Searcharxiv⌕ Search

arXiv subjects

Xinyi Li

Publications and source records attributed to Xinyi Li.

At least 127 records · Page 7Linked to original sources
arXiv2023

MPGemmFI: A Fault Injection Technique for Mixed Precision GEMM in ML Applications

Emerging deep learning workloads urgently need fast general matrix multiplication (GEMM). To meet such demand, one of the critical features of machine-learning-specific accelerators such as NVIDIA Tensor Cores, AMD Matrix Cores, and Google TPUs is the support of mixed-precision enabled GEMM. For DNN models, lower-precision FP data formats and computation offer acceptable correctness but significant performance, area, and memory footprint improvement. While promising, the mixed-precision computation on error resilience remains unexplored. To this end, we develop a fault injection framework that systematically injects fault into the mixed-precision computation results. We investigate how the faults affect the accuracy of machine learning applications. Based on the error resilience characteristics, we offer lightweight error detection and correction solutions that significantly improve the overall model accuracy if the models experience hardware faults. The solutions can be efficiently integrated into the accelerator's pipelines.

cs.DC↗
arXiv2023

Transformation Decoupling Strategy based on Screw Theory for Deterministic Point Cloud Registration with Gravity Prior

Point cloud registration is challenging in the presence of heavy outlier correspondences. This paper focuses on addressing the robust correspondence-based registration problem with gravity prior that often arises in practice. The gravity directions are typically obtained by inertial measurement units (IMUs) and can reduce the degree of freedom (DOF) of rotation from 3 to 1. We propose a novel transformation decoupling strategy by leveraging screw theory. This strategy decomposes the original 4-DOF problem into three sub-problems with 1-DOF, 2-DOF, and 1-DOF, respectively, thereby enhancing the computation efficiency. Specifically, the first 1-DOF represents the translation along the rotation axis and we propose an interval stabbing-based method to solve it. The second 2-DOF represents the pole which is an auxiliary variable in screw theory and we utilize a branch-and-bound method to solve it. The last 1-DOF represents the rotation angle and we propose a global voting method for its estimation. The proposed method sequentially solves three consensus maximization sub-problems, leading to efficient and deterministic registration. In particular, it can even handle the correspondence-free registration problem due to its significant robustness. Extensive experiments on both synthetic and real-world datasets demonstrate that our method is more efficient and robust than state-of-the-art methods, even when dealing with outlier rates exceeding 99%.

cs.CV↗
arXiv2023

Efficient Fully Bayesian Approach to Brain Activity Mapping with Complex-Valued fMRI Data

Functional magnetic resonance imaging (fMRI) enables indirect detection of brain activity changes via the blood-oxygen-level-dependent (BOLD) signal. Conventional analysis methods mainly rely on the real-valued magnitude of these signals. In contrast, research suggests that analyzing both real and imaginary components of the complex-valued fMRI (cv-fMRI) signal provides a more holistic approach that can increase power to detect neuronal activation. We propose a fully Bayesian model for brain activity mapping with cv-fMRI data. Our model accommodates temporal and spatial dynamics. Additionally, we propose a computationally efficient sampling algorithm, which enhances processing speed through image partitioning. Our approach is shown to be computationally efficient via image partitioning and parallel computation while being competitive with state-of-the-art methods. We support these claims with both simulated numerical studies and an application to real cv-fMRI data obtained from a finger-tapping experiment.

stat.ME↗
arXiv2023

Scaling limit of the occupation measure of random walk cut points

We consider the occupation measure of the cut points of a simple random walk on a $d$-dimensional cubic lattice for $d = 2, 3$, and we show that the scaling limit of the occupation measure in weak topology is the natural fractal measure on the Brownian cut points defined via its Minkowski content.

math.PR↗
arXiv2023

Probing the Galactic halo with RR Lyrae stars -- IV. On the Oosterhoff dichotomy of RR Lyrae stars

We use 3653 (2661 RRab, 992 RRc) RR Lyrae stars (RRLs) with 7D (3D position, 3D velocity, and metallicity) information selected from SDSS, LAMOST, and Gaia EDR3, and divide the sample into two Oosterhoff groups (Oo I and Oo II) according to their amplitude-period behaviour in the Bailey Diagram. We present a comparative study of these two groups based on chemistry, kinematics, and dynamics. We find that Oo I RRLs are relatively more metal rich, with predominately radially dominated orbits and large eccentricities, while Oo II RRLs are relatively more metal poor, and have mildly radially dominated orbits. The Oosterhoff dichotomy of the Milky Way's halo is more apparent for the inner-halo region than for the outer-halo region. Additionally, we also search for this phenomenon in the halos of the two largest satellite galaxies, the Large and Small Magellanic clouds (LMC, SMC), and compare over different bins in metallicity. We find that the Oosterhoff dichotomy is not immutable, and varies based on position in the Galaxy and from galaxy-to-galaxy. We conclude that the Oosterhoff dichotomy is the result of a combination of stellar and galactic evolution, and that it is much more complex than the dichotomy originally identified in Galactic globular clusters.

astro-ph.GA↗
arXiv2023

Continual Learning for Abdominal Multi-Organ and Tumor Segmentation

The ability to dynamically extend a model to new data and classes is critical for multiple organ and tumor segmentation. However, due to privacy regulations, accessing previous data and annotations can be problematic in the medical domain. This poses a significant barrier to preserving the high segmentation accuracy of the old classes when learning from new classes because of the catastrophic forgetting problem. In this paper, we first empirically demonstrate that simply using high-quality pseudo labels can fairly mitigate this problem in the setting of organ segmentation. Furthermore, we put forward an innovative architecture designed specifically for continuous organ and tumor segmentation, which incurs minimal computational overhead. Our proposed design involves replacing the conventional output layer with a suite of lightweight, class-specific heads, thereby offering the flexibility to accommodate newly emerging classes. These heads enable independent predictions for newly introduced and previously learned classes, effectively minimizing the impact of new classes on old ones during the course of continual learning. We further propose incorporating Contrastive Language-Image Pretraining (CLIP) embeddings into the organ-specific heads. These embeddings encapsulate the semantic information of each class, informed by extensive image-text co-training. The proposed method is evaluated on both in-house and public abdominal CT datasets under organ and tumor segmentation tasks. Empirical results suggest that the proposed design improves the segmentation performance of a baseline neural network on newly-introduced and previously-learned classes along the learning trajectory.

eess.IV↗
arXiv2023

A Preliminary Study of ChatGPT on News Recommendation: Personalization, Provider Fairness, Fake News

Online news platforms commonly employ personalized news recommendation methods to assist users in discovering interesting articles, and many previous works have utilized language model techniques to capture user interests and understand news content. With the emergence of large language models like GPT-3 and T-5, a new recommendation paradigm has emerged, leveraging pre-trained language models for making recommendations. ChatGPT, with its user-friendly interface and growing popularity, has become a prominent choice for text-based tasks. Considering the growing reliance on ChatGPT for language tasks, the importance of news recommendation in addressing social issues, and the trend of using language models in recommendations, this study conducts an initial investigation of ChatGPT's performance in news recommendations, focusing on three perspectives: personalized news recommendation, news provider fairness, and fake news detection. ChatGPT has the limitation that its output is sensitive to the input phrasing. We therefore aim to explore the constraints present in the generated responses of ChatGPT for each perspective. Additionally, we investigate whether specific prompt formats can alleviate these constraints or if these limitations require further attention from researchers in the future. We also surpass fixed evaluations by developing a webpage to monitor ChatGPT's performance on weekly basis on the tasks and prompts we investigated. Our aim is to contribute to and encourage more researchers to engage in the study of enhancing news recommendation performance through the utilization of large language models such as ChatGPT.

cs.IR↗
arXiv2023

Functional Individualized Treatment Regimes with Imaging Features

Precision medicine seeks to discover an optimal personalized treatment plan and thereby provide informed and principled decision support, based on the characteristics of individual patients. With recent advancements in medical imaging, it is crucial to incorporate patient-specific imaging features in the study of individualized treatment regimes. We propose a novel, data-driven method to construct interpretable image features which can be incorporated, along with other features, to guide optimal treatment regimes. The proposed method treats imaging information as a realization of a stochastic process, and employs smoothing techniques in estimation. We show that the proposed estimators are consistent under mild conditions. The proposed method is applied to a dataset provided by the Alzheimer's Disease Neuroimaging Initiative.

stat.ME↗
arXiv2023

PBNR: Prompt-based News Recommender System

Online news platforms often use personalized news recommendation methods to help users discover articles that align with their interests. These methods typically predict a matching score between a user and a candidate article to reflect the user's preference for the article. Some previous works have used language model techniques, such as the attention mechanism, to capture users' interests based on their past behaviors, and to understand the content of articles. However, these existing model architectures require adjustments if additional information is taken into account. Pre-trained large language models, which can better capture word relationships and comprehend contexts, have seen a significant development in recent years, and these pre-trained models have the advantages of transfer learning and reducing the training time for downstream tasks. Meanwhile, prompt learning is a newly developed technique that leverages pre-trained language models by building task-specific guidance for output generations. To leverage textual information in news articles, this paper introduces the pre-trained large language model and prompt-learning to the community of news recommendation. The proposed model "prompt-based news recommendation" (PBNR) treats the personalized news recommendation as a text-to-text language task and designs personalized prompts to adapt to the pre-trained language model -- text-to-text transfer transformer (T5). Experimental studies using the Microsoft News dataset show that PBNR is capable of making accurate recommendations by taking into account various lengths of past behaviors of different users. PBNR can also easily adapt to new information without changing the model architecture and the training objective. Additionally, PBNR can make recommendations based on users' specific requirements, allowing human-computer interaction in the news recommendation field.

cs.IR↗
arXiv2023

Company Competition Graph

Financial market participants frequently rely on numerous business relationships to make investment decisions. Investors can learn about potential risks and opportunities associated with other connected entities through these corporate connections. Nonetheless, human annotation of a large corpus to extract such relationships is highly time-consuming, not to mention that it requires a considerable amount of industry expertise and professional training. Meanwhile, we have yet to observe means to generate reliable knowledge graphs of corporate relationships due to the lack of impartial and granular data sources. This study proposes a system to process financial reports and construct the public competitor graph to fill the void. Our method can retrieve more than 83\% competition relationship of the S\&P 500 index companies. Based on the output from our system, we construct a knowledge graph with more than 700 nodes and 1200 edges. A demo interactive graph interface is available.

q-fin.GN↗
arXiv2023

CoPracTter: Toward Integrating Personalized Practice Scenarios, Timely Feedback and Social Support into An Online Support Tool for Coping with Stuttering in China

Stuttering is a speech disorder influencing over 70 million people worldwide, including 13 million in China. It causes low self-esteem among other detrimental effects on people who stutter (PwS). Although prior work has explored approaches to assist PwS, they primarily focused on western contexts. In our formative study, we found unique practices and challenges among Chinese PwS. We then iteratively designed an online tool, CoPracTter, to support Chinese PwS practicing speaking fluency with 1) targeted stress-inducing practice scenarios, 2) real-time speech indicators, and 3) personalized timely feedback from the community. We further conducted a seven-day deployment study (N=11) to understand how participants utilized these key features. To our knowledge, it is the first time such a prototype was designed and tested for a long time with multiple PwS participants online simultaneously. Results indicate that personalized practice with targeted scenarios and timely feedback from a supportive community assisted PwS in speaking fluently, staying positive, and facing similar real-life circumstances.

cs.HC↗
arXiv2023

Nonparametric Regression for 3D Point Cloud Learning

Over the past two decades, we have seen an exponentially increased amount of point clouds collected with irregular shapes in various areas. Motivated by the importance of solid modeling for point clouds, we develop a novel and efficient smoothing tool based on multivariate splines over the tetrahedral partitions to extract the underlying signal and build up a 3D solid model from the point cloud. The proposed smoothing method can denoise or deblur the point cloud effectively and provide a multi-resolution reconstruction of the actual signal. In addition, it can handle sparse and irregularly distributed point clouds and recover the underlying trajectory. The proposed smoothing and interpolation method also provides a natural way of numerosity data reduction. Furthermore, we establish the theoretical guarantees of the proposed method. Specifically, we derive the convergence rate and asymptotic normality of the proposed estimator and illustrate that the convergence rate achieves the optimal nonparametric convergence rate. Through extensive simulation studies and a real data example, we demonstrate the superiority of the proposed method over traditional smoothing methods in terms of estimation accuracy and efficiency of data reduction.

stat.CO↗
arXiv2022

The Circular Velocity Curve of the Milky Way from 5 to 25 kpc using luminous red giant branch star

We present a sample of 254,882 luminous red giant branch (LRGB) stars selected from the APOGEE and LAMOST surveys. By combining photometric and astrometric information from the 2MASS and Gaia surveys, the precise distances of the sample stars are determined by a supervised machine learning algorithm: the gradient boosted decision trees. To test the accuracy of the derived distances, member stars of globular clusters (GCs) and open clusters (OCs) are used. The tests by cluster member stars show a precision of about 10 per cent with negligible zero-point offsets, for the derived distances of our sample stars. The final sample covers a large volume of the Galactic disk(s) and halo of $0<R<30$ kpc and $|Z|\leqslant15$ kpc. The rotation curve (RC) of the Milky Way across radius of $5\lesssim R\lesssim25$ kpc have been accurately measured with $\sim$ 54,000 stars of the thin disk population selected from the LRGB sample. The derived RC shows a weak decline along $R$ with a gradient of $-1.83\pm0.02$ $({\rm stat.}) \pm 0.07$ $({\rm sys.})$ km s$^{-1}$ kpc$^{-1}$, in excellent agreement with the results measured by previous studies. The circular velocity at the solar position, yielded by our RC, is $234.04\pm0.08$ $({\rm stat.}) \pm 1.36$ $({\rm sys.})$ km s$^{-1}$, again in great consistent with other independent determinations. From the newly constructed RC, as well as constraints from other data, we have constructed a mass model for our Galaxy, yielding a mass of the dark matter halo of $M_{\rm{200}}$ = ($8.05\pm1.15$)$\times$10$^{11} \rm{M_\odot}$ with a corresponding radius of $R_{\rm{200}}$ = $192.37\pm9.24$ kpc and a local dark matter density of $0.39\pm0.03$ GeV cm$^{-3}$.

astro-ph.GA↗
arXiv2022

Sharp asymptotics for arm probabilities in critical planar percolation

In this work, we consider critical planar site percolation on the triangular lattice and derive sharp estimates on the asymptotics of the probability of half-plane $j$-arm events for $j \geq 1$ and planar (polychromatic) $j$-arm events for $j>1$. These estimates greatly improve previous results and in particular answer (a large part of) a question of Schramm (ICM Proc., 2006).

math.PR↗
arXiv2022

On large deviations and intersection of random interlacements

We investigate random interlacements on $\mathbb{Z}^d$ with $d \geq 3$, and derive the large deviation rate for the probability that the capacity of the interlacement set in a macroscopic box is much smaller than that of the box. As an application, we obtain the large deviation rate for the probability that two independent interlacements have empty intersections in a macroscopic box. We also prove that conditioning on this event, one of them will be sparse in the box in terms of capacity. This result is an example of the entropic repulsion phenomenon for random interlacements.

math.PR↗
arXiv2022

On Natural Measures of SLE- and CLE-Related Random Fractals

In this paper, we construct and then prove the up-to constants uniqueness of the natural measure on several random fractals, namely the SLE cut points, SLE boundary touching points, CLE pivotal points and the CLE carpet/gasket. As an application, we also show the equivalence between our natural measures defined in this paper (i.e. CLE pivotal and gasket measures) and their discrete analogs of counting measures in critical continuum planar Bernoulli percolation in [Garban-Pete-Schramm, J. Amer. Math. Soc.,2013]. Although the existence and uniqueness for the natural measure for CLE carpet/gasket have already been proved in [Miller-Schoug, arXiv:2201.01748], in this paper we provide with a different argument via the coupling of CLE and LQG.

math.PR↗
arXiv2022

Discussion of Multiscale Fisher's Independence Test for Multivariate Dependence

The multiscale Fisher's independence test (MULTIFIT hereafter) proposed by Gorsky & Ma (2022) is a novel method to test independence between two random vectors. By its design, this test is particularly useful in detecting local dependence. Moreover, by adopting a resampling-free approach, it can easily accommodate massive sample sizes. Another benefit of the proposed method is its ability to interpret the nature of dependency. We congratulate the authors, Shai Gorksy and Li Ma, for their very interesting and elegant work. In this comment, we would like to discuss a general framework unifying the MULTIFIT and other tests and compare it with the binary expansion randomized ensemble test (BERET hereafter) proposed by Lee et al. (In press). We also would like to contribute our thoughts on potential extensions of the method.

math.ST↗