher long-run profit growth.
Tube waves present a significant challenge in vertical seismic profiling data, often obscuring critical seismic signals from seismic acquisition. In this study, we introduce the Seismic Diffusion Model for Denoising, a fast diffusion model specifically designed to remove the noise from seismic shotgather effectively. Our approach balances computational efficiency with high-quality image denoising, ensuring that the method is practical and robust for real-world applications. We validate the effectiveness of the proposed method through rigorous testing on both synthetic and field data, demonstrating its capability to preserve essential seismic signals while eliminating unwanted coherent noise. The results suggest that the proposed method enhances data quality and supports continuous production during seismic acquisition, paving the way for improved subsurface monitoring and analysis.
Inverse molecular design is critical in material science and drug discovery, where the generated molecules should satisfy certain desirable properties. In this paper, we propose equivariant energy-guided stochastic differential equations (EEGSDE), a flexible framework for controllable 3D molecule generation under the guidance of an energy function in diffusion models. Formally, we show that EEGSDE naturally exploits the geometric symmetry in 3D molecular conformation, as long as the energy function is invariant to orthogonal transformations. Empirically, under the guidance of designed energy functions, EEGSDE significantly improves the baseline on QM9, in inverse molecular design targeted to quantum properties and molecular structures. Furthermore, EEGSDE is able to generate molecules with multiple target properties by combining the corresponding energy functions linearly.
In the question answering(QA) task, multi-hop reasoning framework has been extensively studied in recent years to perform more efficient and interpretable answer reasoning on the Knowledge Graph(KG). However, multi-hop reasoning is inapplicable for answering n-ary fact questions due to its linear reasoning nature. We discover that there are two feasible improvements: 1) upgrade the basic reasoning unit from entity or relation to fact; and 2) upgrade the reasoning structure from chain to tree. Based on these, we propose a novel fact-tree reasoning framework, through transforming the question into a fact tree and performing iterative fact reasoning on it to predict the correct answer. Through a comprehensive evaluation on the n-ary fact KGQA dataset introduced by this work, we demonstrate that the proposed fact-tree reasoning framework has the desired advantage of high answer prediction accuracy. In addition, we also evaluate the fact-tree reasoning framework on two binary KGQA datasets and show that our approach also has a strong reasoning ability compared with several excellent baselines. This work has direct implications for exploring complex reasoning scenarios and provides a preliminary baseline approach.
Useful information (UI) is an elusive concept in neural networks. A quantitative measurement of UI is absent, despite the variations of UI can be recognized by prior knowledge. The communication bandwidth of feature maps decreases after downscaling operations, but UI flows smoothly after training due to lower Nyquist frequency. Inspired by the low-Nyqusit-frequency nature of UI, we propose the use of spectral roll-off points (SROPs) to estimate UI on variations. The computation of an SROP is extended from a 1-D signal to a 2-D image by the required rotation invariance in image classification tasks. SROP statistics across feature maps are implemented as layer-wise useful information estimates. We design sanity checks to explore SROP variations when UI variations are produced by variations in model input, model architecture and training stages. The variations of SROP is synchronizes with UI variations in various randomized and sufficiently trained model structures. Therefore, SROP variations is an accurate and convenient sign of UI variations, which promotes the explainability of data representations with respect to frequency-domain knowledge.
The novel coronavirus (SARS-CoV-2) which causes COVID-19 is an ongoing pandemic. There are ongoing studies with up to hundreds of publications uploaded to databases daily. We are exploring the use-case of artificial intelligence and natural language processing in order to efficiently sort through these publications. We demonstrate that clinical trial information, preclinical studies, and a general topic model can be used as text mining data intelligence tools for scientists all over the world to use as a resource for their own research. To evaluate our method, several metrics are used to measure the information extraction and clustering results. In addition, we demonstrate that our workflow not only have a use-case for COVID-19, but for other disease areas as well. Overall, our system aims to allow scientists to more efficiently research coronavirus. Our automatically updating modules are available on our information portal at https://ghddi-ailab.github.io/Targeting2019-nCoV/ for public viewing.
Background: Elderly patients with MODS have high risk of death and poor prognosis. The performance of current scoring systems assessing the severity of MODS and its mortality remains unsatisfactory. This study aims to develop an interpretable and generalizable model for early mortality prediction in elderly patients with MODS. Methods: The MIMIC-III, eICU-CRD and PLAGH-S databases were employed for model generation and evaluation. We used the eXtreme Gradient Boosting model with the SHapley Additive exPlanations method to conduct early and interpretable predictions of patients' hospital outcome. Three types of data source combinations and five typical evaluation indexes were adopted to develop a generalizable model. Findings: The interpretable model, with optimal performance developed by using MIMIC-III and eICU-CRD datasets, was separately validated in MIMIC-III, eICU-CRD and PLAGH-S datasets (no overlapping with training set). The performances of the model in predicting hospital mortality as validated by the three datasets were: AUC of 0.858, sensitivity of 0.834 and specificity of 0.705; AUC of 0.849, sensitivity of 0.763 and specificity of 0.784; and AUC of 0.838, sensitivity of 0.882 and specificity of 0.691, respectively. Comparisons of AUC between this model and baseline models with MIMIC-III dataset validation showed superior performances of this model; In addition, comparisons in AUC between this model and commonly used clinical scores showed significantly better performance of this model. Interpretation: The interpretable machine learning model developed in this study using fused datasets with large sample sizes was robust and generalizable. This model outperformed the baseline models and several clinical scores for early prediction of mortality in elderly ICU patients. The interpretative nature of this model provided clinicians with the ranking of mortality risk features.
Understanding the sleep quality and architecture is essential to human being's health, which is usually represented using multiple sleep stages. A standard sleep stage determination requires Electroencephalography (EEG) signals during the expensive and labor-intensive Polysomnography (PSG) test. To overcome this inconvenience, cardiorespiratory signals are proposed for the same purpose because of the easy and comfortable acquisition by simplified devices. In this paper, we leverage our low-cost wearable multi-sensor system to acquire the cardiorespiratory signals from subjects. Three novel features are designed during the feature extraction. We then apply a Bi-directional Recurrent Neural Network architecture with Long Short-term Memory (BLSTM) to predict the four-class sleep stages. Our prediction accuracy is 80.25% on a large public dataset (417 subjects), and 80.75% on our 32 enrolled subjects, respectively. Our results outperform the previous works which either used small data sets and had the potential over-fitting issues, or used the conventional machine learning methods on large data sets.