arXiv subjects

Yutong Wu

Publications and source records attributed to Yutong Wu.

At least 37 records · Page 2Linked to original sources

arXiv2025

Early Lung Cancer Diagnosis from Virtual Follow-up LDCT Generation via Correlational Autoencoder and Latent Flow Matching

Lung cancer is one of the most commonly diagnosed cancers, and early diagnosis is critical because the survival rate declines sharply once the disease progresses to advanced stages. However, achieving an early diagnosis remains challenging, particularly in distinguishing subtle early signals of malignancy from those of benign conditions. In clinical practice, a patient with a high risk may need to undergo an initial baseline and several annual follow-up examinations (e.g., CT scans) before receiving a definitive diagnosis, which can result in missing the optimal treatment. Recently, Artificial Intelligence (AI) methods have been increasingly used for early diagnosis of lung cancer, but most existing algorithms focus on radiomic features extraction from single early-stage CT scans. Inspired by recent advances in diffusion models for image generation, this paper proposes a generative method, named CorrFlowNet, which creates a virtual, one-year follow-up CT scan after the initial baseline scan. This virtual follow-up would allow for an early detection of malignant/benign nodules, reducing the need to wait for clinical follow-ups. During training, our approach employs a correlational autoencoder to encode both early baseline and follow-up CT images into a latent space that captures the dynamics of nodule progression as well as the correlations between them, followed by a flow matching algorithm on the latent space with a neural ordinary differential equation. An auxiliary classifier is used to further enhance the diagnostic accuracy. Evaluations on a real clinical dataset show our method can significantly improve downstream lung nodule risk assessment compared with existing baseline models. Moreover, its diagnostic accuracy is comparable with real clinical CT follow-ups, highlighting its potential to improve cancer diagnosis.

cs.CV↗

arXiv2025

Cost-Optimal Grouped-Query Attention for Long-Context Modeling

Grouped-Query Attention (GQA) is a widely adopted strategy for reducing the computational cost of attention layers in large language models (LLMs). However, current GQA configurations are often suboptimal because they overlook how context length influences inference cost. Since inference cost grows with context length, the most cost-efficient GQA configuration should also vary accordingly. In this work, we analyze the relationship among context length, model size, GQA configuration, and model loss, and introduce two innovations: (1) we decouple the total head size from the hidden size, enabling more flexible control over attention FLOPs; and (2) we jointly optimize the model size and the GQA configuration to arrive at a better allocation of inference resources between attention layers and other components. Our analysis reveals that commonly used GQA configurations are highly suboptimal for long-context scenarios. More importantly, we propose a recipe for deriving cost-optimal GQA configurations. Our results show that for long-context scenarios, one should use fewer attention heads while scaling up model size. Configurations selected by our recipe can reduce both memory usage and FLOPs by more than 50% compared to Llama-3's GQA, with *no degradation in model capabilities*. Our findings offer valuable insights for designing efficient long-context LLMs. The code is available at https://www.github.com/THUNLP/cost-optimal-gqa .

cs.CL↗

arXiv2025

StepFun-Prover Preview: Let's Think and Verify Step by Step

We present StepFun-Prover Preview, a large language model designed for formal theorem proving through tool-integrated reasoning. Using a reinforcement learning pipeline that incorporates tool-based interactions, StepFun-Prover can achieve strong performance in generating Lean 4 proofs with minimal sampling. Our approach enables the model to emulate human-like problem-solving strategies by iteratively refining proofs based on real-time environment feedback. On the miniF2F-test benchmark, StepFun-Prover achieves a pass@1 success rate of $70.0\%$. Beyond advancing benchmark performance, we introduce an end-to-end training framework for developing tool-integrated reasoning models, offering a promising direction for automated theorem proving and Math AI assistant.

cs.AI↗

arXiv2025

Cowpox: Towards the Immunity of VLM-based Multi-Agent Systems

Vision Language Model (VLM)-based agents are stateful, autonomous entities capable of perceiving and interacting with their environments through vision and language. Multi-agent systems comprise specialized agents who collaborate to solve a (complex) task. A core security property is robustness, stating that the system should maintain its integrity under adversarial attacks. However, the design of existing multi-agent systems lacks the robustness consideration, as a successful exploit against one agent can spread and infect other agents to undermine the entire system's assurance. To address this, we propose a new defense approach, Cowpox, to provably enhance the robustness of multi-agent systems. It incorporates a distributed mechanism, which improves the recovery rate of agents by limiting the expected number of infections to other agents. The core idea is to generate and distribute a special cure sample that immunizes an agent against the attack before exposure and helps recover the already infected agents. We demonstrate the effectiveness of Cowpox empirically and provide theoretical robustness guarantees.

cs.MA↗

arXiv2025

Data-Efficient Neural Training with Dynamic Connectomes

The study of dynamic functional connectomes has provided valuable insights into how patterns of brain activity change over time. Neural networks process information through artificial neurons, conceptually inspired by patterns of activation in the brain. However, their hierarchical structure and high-dimensional parameter space pose challenges for understanding and controlling training dynamics. In this study, we introduce a novel approach to characterize training dynamics in neural networks by representing evolving neural activations as functional connectomes and extracting dynamic signatures of activity throughout training. Our results show that these signatures effectively capture key transitions in the functional organization of the network. Building on this analysis, we propose the use of a time series of functional connectomes as an intrinsic indicator of learning progress, enabling a principled early stopping criterion. Our framework performs robustly across benchmarks and provides new insights into neural network training dynamics.

q-bio.NC↗

arXiv2025

Functional connectomes of neural networks

The human brain is a complex system, and understanding its mechanisms has been a long-standing challenge in neuroscience. The study of the functional connectome, which maps the functional connections between different brain regions, has provided valuable insights through various advanced analysis techniques developed over the years. Similarly, neural networks, inspired by the brain's architecture, have achieved notable success in diverse applications but are often noted for their lack of interpretability. In this paper, we propose a novel approach that bridges neural networks and human brain functions by leveraging brain-inspired techniques. Our approach, grounded in the insights from the functional connectome, offers scalable ways to characterize topology of large neural networks using stable statistical and machine learning techniques. Our empirical analysis demonstrates its capability to enhance the interpretability of neural networks, providing a deeper understanding of their underlying mechanisms.

cs.NE↗

arXiv2024

InverseCoder: Self-improving Instruction-Tuned Code LLMs with Inverse-Instruct

Recent advancements in open-source code large language models (LLMs) have been driven by fine-tuning on the data generated from powerful closed-source LLMs, which are expensive to obtain. This paper explores whether it is possible to use a fine-tuned open-source model to generate additional data to augment its instruction-tuning dataset. We make two observations: (1) A code snippet can serve as the response to different instructions. (2) Instruction-tuned code LLMs perform better at translating code into instructions than the reverse. Based on these observations, we propose Inverse-Instruct, a data augmentation technique that uses a fine-tuned LLM to generate additional instructions of code responses from its own training dataset. The additional instruction-response pairs are added to the original dataset, and a stronger code LLM can be obtained by fine-tuning on the augmented dataset. We empirically validate Inverse-Instruct on a range of open-source code models (e.g. CodeLlama-Python and DeepSeek-Coder) and benchmarks (e.g., HumanEval(+), MBPP(+), DS-1000 and MultiPL-E), showing it consistently improves the base models.

cs.CL↗

arXiv2024

BMRMM: An R Package for Bayesian Markov (Renewal) Mixed Models

We introduce the BMRMM package implementing Bayesian inference for a class of Markov renewal mixed models which can characterize the stochastic dynamics of a collection of sequences, each comprising alternative instances of categorical states and associated continuous duration times, while being influenced by a set of exogenous factors as well as a 'random' individual. The default setting flexibly models the state transition probabilities using mixtures of Dirichlet distributions and the duration times using mixtures of gamma kernels while also allowing variable selection for both. Modeling such data using simpler Markov mixed models also remains an option, either by ignoring the duration times altogether or by replacing them with instances of an additional category obtained by discretizing them by a user-specified unit. The option is also useful when data on duration times may not be available in the first place. We demonstrate the package's utility using two data sets.

stat.ME↗

arXiv2024

MMToM-QA: Multimodal Theory of Mind Question Answering

Theory of Mind (ToM), the ability to understand people's mental states, is an essential ingredient for developing machines with human-level social intelligence. Recent machine learning models, particularly large language models, seem to show some aspects of ToM understanding. However, existing ToM benchmarks use unimodal datasets - either video or text. Human ToM, on the other hand, is more than video or text understanding. People can flexibly reason about another person's mind based on conceptual representations (e.g., goals, beliefs, plans) extracted from any available data. To address this, we introduce a multimodal Theory of Mind question answering (MMToM-QA) benchmark. MMToM-QA comprehensively evaluates machine ToM both on multimodal data and on different kinds of unimodal data about a person's activity in a household environment. To engineer multimodal ToM capacity, we propose a novel method, BIP-ALM (Bayesian Inverse Planning Accelerated by Language Models). BIP-ALM extracts unified representations from multimodal data and utilizes language models for scalable Bayesian inverse planning. We conducted a systematic comparison of human performance, BIP-ALM, and state-of-the-art models, including GPT-4. The experiments demonstrate that large language models and large multimodal models still lack robust ToM capacity. BIP-ALM, on the other hand, shows promising results, by leveraging the power of both model-based mental inference and language models.

cs.AI↗

arXiv2024

Intelligent EC Rearview Mirror: Enhancing Driver Safety with Dynamic Glare Mitigation via Cloud Edge Collaboration

Sudden glare from trailing vehicles significantly increases driving safety risks. Existing anti-glare technologies such as electronic, manually-adjusted, and electrochromic rearview mirrors, are expensive and lack effective adaptability in different lighting conditions. To address these issues, our research introduces an intelligent rearview mirror system utilizing novel all-liquid electrochromic technology. This system integrates IoT with ensemble and federated learning within a cloud edge collaboration framework, dynamically controlling voltage to effectively eliminate glare and maintain clear visibility. Utilizing an ensemble learning model, it automatically adjusts mirror transmittance based on light intensity, achieving a low RMSE of 0.109 on the test set. Furthermore, the system leverages federated learning for distributed data training across devices, which enhances privacy and updates the cloud model continuously. Distinct from conventional methods, our experiment utilizes the Schmidt-Clausen and Bindels de Boer 9-point scale with TOPSIS for comprehensive evaluation of rearview mirror glare. Designed to be convenient and costeffective, this system demonstrates how IoT and AI can significantly enhance rearview mirror anti-glare performance.

cs.HC↗

arXiv2024

Giant Nonreciprocity of Surface Acoustic Waves induced by a positive-negative magnetostrictive heterostructure

Lack of nonreciprocity is one of the major drawbacks of solid-state acoustic devices, which has hindered the development of microwave-frequency acoustic isolators and circulators. Here we report giant nonreciprocal transmission of shear-horizontal surface acoustic waves (SH-SAWs) on a LiTaO3 substrate coated with a negative-positive magnetostrictive bilayer structure of Ni/Ti/FeCoSiB. Although the static magnetic moments of two layers are parallel, SH-SAWs can excite optical-mode spin waves much stronger than acoustic-mode ones at relatively low frequencies via magnetoelastic coupling. The measured magnitude nonreciprocity exceeds 40 dB (or 80 dB/mm) at 2.333 GHz. In addition, maximum nonreciprocal phase accumulation reaches 188° (376°/mm), which is desired for an effective SAW circulator. Our theoretical model and calculations provide an insight into the observed phenomena and demonstrate a pathway for further improvement of nonreciprocal acoustic devices.

physics.app-ph↗

arXiv2024

Reinforcement Learning-assisted Evolutionary Algorithm: A Survey and Research Opportunities

Evolutionary algorithms (EA), a class of stochastic search methods based on the principles of natural evolution, have received widespread acclaim for their exceptional performance in various real-world optimization problems. While researchers worldwide have proposed a wide variety of EAs, certain limitations remain, such as slow convergence speed and poor generalization capabilities. Consequently, numerous scholars actively explore improvements to algorithmic structures, operators, search patterns, etc., to enhance their optimization performance. Reinforcement learning (RL) integrated as a component in the EA framework has demonstrated superior performance in recent years. This paper presents a comprehensive survey on integrating reinforcement learning into the evolutionary algorithm, referred to as reinforcement learning-assisted evolutionary algorithm (RL-EA). We begin with the conceptual outlines of reinforcement learning and the evolutionary algorithm. We then provide a taxonomy of RL-EA. Subsequently, we discuss the RL-EA integration method, the RL-assisted strategy adopted by RL-EA, and its applications according to the existing literature. The RL-assisted procedure is divided according to the implemented functions including solution generation, learnable objective function, algorithm/operator/sub-population selection, parameter adaptation, and other strategies. Additionally, different attribute settings of RL in RL-EA are discussed. In the applications of RL-EA section, we also demonstrate the excellent performance of RL-EA on several benchmarks and a range of public datasets to facilitate a quick comparative study. Finally, we analyze potential directions for future research.

cs.NE↗

arXiv2023

Personalization as a Shortcut for Few-Shot Backdoor Attack against Text-to-Image Diffusion Models

Although recent personalization methods have democratized high-resolution image synthesis by enabling swift concept acquisition with minimal examples and lightweight computation, they also present an exploitable avenue for high accessible backdoor attacks. This paper investigates a critical and unexplored aspect of text-to-image (T2I) diffusion models - their potential vulnerability to backdoor attacks via personalization. Our study focuses on a zero-day backdoor vulnerability prevalent in two families of personalization methods, epitomized by Textual Inversion and DreamBooth.Compared to traditional backdoor attacks, our proposed method can facilitate more precise, efficient, and easily accessible attacks with a lower barrier to entry. We provide a comprehensive review of personalization in T2I diffusion models, highlighting the operation and exploitation potential of this backdoor vulnerability. To be specific, by studying the prompt processing of Textual Inversion and DreamBooth, we have devised dedicated backdoor attacks according to the different ways of dealing with unseen tokens and analyzed the influence of triggers and concept images on the attack effect. Through comprehensive empirical study, we endorse the utilization of the nouveau-token backdoor attack due to its impressive effectiveness, stealthiness, and integrity, markedly outperforming the legacy-token backdoor attack.

cs.CV↗

arXiv2023

Ensemble Reinforcement Learning: A Survey

Reinforcement Learning (RL) has emerged as a highly effective technique for addressing various scientific and applied problems. Despite its success, certain complex tasks remain challenging to be addressed solely with a single model and algorithm. In response, ensemble reinforcement learning (ERL), a promising approach that combines the benefits of both RL and ensemble learning (EL), has gained widespread popularity. ERL leverages multiple models or training algorithms to comprehensively explore the problem space and possesses strong generalization capabilities. In this study, we present a comprehensive survey on ERL to provide readers with an overview of recent advances and challenges in the field. Firstly, we provide an introduction to the background and motivation for ERL. Secondly, we conduct a detailed analysis of strategies such as model selection and combination that have been successfully implemented in ERL. Subsequently, we explore the application of ERL, summarize the datasets, and analyze the algorithms employed. Finally, we outline several open questions and discuss future research directions of ERL. By offering guidance for future scientific research and engineering applications, this survey significantly contributes to the advancement of ERL.

cs.LG↗

arXiv2023

some properties for asymptotically tracially approximation of C*-algebras

Let $Ω$ be a class of unital $\rm C^{*}$-algebras. The class of ${\rm C^*}$-algebras which are asymptotical tracially in $Ω$, denoted by ${\rm AT}Ω$. In this paper, we will show that the following class of ${\rm C^*}$-algebras in the class $Ω$ are inherited by simple unital ${\rm C^*}$-algebras in the class $\rm TΩ$ $(1)$ the class of real rank zero ${\rm C^*}$-algebras, $(2)$ the class of ${\rm C^*}$-algebras with the radius of comparison $n$, and $(3)$ the class of $\rm ATΩ$. As an application, let $Ω$ be a class of unital ${\rm C^*}$-algebras which have generalized tracial rank at most one (or has tracial topological rank zero, or has tracial topological rank one). Let $A$ be a unital separable simple ${\rm C^*}$-algebra such that $A\in \rm ATΩ$, then $A$ has generalized tracial rank at most one (or has tracial topological rank zero, or has tracial topological rank one).

math.OA↗

arXiv2023

Large Nonreciprocity of Shear-Horizontal Surface Acoustic Waves induced by Magnetoelastic Bilayers

We report large nonreciprocity in the transmission of shear-horizontal surface acoustic waves (SAWs) on LiTaO3 substrate coated with a FeCoSiB/NiFeCu magnetoelastic bilayer. The large difference in saturation magnetization of the two layers not only brings nonreciprocal spin waves (SWs), but also ensures the phonon-magnon (SAWs-SWs) coupling at relatively low wavenumbers. It is found that the angle between the magnetization and the wavevector play important roles in determining the strength of magnetoelastic coupling and nonreciprocity, simultaneously. A large nonreciprocal transmission of SAWs about 30 dB (i.e. 60 dB/mm) is demonstrated at 2.33 GHz. In addition, the dispersion relation between coupled SH-SAWs and nonreciprocal SWs is developed, which provide a good insight into the observed phenomena. Our results offer a convenient approach to implement nonreciprocal SAW isolators or circulators.

physics.app-ph↗

arXiv2023

Backdooring Textual Inversion for Concept Censorship

Recent years have witnessed success in AIGC (AI Generated Content). People can make use of a pre-trained diffusion model to generate images of high quality or freely modify existing pictures with only prompts in nature language. More excitingly, the emerging personalization techniques make it feasible to create specific-desired images with only a few images as references. However, this induces severe threats if such advanced techniques are misused by malicious users, such as spreading fake news or defaming individual reputations. Thus, it is necessary to regulate personalization models (i.e., concept censorship) for their development and advancement. In this paper, we focus on the personalization technique dubbed Textual Inversion (TI), which is becoming prevailing for its lightweight nature and excellent performance. TI crafts the word embedding that contains detailed information about a specific object. Users can easily download the word embedding from public websites like Civitai and add it to their own stable diffusion model without fine-tuning for personalization. To achieve the concept censorship of a TI model, we propose leveraging the backdoor technique for good by injecting backdoors into the Textual Inversion embeddings. Briefly, we select some sensitive words as triggers during the training of TI, which will be censored for normal use. In the subsequent generation stage, if the triggers are combined with personalized embeddings as final prompts, the model will output a pre-defined target image rather than images including the desired malicious concept. To demonstrate the effectiveness of our approach, we conduct extensive experiments on Stable Diffusion, a prevailing open-sourced text-to-image model. Our code, data, and results are available at https://concept-censorship.github.io.

cs.CR↗

arXiv2023

Prophet Inequality on I.I.D. Distributions: Beating 1-1/e with a Single Query

In this work, we study the single-choice prophet inequality problem, where a gambler faces a sequence of~$n$ online i.i.d. random variables drawn from an unknown distribution. When a variable reveals its value, the gambler needs to decide irrevocably whether or not to accept the value. The goal is to maximize the competitive ratio between the expected gain of the gambler and that of the maximum variable. It is shown by Correa et al. that when the distribution is unknown or only $o(n)$ uniform samples from the distribution are given, the best an algorithm can do is $1/e$-competitive. In contrast, when the distribution is known or $Ω(n)$ uniform samples are given, the optimal competitive ratio of 0.7451 can be achieved. In this paper, we study a new model in which the algorithm has access to an oracle that answers quantile queries about the distribution and investigate to what extent we can use a small number of queries to achieve good competitive ratios. We first use the answers from the queries to implement the threshold-based algorithms and show that with two thresholds our algorithm achieves a competitive ratio of $0.6786$. Motivated by the two-threshold algorithm, we design the observe-and-accept algorithm that requires only a single query. This algorithm sets a threshold in the first phase by making a query and uses the maximum realization from the first phase as the threshold for the second phase. It can be viewed as a natural combination of the single-threshold algorithm and the algorithm for the secretary problem. By properly choosing the quantile to query and the break-point between the two phases, we achieve a competitive ratio of $0.6718$, beating the benchmark of $1-1/e$.

cs.DS↗