arXiv subjects

Can Yang

Publications and source records attributed to Can Yang.

At least 37 records · Page 2Linked to original sources

arXiv2025

Distilled Prompt Learning for Incomplete Multimodal Survival Prediction

The integration of multimodal data including pathology images and gene profiles is widely applied in precise survival prediction. Despite recent advances in multimodal survival models, collecting complete modalities for multimodal fusion still poses a significant challenge, hindering their application in clinical settings. Current approaches tackling incomplete modalities often fall short, as they typically compensate for only a limited part of the knowledge of missing modalities. To address this issue, we propose a Distilled Prompt Learning framework (DisPro) to utilize the strong robustness of Large Language Models (LLMs) to missing modalities, which employs two-stage prompting for compensation of comprehensive information for missing modalities. In the first stage, Unimodal Prompting (UniPro) distills the knowledge distribution of each modality, preparing for supplementing modality-specific knowledge of the missing modality in the subsequent stage. In the second stage, Multimodal Prompting (MultiPro) leverages available modalities as prompts for LLMs to infer the missing modality, which provides modality-common information. Simultaneously, the unimodal knowledge acquired in the first stage is injected into multimodal inference to compensate for the modality-specific knowledge of the missing modality. Extensive experiments covering various missing scenarios demonstrated the superiority of the proposed method. The code is available at https://github.com/Innse/DisPro.

cs.LG↗

arXiv2025

Can LLMs Solve longer Math Word Problems Better?

Math Word Problems (MWPs) play a vital role in assessing the capabilities of Large Language Models (LLMs), yet current research primarily focuses on questions with concise contexts. The impact of longer contexts on mathematical reasoning remains under-explored. This study pioneers the investigation of Context Length Generalizability (CoLeG), which refers to the ability of LLMs to solve MWPs with extended narratives. We introduce Extended Grade-School Math (E-GSM), a collection of MWPs featuring lengthy narratives, and propose two novel metrics to evaluate the efficacy and resilience of LLMs in tackling these problems. Our analysis of existing zero-shot prompting techniques with proprietary LLMs along with open-source LLMs reveals a general deficiency in CoLeG. To alleviate these issues, we propose tailored approaches for different categories of LLMs. For proprietary LLMs, we introduce a new instructional prompt designed to mitigate the impact of long contexts. For open-source LLMs, we develop a novel auxiliary task for fine-tuning to enhance CoLeG. Our comprehensive results demonstrate the effectiveness of our proposed methods, showing improved performance on E-GSM. Additionally, we conduct an in-depth analysis to differentiate the effects of semantic understanding and reasoning efficacy, showing that our methods improves the latter. We also establish the generalizability of our methods across several other MWP benchmarks. Our findings highlight the limitations of current LLMs and offer practical solutions correspondingly, paving the way for further exploration of model generalizability and training methodologies.

cs.CL↗

arXiv2025

UGMathBench: A Diverse and Dynamic Benchmark for Undergraduate-Level Mathematical Reasoning with Large Language Models

Large Language Models (LLMs) have made significant strides in mathematical reasoning, underscoring the need for a comprehensive and fair evaluation of their capabilities. However, existing benchmarks often fall short, either lacking extensive coverage of undergraduate-level mathematical problems or probably suffering from test-set contamination. To address these issues, we introduce UGMathBench, a diverse and dynamic benchmark specifically designed for evaluating undergraduate-level mathematical reasoning with LLMs. UGMathBench comprises 5,062 problems across 16 subjects and 111 topics, featuring 10 distinct answer types. Each problem includes three randomized versions, with additional versions planned for release as leading open-source LLMs become saturated in UGMathBench. Furthermore, we propose two key metrics: effective accuracy (EAcc), which measures the percentage of correctly solved problems across all three versions, and reasoning gap ($Δ$), which assesses reasoning robustness by calculating the difference between the average accuracy across all versions and EAcc. Our extensive evaluation of 23 leading LLMs reveals that the highest EAcc achieved is 56.3\% by OpenAI-o1-mini, with large $Δ$ values observed across different models. This highlights the need for future research aimed at developing "large reasoning models" with high EAcc and $Δ= 0$. We anticipate that the release of UGMathBench, along with its detailed evaluation codes, will serve as a valuable resource to advance the development of LLMs in solving mathematical problems. Codes and data are available at https://github.com/YangLabHKUST/UGMathBench

cs.CL↗

arXiv2024

Reconstructing Close Human Interactions from Multiple Views

This paper addresses the challenging task of reconstructing the poses of multiple individuals engaged in close interactions, captured by multiple calibrated cameras. The difficulty arises from the noisy or false 2D keypoint detections due to inter-person occlusion, the heavy ambiguity in associating keypoints to individuals due to the close interactions, and the scarcity of training data as collecting and annotating motion data in crowded scenes is resource-intensive. We introduce a novel system to address these challenges. Our system integrates a learning-based pose estimation component and its corresponding training and inference strategies. The pose estimation component takes multi-view 2D keypoint heatmaps as input and reconstructs the pose of each individual using a 3D conditional volumetric network. As the network doesn't need images as input, we can leverage known camera parameters from test scenes and a large quantity of existing motion capture data to synthesize massive training data that mimics the real data distribution in test scenes. Extensive experiments demonstrate that our approach significantly surpasses previous approaches in terms of pose accuracy and is generalizable across various camera setups and population sizes. The code is available on our project page: https://github.com/zju3dv/CloseMoCap.

cs.CV↗

arXiv2023

Flexible and Accurate Methods for Estimation and Inference of Gaussian Graphical Models with Applications

The Gaussian graphical model (GGM) incorporates an undirected graph to represent the conditional dependence between variables, with the precision matrix encoding partial correlation between pair of variables given the others. To achieve flexible and accurate estimation and inference of GGM, we propose the novel method FLAG, which utilizes the random effects model for pairwise conditional regression to estimate the precision matrix and applies statistical tests to recover the graph. Compared with existing methods, FLAG has several unique advantages: (i) it provides accurate estimation without sparsity assumptions on the precision matrix, (ii) it allows for element-wise inference of the precision matrix, (iii) it achieves computational efficiency by developing an efficient PX-EM algorithm and a MM algorithm accelerated with low-rank updates, and (iv) it enables joint estimation of multiple graphs using FLAG-Meta or FLAG-CA. The proposed methods are evaluated using various simulation settings and real data applications, including gene expression in the human brain, term association in university websites, and stock prices in the U.S. financial market. The results demonstrate that FLAG and its extensions provide accurate precision estimation and graph recovery.

stat.ME↗

arXiv2023

Option Dynamic Hedging Using Reinforcement Learning

This work focuses on the dynamic hedging of financial derivatives, where a reinforcement learning algorithm is designed to minimize the variance of the delta hedging process. In contrast to previous research in this area, we apply uncertainty estimation technology to measure the uncertainty of the agent's decision, which can further reduce unnecessary wear and tear in the hedging process and control model overconfidence that may lead to significant losses. Numerical experiments show the superiority of our strategy in Monte Carlo simulations and SP 500 option data.

math.OC↗

arXiv2023

Optimal Execution Using Reinforcement Learning

This work is about optimal order execution, where a large order is split into several small orders to maximize the implementation shortfall. Based on the diversity of cryptocurrency exchanges, we attempt to extract cross-exchange signals by aligning data from multiple exchanges for the first time. Unlike most previous studies that focused on using single-exchange information, we discuss the impact of cross-exchange signals on the agent's decision-making in the optimal execution problem. Experimental results show that cross-exchange signals can provide additional information for the optimal execution of cryptocurrency to facilitate the optimal execution process.

q-fin.TR↗

arXiv2023

Integrating Tick-level Data and Periodical Signal for High-frequency Market Making

We focus on the problem of market making in high-frequency trading. Market making is a critical function in financial markets that involves providing liquidity by buying and selling assets. However, the increasing complexity of financial markets and the high volume of data generated by tick-level trading makes it challenging to develop effective market making strategies. To address this challenge, we propose a deep reinforcement learning approach that fuses tick-level data with periodic prediction signals to develop a more accurate and robust market making strategy. Our results of market making strategies based on different deep reinforcement learning algorithms under the simulation scenarios and real data experiments in the cryptocurrency markets show that the proposed framework outperforms existing methods in terms of profitability and risk management.

q-fin.TR↗

arXiv2022

Modx: Binary Level Partial Imported Third-Party Library Detection through Program Modularization and Semantic Matching

With the rapid growth of software, using third-party libraries (TPLs) has become increasingly popular. The prosperity of the library usage has provided the software engineers with handful of methods to facilitate and boost the program development. Unfortunately, it also poses great challenges as it becomes much more difficult to manage the large volume of libraries. Researches and studies have been proposed to detect and understand the TPLs in the software. However, most existing approaches rely on syntactic features, which are not robust when these features are changed or deliberately hidden by the adversarial parties. Moreover, these approaches typically model each of the imported libraries as a whole, therefore, cannot be applied to scenarios where the host software only partially uses the library code segments. To detect both fully and partially imported TPLs at the semantic level, we propose ModX, a framework that leverages novel program modularization techniques to decompose the program into finegrained functionality-based modules. By extracting both syntactic and semantic features, it measures the distance between modules to detect similar library module reuse in the program. Experimental results show that ModX outperforms other modularization tools by distinguishing more coherent program modules with 353% higher module quality scores and beats other TPL detection tools with on average 17% better in precision and 8% better in recall.

cs.SE↗

arXiv2021

Deep Generative Learning via Schrödinger Bridge

We propose to learn a generative model via entropy interpolation with a Schrödinger Bridge. The generative learning task can be formulated as interpolating between a reference distribution and a target distribution based on the Kullback-Leibler divergence. At the population level, this entropy interpolation is characterized via an SDE on $[0,1]$ with a time-varying drift term. At the sample level, we derive our Schrödinger Bridge algorithm by plugging the drift term estimated by a deep score estimator and a deep density ratio estimator into the Euler-Maruyama method. Under some mild smoothness assumptions of the target distribution, we prove the consistency of both the score estimator and the density ratio estimator, and then establish the consistency of the proposed Schrödinger Bridge approach. Our theoretical results guarantee that the distribution learned by our approach converges to the target distribution. Experimental results on multimodal synthetic data and benchmark data support our theoretical findings and indicate that the generative model via Schrödinger Bridge is comparable with state-of-the-art GANs, suggesting a new formulation of generative learning. We demonstrate its usefulness in image interpolation and image inpainting.

cs.LG↗

arXiv2020

BOLT-SSI: A Statistical Approach to Screening Interaction Effects for Ultra-High Dimensional Data

Detecting interaction effects among predictors on the response variable is a crucial step in various applications. In this paper, we first propose a simple method for sure screening interactions (SSI). Although its computation complexity is $O(p^2n)$, SSI works well for problems of moderate dimensionality (e.g., $p=10^3\sim10^4$), without the heredity assumption. To ultra-high dimensional problems (e.g., $p = 10^6$), motivated by discretization associated Boolean representation and operations and the contingency table for discrete variables, we propose a fast algorithm, named "BOLT-SSI". The statistical theory has been established for SSI and BOLT-SSI, guaranteeing their sure screening property. The performance of SSI and BOLT-SSI are evaluated by comprehensive simulation and real case studies. Numerical results demonstrate that SSI and BOLT-SSI can often outperform their competitors in terms of computational efficiency and statistical accuracy. The proposed method can be applied for fully detecting interactions with more than 300,000 predictors. Based on this study, we believe that there is a great need to rethink the relationship between statistical accuracy and computational efficiency. We have shown that the computational performance of a statistical method can often be greatly improved by exploring the advantages of computational architecture with a tolerable loss of statistical accuracy.

stat.ME↗

arXiv2020

Learning Hybrid Representations for Automatic 3D Vessel Centerline Extraction

Automatic blood vessel extraction from 3D medical images is crucial for vascular disease diagnoses. Existing methods based on convolutional neural networks (CNNs) may suffer from discontinuities of extracted vessels when segmenting such thin tubular structures from 3D images. We argue that preserving the continuity of extracted vessels requires to take into account the global geometry. However, 3D convolutions are computationally inefficient, which prohibits the 3D CNNs from sufficiently large receptive fields to capture the global cues in the entire image. In this work, we propose a hybrid representation learning approach to address this challenge. The main idea is to use CNNs to learn local appearances of vessels in image crops while using another point-cloud network to learn the global geometry of vessels in the entire image. In inference, the proposed approach extracts local segments of vessels using CNNs, classifies each segment based on global geometry using the point-cloud network, and finally connects all the segments that belong to the same vessel using the shortest-path algorithm. This combination results in an efficient, fully-automatic and template-free approach to centerline extraction from 3D images. We validate the proposed approach on CTA datasets and demonstrate its superior performance compared to both traditional and CNN-based baselines.

eess.IV↗

arXiv2020

On the Bound of Cumulative Return in Trading Series and the Verification Using Technical Trading Rules

Although there is a wide use of technical trading rules in stock markets, the profitability of them still remains controversial. This paper first presents and proves the upper bound of cumulative return, and then introduces many of conventional technical trading rules. Furthermore, with the help of bootstrap methodology, we investigate the profitability of technical trading rules on different international stock markets, including developed markets and emerging markets. At last, the results show that the technical trading rules are hard to beat the market, and even less profitable than the random trading strategy.

q-fin.ST↗

arXiv2019

A Coding-free Software Framework of Developing Web Data Management Systems

More and more enterprises recently intend to deploy data management systems in the cloud. Due to the professionalism of software development, it has still been difficult for non-programmers to develop this kind of systems, even a small one. However, the development of SaaS brings forth the more feasibility of coding-free software development than before. Based on the SaaS architecture, this paper presents a set of theory and method for coding-free construction of a data management system, on which our contributions involve in a practical application platform, a set of construction method and a set of interface on data exchange. By abstracting the common features of data management systems, we design a universal web platform to quickly generate and publish customized system instances. Moreover, we propose a kind of method to develop a data management system using a specific requirements table in spreadsheet. The corresponding platform maps the requirements table into a system instance through parsing the table model and implementing the objective system in the running stage. Finally, we implement the proposed framework and deploy it on web. The empirical result demonstrates the feasibility and availability of the coding-free method in developing web data management systems.

cs.SE↗

arXiv2019

Deep Generative Learning via Variational Gradient Flow

We propose a general framework to learn deep generative models via \textbf{V}ariational \textbf{Gr}adient Fl\textbf{ow} (VGrow) on probability spaces. The evolving distribution that asymptotically converges to the target distribution is governed by a vector field, which is the negative gradient of the first variation of the $f$-divergence between them. We prove that the evolving distribution coincides with the pushforward distribution through the infinitesimal time composition of residual maps that are perturbations of the identity map along the vector field. The vector field depends on the density ratio of the pushforward distribution and the target distribution, which can be consistently learned from a binary classification problem. Connections of our proposed VGrow method with other popular methods, such as VAE, GAN and flow-based methods, have been established in this framework, gaining new insights of deep generative learning. We also evaluated several commonly used divergences, including Kullback-Leibler, Jensen-Shannon, Jeffrey divergences as well as our newly discovered `logD' divergence which serves as the objective function of the logD-trick GAN. Experimental results on benchmark datasets demonstrate that VGrow can generate high-fidelity images in a stable and efficient manner, achieving competitive performance with state-of-the-art GANs.

cs.LG↗

arXiv2019

Bayesian Weighted Mendelian Randomization for Causal Inference based on Summary Statistics

The results from Genome-Wide Association Studies (GWAS) on thousands of phenotypes provide an unprecedented opportunity to infer the causal effect of one phenotype (exposure) on another (outcome). Mendelian randomization (MR), an instrumental variable (IV) method, has been introduced for causal inference using GWAS data. Due to the polygenic architecture of complex traits/diseases and the ubiquity of pleiotropy, however, MR has many unique challenges compared to conventional IV methods. We propose a Bayesian weighted Mendelian randomization (BWMR) for causal inference to address these challenges. In our BWMR model, the uncertainty of weak effects owing to polygenicity has been taken into account and the violation of IV assumption due to pleiotropy has been addressed through outlier detection by Bayesian weighting. To make the causal inference based on BWMR computationally stable and efficient, we developed a variational expectation-maximization (VEM) algorithm. Moreover, we have also derived an exact closed-form formula to correct the posterior covariance which is often underestimated in variational inference. Through comprehensive simulation studies, we evaluated the performance of BWMR, demonstrating the advantage of BWMR over its competitors. Then we applied BWMR to make causal inference between 130 metabolites and 93 complex human traits, uncovering novel causal relationship between exposure and outcome traits. The BWMR software is available at https://github.com/jiazhao97/BWMR.

stat.ME↗

arXiv2019

A Unified Primal Dual Active Set Algorithm for Nonconvex Sparse Recovery

In this paper, we consider the problem of recovering a sparse signal based on penalized least squares formulations. We develop a novel algorithm of primal-dual active set type for a class of nonconvex sparsity-promoting penalties, including $\ell^0$, bridge, smoothly clipped absolute deviation, capped $\ell^1$ and minimax concavity penalty. First we establish the existence of a global minimizer for the related optimization problems. Then we derive a novel necessary optimality condition for the global minimizer using the associated thresholding operator. The solutions to the optimality system are coordinate-wise minimizers, and under minor conditions, they are also local minimizers. Upon introducing the dual variable, the active set can be determined using the primal and dual variables together. Further, this relation lends itself to an iterative algorithm of active set type which at each step involves first updating the primal variable only on the active set and then updating the dual variable explicitly. When combined with a continuation strategy on the regularization parameter, the primal dual active set method is shown to converge globally to the underlying regression target under certain regularity conditions. Extensive numerical experiments with both simulated and real data demonstrate its superior performance in efficiency and accuracy compared with the existing sparse recovery methods.

math.OC↗

arXiv2019

Wasserstein-Wasserstein Auto-Encoders

To address the challenges in learning deep generative models (e.g.,the blurriness of variational auto-encoder and the instability of training generative adversarial networks, we propose a novel deep generative model, named Wasserstein-Wasserstein auto-encoders (WWAE). We formulate WWAE as minimization of the penalized optimal transport between the target distribution and the generated distribution. By noticing that both the prior $P_Z$ and the aggregated posterior $Q_Z$ of the latent code Z can be well captured by Gaussians, the proposed WWAE utilizes the closed-form of the squared Wasserstein-2 distance for two Gaussians in the optimization process. As a result, WWAE does not suffer from the sampling burden and it is computationally efficient by leveraging the reparameterization trick. Numerical results evaluated on multiple benchmark datasets including MNIST, fashion- MNIST and CelebA show that WWAE learns better latent structures than VAEs and generates samples of better visual quality and higher FID scores than VAEs and GANs.

cs.LG↗