arXiv subjects

Zhen Wu

Publications and source records attributed to Zhen Wu.

At least 73 records · Page 4Linked to original sources

arXiv2025

Zero-Shot Human-Object Interaction Synthesis with Multimodal Priors

Human-object interaction (HOI) synthesis is important for various applications, ranging from virtual reality to robotics. However, acquiring 3D HOI data is challenging due to its complexity and high cost, limiting existing methods to the narrow diversity of object types and interaction patterns in training datasets. This paper proposes a novel zero-shot HOI synthesis framework without relying on end-to-end training on currently limited 3D HOI datasets. The core idea of our method lies in leveraging extensive HOI knowledge from pre-trained Multimodal Models. Given a text description, our system first obtains temporally consistent 2D HOI image sequences using image or video generation models, which are then uplifted to 3D HOI milestones of human and object poses. We employ pre-trained human pose estimation models to extract human poses and introduce a generalizable category-level 6-DoF estimation method to obtain the object poses from 2D HOI images. Our estimation method is adaptive to various object templates obtained from text-to-3D models or online retrieval. A physics-based tracking of the 3D HOI kinematic milestone is further applied to refine both body motions and object poses, yielding more physically plausible HOI generation results. The experimental results demonstrate that our method is capable of generating open-vocabulary HOIs with physical realism and semantic diversity.

cs.GR↗

arXiv2025

Rethinking Relation Extraction: Beyond Shortcuts to Generalization with a Debiased Benchmark

Benchmarks are crucial for evaluating machine learning algorithm performance, facilitating comparison and identifying superior solutions. However, biases within datasets can lead models to learn shortcut patterns, resulting in inaccurate assessments and hindering real-world applicability. This paper addresses the issue of entity bias in relation extraction tasks, where models tend to rely on entity mentions rather than context. We propose a debiased relation extraction benchmark DREB that breaks the pseudo-correlation between entity mentions and relation types through entity replacement. DREB utilizes Bias Evaluator and PPL Evaluator to ensure low bias and high naturalness, providing a reliable and accurate assessment of model generalization in entity bias scenarios. To establish a new baseline on DREB, we introduce MixDebias, a debiasing method combining data-level and model training-level techniques. MixDebias effectively improves model performance on DREB while maintaining performance on the original dataset. Extensive experiments demonstrate the effectiveness and robustness of MixDebias compared to existing methods, highlighting its potential for improving the generalization ability of relation extraction models. We will release DREB and MixDebias publicly.

cs.AI↗

arXiv2024

Dynamic Demonstrations Controller for In-Context Learning

In-context learning (ICL) is a new paradigm for natural language processing (NLP), where a large language model (LLM) observes a small number of demonstrations and a test instance as its input, and directly makes predictions without updating model parameters. Previous studies have revealed that ICL is sensitive to the selection and the ordering of demonstrations. However, there are few studies regarding the impact of the demonstration number on the ICL performance within a limited input length of LLM, because it is commonly believed that the number of demonstrations is positively correlated with model performance. In this paper, we found this conclusion does not always hold true. Through pilot experiments, we discover that increasing the number of demonstrations does not necessarily lead to improved performance. Building upon this insight, we propose a Dynamic Demonstrations Controller (D$^2$Controller), which can improve the ICL performance by adjusting the number of demonstrations dynamically. The experimental results show that D$^2$Controller yields a 4.6% relative improvement on ten different sizes of LLMs across ten datasets. Moreover, we also extend our method to previous ICL models and achieve competitive results.

cs.CL↗

arXiv2024

Linear-Quadratic Mean Field Control with Non-Convex Data

In this manuscript, we study a class of linear-quadratic (LQ) mean field control problems with a common noise and their corresponding $N$-particle systems. The mean field control problems considered are not standard LQ mean field control problems in the sense that their dependence on the mean field terms can be non-linear and non-convex. Therefore, all the existing methods to deal with LQ mean field control problems fail. The key idea to solve our LQ mean field control problem is to utilize the common noise. We first prove the global well-posedness of the corresponding Hamilton-Jacobi equations via the non-degeneracy of the common noise. In contrast to the LQ mean field games master equations, the Hamilton-Jacobi equations for the LQ mean field control problems can not be reduced to finite-dimensional PDEs. We then globally solve the Hamilton-Jacobi equations for $N$-particle systems. As byproducts, we derive the optimal quantitative convergence results from the $N$-particle systems to the mean field control problem and the propagation of chaos property for the related optimal trajectories. This paper extends the results in [{\sc M. Li, C. Mou, Z. Wu and C. Zhou}, \emph{Trans. Amer. Math. Soc.}, 376(06) (2023), pp.~4105--4143] to the LQ mean field control problems.

math.OC↗

arXiv2024

AlignGPT: Multi-modal Large Language Models with Adaptive Alignment Capability

Multimodal Large Language Models (MLLMs) are widely regarded as crucial in the exploration of Artificial General Intelligence (AGI). The core of MLLMs lies in their capability to achieve cross-modal alignment. To attain this goal, current MLLMs typically follow a two-phase training paradigm: the pre-training phase and the instruction-tuning phase. Despite their success, there are shortcomings in the modeling of alignment capabilities within these models. Firstly, during the pre-training phase, the model usually assumes that all image-text pairs are uniformly aligned, but in fact the degree of alignment between different image-text pairs is inconsistent. Secondly, the instructions currently used for finetuning incorporate a variety of tasks and different tasks usually require different levels of alignment capabilities, but previous MLLMs overlook these differentiated alignment needs. To tackle these issues, we propose a new multimodal large language model AlignGPT. In the pre-training stage, instead of treating all image-text pairs equally, we divide them into different groups according to the degrees of alignment of them. Then, the model is trained to learn the representations of different alignment levels. In the instruction-tuning phase, we adaptively combine these representations of alignment levels to meet the dynamic alignment needs of different tasks. Extensive experimental results show that our model achieves competitive performance on 12 benchmarks.

cs.CL↗

arXiv2024

Quantum Mutual Information in Time

While the quantum mutual information is a fundamental measure of quantum information, it is only defined for spacelike-separated quantum systems. Such a limitation is not present in the theory of classical information, where the mutual information between two random variables is well-defined irrespective of whether or not the variables are separated in space or separated in time. Motivated by this disparity between the classical and quantum mutual information, we employ the pseudo-density matrix formalism to define a simple extension of quantum mutual information into the time domain. As in the spatial case, we show that such a notion of quantum mutual information in time serves as a natural measure of correlation between timelike-separated systems, while also highlighting ways in which quantum correlations distinguish between space and time. We also show how such quantum mutual information is time-symmetric with respect to quantum Bayesian inversion, and then we conclude by showing how mutual information in time yields a Holevo bound for the amount of classical information that may be extracted from sequential measurements on an ensemble of quantum states.

quant-ph↗

arXiv2024

EFUF: Efficient Fine-grained Unlearning Framework for Mitigating Hallucinations in Multimodal Large Language Models

Multimodal large language models (MLLMs) have attracted increasing attention in the past few years, but they may still generate descriptions that include objects not present in the corresponding images, a phenomenon known as object hallucination. To eliminate hallucinations, existing methods manually annotate paired responses with and without hallucinations, and then employ various alignment algorithms to improve the alignment capability between images and text. However, they not only demand considerable computation resources during the finetuning stage but also require expensive human annotation to construct paired data needed by the alignment algorithms. To address these issues, we borrow the idea of unlearning and propose an efficient fine-grained unlearning framework (EFUF), which can eliminate hallucinations without the need for paired data. Extensive experiments show that our method consistently reduces hallucinations while preserving the generation quality with modest computational overhead. Our code and datasets will be publicly available.

cs.CL↗

arXiv2024

RAGLAB: A Modular and Research-Oriented Unified Framework for Retrieval-Augmented Generation

Large Language Models (LLMs) demonstrate human-level capabilities in dialogue, reasoning, and knowledge retention. However, even the most advanced LLMs face challenges such as hallucinations and real-time updating of their knowledge. Current research addresses this bottleneck by equipping LLMs with external knowledge, a technique known as Retrieval Augmented Generation (RAG). However, two key issues constrained the development of RAG. First, there is a growing lack of comprehensive and fair comparisons between novel RAG algorithms. Second, open-source tools such as LlamaIndex and LangChain employ high-level abstractions, which results in a lack of transparency and limits the ability to develop novel algorithms and evaluation metrics. To close this gap, we introduce RAGLAB, a modular and research-oriented open-source library. RAGLAB reproduces 6 existing algorithms and provides a comprehensive ecosystem for investigating RAG algorithms. Leveraging RAGLAB, we conduct a fair comparison of 6 RAG algorithms across 10 benchmarks. With RAGLAB, researchers can efficiently compare the performance of various algorithms and develop novel algorithms.

cs.CL↗

arXiv2024

Stochastic Singular Linear Systems and Related Linear-Quadratic Optimal Control Problems under Finite and Infinite Horizons

In this paper, we study the necessary and sufficient conditions for ensuring the well-posedness of the stochastic singular systems. Moreover, we investigate the stochastic singular linear-quadratic control problems, considering both finite and infinite horizons, and transform each of these problems into their corresponding normal linear-quadratic control problem. To guarantee the finiteness of the infinite-horizon linear-quadratic control problem, we establish the Popov-Belevitch-Hautus rank criterion for accessing the controllability of the stochastic system. Furthermore, we derive the feedback form of the optimal control. Finally, we provide solutions for illustrated examples of the stochastic singular linear-quadratic control problem in both finite and infinite horizons.

math.OC↗

arXiv2024

Linear-Quadratic Mean-Field Game for Stochastic Systems with Partial Observation

This paper is concerned with a class of linear-quadratic stochastic large-population problems with partial information, where the individual agent only has access to a noisy observation process related to the state. The dynamics of each agent follows a linear stochastic differential equation driven by individual noise, and all agents are coupled together via the control average term. Using the mean-field game approach and the backward separation principle with a state decomposition technique, the decentralized optimal control can be obtained in the open-loop form through a forward-backward stochastic differential equation with the conditional expectation. The optimal filtering equation is also provided. By the decoupling method, the decentralized optimal control can also be further presented as the feedback of state filtering via the Riccati equation. The explicit solution of the control average limit is given, and the consistency condition system is discussed. Moreover, the related $\varepsilon$-Nash equilibrium property is verified. To illustrate the good performance of theoretical results, an example in finance is studied.

math.OC↗

arXiv2024

Finite-time and bumpless transfer control of asynchronously switched systems: An output feedback control approach

In this paper, the finite-time control and bumpless transfer control are investigated for switched systems under asynchronously switching. First, a class of dynamic output feedback controllers are designed to stabilize the switched system with measurable system outputs. Considering the improvement of transient performance, the bumpless transfer control and finite-time control are further studied in the controller design. To avoid the control bumps, a practical filter is introduced to make the control signal smoother and continuous. Furthermore, to derive a finite-time bounded system state over short-time intervals, the finite-time analysis is considered in managing the switching process with the average dwell time. New criteria are proposed to analyze the finite-time stability and finite-time boundedness for the closed-loop system and solvable conditions are newly proposed to optimize the controller gain. Finally, the superiorities of the proposed method are validated through an application to a boost converter.

eess.SY↗

arXiv2024

General Communication Enhancement via the Quantum Switch

Recent studies have shown that quantum information may be effectively transmitted by a finite collection of completely depolarizing channels in a coherent superposition of different orders, via an operation known as the quantum $\tt SWITCH$. Such results are quite remarkable, as completely depolarizing channels taken in isolation and in a definite order can only output white noise. For general channels however, little is known about the potential communication enhancement provided by the quantum $\tt SWITCH$. In this Letter, we define an easily computable quantity $\mathcal{P}_n$ associated with the quantum ${\tt SWITCH}$ of $n$ copies of a fixed channel, and we conjecture that $\mathcal{P}_n>0$ is both a necessary and sufficient condition for communication enhancement via the quantum $\tt SWITCH$. In support of our conjecture, we derive a simple analytic expression for the classical capacity of the quantum $\tt SWITCH$ of $n$ copies of an arbitrary Pauli channel in terms of the quantity $\mathcal{P}_n$, which we then use to show that our conjecture indeed holds in the space of all Pauli channels. Utilizing such results, we then formulate a communication protocol involving the quantum $\tt SWITCH$ which enhances the private capacity of the BB84 channel.

quant-ph↗

arXiv2024

Leveraging Machine-Generated Rationales to Facilitate Social Meaning Detection in Conversations

We present a generalizable classification approach that leverages Large Language Models (LLMs) to facilitate the detection of implicitly encoded social meaning in conversations. We design a multi-faceted prompt to extract a textual explanation of the reasoning that connects visible cues to underlying social meanings. These extracted explanations or rationales serve as augmentations to the conversational text to facilitate dialogue understanding and transfer. Our empirical results over 2,340 experimental settings demonstrate the significant positive impact of adding these rationales. Our findings hold true for in-domain classification, zero-shot, and few-shot domain transfer for two different social meaning detection tasks, each spanning two different corpora.

cs.CL↗

arXiv2024

AutoSurvey: Large Language Models Can Automatically Write Surveys

This paper introduces AutoSurvey, a speedy and well-organized methodology for automating the creation of comprehensive literature surveys in rapidly evolving fields like artificial intelligence. Traditional survey paper creation faces challenges due to the vast volume and complexity of information, prompting the need for efficient survey methods. While large language models (LLMs) offer promise in automating this process, challenges such as context window limitations, parametric knowledge constraints, and the lack of evaluation benchmarks remain. AutoSurvey addresses these challenges through a systematic approach that involves initial retrieval and outline generation, subsection drafting by specialized LLMs, integration and refinement, and rigorous evaluation and iteration. Our contributions include a comprehensive solution to the survey problem, a reliable evaluation method, and experimental validation demonstrating AutoSurvey's effectiveness.We open our resources at \url{https://github.com/AutoSurveys/AutoSurvey}.

cs.IR↗

arXiv2024

Extroversion or Introversion? Controlling The Personality of Your Large Language Models

Large language models (LLMs) exhibit robust capabilities in text generation and comprehension, mimicking human behavior and exhibiting synthetic personalities. However, some LLMs have displayed offensive personality, propagating toxic discourse. Existing literature neglects the origin and evolution of LLM personalities, as well as the effective personality control. To fill these gaps, our study embarked on a comprehensive investigation into LLM personality control. We investigated several typical methods to influence LLMs, including three training methods: Continual Pre-training, Supervised Fine-Tuning (SFT), and Reinforcement Learning from Human Feedback (RLHF), along with inference phase considerations (prompts). Our investigation revealed a hierarchy of effectiveness in control: Prompt > SFT > RLHF > Continual Pre-train. Notably, SFT exhibits a higher control success rate compared to prompt induction. While prompts prove highly effective, we found that prompt-induced personalities are less robust than those trained, making them more prone to showing conflicting personalities under reverse personality prompt induction. Besides, harnessing the strengths of both SFT and prompt, we proposed $\underline{\text{P}}$rompt $\underline{\text{I}}$nduction post $\underline{\text{S}}$upervised $\underline{\text{F}}$ine-tuning (PISF), which emerges as the most effective and robust strategy for controlling LLMs' personality, displaying high efficacy, high success rates, and high robustness. Even under reverse personality prompt induction, LLMs controlled by PISF still exhibit stable and robust personalities.

cs.CL↗

arXiv2024

Problem of eigenvalues of stochastic Hamiltonian systems with boundary conditions and Markov chain

In this paper, we study the eigenvalue problem of stochastic Hamiltonian system driven by Brownian motion and Markov chain with boundary conditions and time-dependent coefficients. For any dimensional case, the existence of the first eigenvalue is proven and the corresponding eigenfunctions are constructed by virtue of dual transformation and generalized Riccati equation system. Furthermore, we have more finely characterized the existence of all eigenvalues and constructed the related eigenfunctions for one-dimensional Hamiltonian system. Moreover, the increasing order of these eigenvalues have also been given.

math.PR↗

arXiv2024

Viscosity solutions for mean field optimal switching with a two-time-scale Markov chain

In this paper, we consider the mean field optimal switching problem with a Markov chain under viscosity solution notion. Based on the conditional distribution of the Markov chain, the value function and corresponding dynamic programming principle (DPP) are established. The switching problem is characterized by an obstacle equation on the Wasserstein space, and the existence, stability, and comparison principle are obtained in the sense of viscosity solution. In particular, we consider a two-time-scale structure and obtain the convergence of the limit system. As an application of our theoretical results, an innovative example concerning the stock trading problem in a regime switching market is solved.

math.OC↗

arXiv2024

Emotion-Anchored Contrastive Learning Framework for Emotion Recognition in Conversation

Emotion Recognition in Conversation (ERC) involves detecting the underlying emotion behind each utterance within a conversation. Effectively generating representations for utterances remains a significant challenge in this task. Recent works propose various models to address this issue, but they still struggle with differentiating similar emotions such as excitement and happiness. To alleviate this problem, We propose an Emotion-Anchored Contrastive Learning (EACL) framework that can generate more distinguishable utterance representations for similar emotions. To achieve this, we utilize label encodings as anchors to guide the learning of utterance representations and design an auxiliary loss to ensure the effective separation of anchors for similar emotions. Moreover, an additional adaptation process is proposed to adapt anchors to serve as effective classifiers to improve classification performance. Across extensive experiments, our proposed EACL achieves state-of-the-art emotion recognition performance and exhibits superior performance on similar emotions. Our code is available at https://github.com/Yu-Fangxu/EACL.

cs.CL↗