arXiv subjects

Xin Chen

Publications and source records attributed to Xin Chen.

At least 325 records · Page 18Linked to original sources

arXiv2024

MikuDance: Animating Character Art with Mixed Motion Dynamics

We propose MikuDance, a diffusion-based pipeline incorporating mixed motion dynamics to animate stylized character art. MikuDance consists of two key techniques: Mixed Motion Modeling and Mixed-Control Diffusion, to address the challenges of high-dynamic motion and reference-guidance misalignment in character art animation. Specifically, a Scene Motion Tracking strategy is presented to explicitly model the dynamic camera in pixel-wise space, enabling unified character-scene motion modeling. Building on this, the Mixed-Control Diffusion implicitly aligns the scale and body shape of diverse characters with motion guidance, allowing flexible control of local character motion. Subsequently, a Motion-Adaptive Normalization module is incorporated to effectively inject global scene motion, paving the way for comprehensive character art animation. Through extensive experiments, we demonstrate the effectiveness and generalizability of MikuDance across various character art and motion guidance, consistently producing high-quality animations with remarkable motion dynamics.

cs.CV↗

arXiv2024

SciDFM: A Large Language Model with Mixture-of-Experts for Science

Recently, there has been a significant upsurge of interest in leveraging large language models (LLMs) to assist scientific discovery. However, most LLMs only focus on general science, while they lack domain-specific knowledge, such as chemical molecules and amino acid sequences. To bridge these gaps, we introduce SciDFM, a mixture-of-experts LLM, which is trained from scratch and is able to conduct college-level scientific reasoning and understand molecules and amino acid sequences. We collect a large-scale training corpus containing numerous scientific papers and books from different disciplines as well as data from domain-specific databases. We further fine-tune the pre-trained model on lots of instruction data to improve performances on downstream benchmarks. From experiment results, we show that SciDFM achieves strong performance on general scientific benchmarks such as SciEval and SciQ, and it reaches a SOTA performance on domain-specific benchmarks among models of similar size. We further analyze the expert layers and show that the results of expert selection vary with data from different disciplines. To benefit the broader research community, we open-source SciDFM at https://huggingface.co/OpenDFM/SciDFM-MoE-A5.6B-v1.0.

cs.CL↗

arXiv2024

Content-Based Collaborative Generation for Recommender Systems

Generative models have emerged as a promising utility to enhance recommender systems. It is essential to model both item content and user-item collaborative interactions in a unified generative framework for better recommendation. Although some existing large language model (LLM)-based methods contribute to fusing content information and collaborative signals, they fundamentally rely on textual language generation, which is not fully aligned with the recommendation task. How to integrate content knowledge and collaborative interaction signals in a generative framework tailored for item recommendation is still an open research challenge. In this paper, we propose content-based collaborative generation for recommender systems, namely ColaRec. ColaRec is a sequence-to-sequence framework which is tailored for directly generating the recommended item identifier. Precisely, the input sequence comprises data pertaining to the user's interacted items, and the output sequence represents the generative identifier (GID) for the suggested item. To model collaborative signals, the GIDs are constructed from a pretrained collaborative filtering model, and the user is represented as the content aggregation of interacted items. To this end, ColaRec captures both collaborative signals and content information in a unified framework. Then an item indexing task is proposed to conduct the alignment between the content-based semantic space and the interaction-based collaborative space. Besides, a contrastive loss is further introduced to ensure that items with similar collaborative GIDs have similar content representations. To verify the effectiveness of ColaRec, we conduct experiments on four benchmark datasets. Empirical results demonstrate the superior performance of ColaRec.

cs.IR↗

arXiv2024

Quantum-Powered Optimization for Electric Vehicle Charging Infrastructure Deployment

The infrastructure development of electric vehicle charging stations (EVCS) is critical to the integration of electrical vehicles (EVs) into transportation systems, which requires significant investment and has long-term impact on the adoption of EVs. In this paper, a mathematical model is developed to identify the optimal placement of EVCS by utilizing a novel quantum annealing (QA) algorithm and quantum computation (QC). The objective of the optimization model is to determine the locations of EVCS that maximize their service quality for EV users. The model is validated using a real-world case study and solved using commercially available quantum computers from D-Wave. The case study shows that the QA algorithm can find the optimal placement of EVCS within seconds. The quality of the solutions obtained using QC is not sensitive to the shape or size of the area where EVCS are to be deployed.

quant-ph↗

arXiv2024

The Fairness of Maximum Nash Social Welfare Under Matroid Constraints and Beyond

We study the problem of fair allocation of a set of indivisible items among agents with additive valuations, under matroid constraints and two generalizations: $p$-extendible system and independence system constraints. The objective is to find fair and efficient allocations in which the subset of items assigned to every agent satisfies the given constraint. We focus on a common fairness notion of envy-freeness up to one item (EF1) and a well-known efficient (and fair) notion of the maximum Nash social welfare (Max-NSW). By using properties of matroids, we demonstrate that the Max-NSW allocation, implying Pareto optimality (PO), achieves a tight $1/2$-EF1 under matroid constraints. This result resolves an open question proposed in prior literature [26]. In particular, if agents have 2-valued ($\{1, a\}$) valuations, we prove that the Max-NSW allocation admits $\max\{1/a^2, 1/2\}$-EF1 and PO. Under strongly $p$-extendible system constraints, we show that the Max-NSW allocation guarantees $\max\{1/p, 1/4\}$-EF1 and PO for identical binary valuations. Indeed, the approximation of $1/4$ is the ratio for independence system constraints and additive valuations. Additionally, for lexicographic preferences, we study possibly feasible allocations other than Max-NSW admitting exactly EF1 and PO under the above constraints.

cs.GT↗

arXiv2024

MVPaint: Synchronized Multi-View Diffusion for Painting Anything 3D

Texturing is a crucial step in the 3D asset production workflow, which enhances the visual appeal and diversity of 3D assets. Despite recent advancements in Text-to-Texture (T2T) generation, existing methods often yield subpar results, primarily due to local discontinuities, inconsistencies across multiple views, and their heavy dependence on UV unwrapping outcomes. To tackle these challenges, we propose a novel generation-refinement 3D texturing framework called MVPaint, which can generate high-resolution, seamless textures while emphasizing multi-view consistency. MVPaint mainly consists of three key modules. 1) Synchronized Multi-view Generation (SMG). Given a 3D mesh model, MVPaint first simultaneously generates multi-view images by employing an SMG model, which leads to coarse texturing results with unpainted parts due to missing observations. 2) Spatial-aware 3D Inpainting (S3I). To ensure complete 3D texturing, we introduce the S3I method, specifically designed to effectively texture previously unobserved areas. 3) UV Refinement (UVR). Furthermore, MVPaint employs a UVR module to improve the texture quality in the UV space, which first performs a UV-space Super-Resolution, followed by a Spatial-aware Seam-Smoothing algorithm for revising spatial texturing discontinuities caused by UV unwrapping. Moreover, we establish two T2T evaluation benchmarks: the Objaverse T2T benchmark and the GSO T2T benchmark, based on selected high-quality 3D meshes from the Objaverse dataset and the entire GSO dataset, respectively. Extensive experimental results demonstrate that MVPaint surpasses existing state-of-the-art methods. Notably, MVPaint could generate high-fidelity textures with minimal Janus issues and highly enhanced cross-view consistency.

cs.CV↗

arXiv2024

Interpreting Inflammation Prediction Model via Tag-based Cohort Explanation

Machine learning is revolutionizing nutrition science by enabling systems to learn from data and make intelligent decisions. However, the complexity of these models often leads to challenges in understanding their decision-making processes, necessitating the development of explainability techniques to foster trust and increase model transparency. An under-explored type of explanation is cohort explanation, which provides explanations to groups of instances with similar characteristics. Unlike traditional methods that focus on individual explanations or global model behavior, cohort explainability bridges the gap by providing unique insights at an intermediate granularity. We propose a novel framework for identifying cohorts within a dataset based on local feature importance scores, aiming to generate concise descriptions of the clusters via tags. We evaluate our framework on a food-based inflammation prediction model and demonstrated that the framework can generate reliable explanations that match domain knowledge.

cs.LG↗

arXiv2024

Devil in the Tail: A Multi-Modal Framework for Drug-Drug Interaction Prediction in Long Tail Distinction

Drug-drug interaction (DDI) identification is a crucial aspect of pharmacology research. There are many DDI types (hundreds), and they are not evenly distributed with equal chance to occur. Some of the rarely occurred DDI types are often high risk and could be life-critical if overlooked, exemplifying the long-tailed distribution problem. Existing models falter against this distribution challenge and overlook the multi-faceted nature of drugs in DDI prediction. In this paper, a novel multi-modal deep learning-based framework, namely TFDM, is introduced to leverage multiple properties of a drug to achieve DDI classification. The proposed framework fuses multimodal features of drugs, including graph-based, molecular structure, Target and Enzyme, for DDI identification. To tackle the challenge posed by the distribution skewness across categories, a novel loss function called Tailed Focal Loss is introduced, aimed at further enhancing the model performance and address gradient vanishing problem of focal loss in extremely long-tailed dataset. Intensive experiments over 4 challenging long-tailed dataset demonstrate that the TFMD outperforms the most recent SOTA methods in long-tailed DDI classification tasks. The source code is released to reproduce our experiment results: https://github.com/IcurasLW/TFMD_Longtailed_DDI.git

cs.LG↗

arXiv2024

Exploring and Addressing Reward Confusion in Offline Preference Learning

Spurious correlations in a reward model's training data can prevent Reinforcement Learning from Human Feedback (RLHF) from identifying the desired goal and induce unwanted behaviors. This paper shows that offline RLHF is susceptible to reward confusion, especially in the presence of spurious correlations in offline data. We create a benchmark to study this problem and propose a method that can significantly reduce reward confusion by leveraging transitivity of preferences while building a global preference chain with active learning.

cs.LG↗

arXiv2024

The Brain-Inspired Cooperative Shared Control Framework for Brain-Machine Interface

In brain-machine interface (BMI) applications, a key challenge is the low information content and high noise level in neural signals, severely affecting stable robotic control. To address this challenge, we proposes a cooperative shared control framework based on brain-inspired intelligence, where control signals are decoded from neural activity, and the robot handles the fine control. This allows for a combination of flexible and adaptive interaction control between the robot and the brain, making intricate human-robot collaboration feasible. The proposed framework utilizes spiking neural networks (SNNs) for controlling robotic arm and wheel, including speed and steering. While full integration of the system remains a future goal, individual modules for robotic arm control, object tracking, and map generation have been successfully implemented. The framework is expected to significantly enhance the performance of BMI. In practical settings, the BMI with cooperative shared control, utilizing a brain-inspired algorithm, will greatly enhance the potential for clinical applications.

cs.RO↗

arXiv2024

Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models

Diffusion models (DMs) have achieved remarkable success in text-to-image generation, but they also pose safety risks, such as the potential generation of harmful content and copyright violations. The techniques of machine unlearning, also known as concept erasing, have been developed to address these risks. However, these techniques remain vulnerable to adversarial prompt attacks, which can prompt DMs post-unlearning to regenerate undesired images containing concepts (such as nudity) meant to be erased. This work aims to enhance the robustness of concept erasing by integrating the principle of adversarial training (AT) into machine unlearning, resulting in the robust unlearning framework referred to as AdvUnlearn. However, achieving this effectively and efficiently is highly nontrivial. First, we find that a straightforward implementation of AT compromises DMs' image generation quality post-unlearning. To address this, we develop a utility-retaining regularization on an additional retain set, optimizing the trade-off between concept erasure robustness and model utility in AdvUnlearn. Moreover, we identify the text encoder as a more suitable module for robustification compared to UNet, ensuring unlearning effectiveness. And the acquired text encoder can serve as a plug-and-play robust unlearner for various DM types. Empirically, we perform extensive experiments to demonstrate the robustness advantage of AdvUnlearn across various DM unlearning scenarios, including the erasure of nudity, objects, and style concepts. In addition to robustness, AdvUnlearn also achieves a balanced tradeoff with model utility. To our knowledge, this is the first work to systematically explore robust DM unlearning through AT, setting it apart from existing methods that overlook robustness in concept erasing. Codes are available at: https://github.com/OPTML-Group/AdvUnlearn

cs.CV↗

arXiv2024

MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers

Recently, 3D assets created via reconstruction and generation have matched the quality of manually crafted assets, highlighting their potential for replacement. However, this potential is largely unrealized because these assets always need to be converted to meshes for 3D industry applications, and the meshes produced by current mesh extraction methods are significantly inferior to Artist-Created Meshes (AMs), i.e., meshes created by human artists. Specifically, current mesh extraction methods rely on dense faces and ignore geometric features, leading to inefficiencies, complicated post-processing, and lower representation quality. To address these issues, we introduce MeshAnything, a model that treats mesh extraction as a generation problem, producing AMs aligned with specified shapes. By converting 3D assets in any 3D representation into AMs, MeshAnything can be integrated with various 3D asset production methods, thereby enhancing their application across the 3D industry. The architecture of MeshAnything comprises a VQ-VAE and a shape-conditioned decoder-only transformer. We first learn a mesh vocabulary using the VQ-VAE, then train the shape-conditioned decoder-only transformer on this vocabulary for shape-conditioned autoregressive mesh generation. Our extensive experiments show that our method generates AMs with hundreds of times fewer faces, significantly improving storage, rendering, and simulation efficiencies, while achieving precision comparable to previous methods.

cs.CV↗

arXiv2024

Error Reduction from Stacked Regressions

Stacking regressions is an ensemble technique that forms linear combinations of different regression estimators to enhance predictive accuracy. The conventional approach uses cross-validation data to generate predictions from the constituent estimators, and least-squares with nonnegativity constraints to learn the combination weights. In this paper, we learn these weights analogously by minimizing a regularized version of the empirical risk subject to a nonnegativity constraint. When the constituent estimators are linear least-squares projections onto nested subspaces separated by at least three dimensions, we show that thanks to an adaptive shrinkage effect, the resulting stacked estimator has strictly smaller population risk than best single estimator among them, with more significant gains when the signal-to-noise ratio is small. Here "best" refers to an estimator that minimizes a model selection criterion such as AIC or BIC. In other words, in this setting, the best single estimator is inadmissible. Because the optimization problem can be reformulated as isotonic regression, the stacked estimator requires the same order of computation as the best single estimator, making it an attractive alternative in terms of both performance and implementation.

stat.ML↗

arXiv2024

Understanding Reasoning in Chain-of-Thought from the Hopfieldian View

Large Language Models have demonstrated remarkable abilities across various tasks, with Chain-of-Thought (CoT) prompting emerging as a key technique to enhance reasoning capabilities. However, existing research primarily focuses on improving performance, lacking a comprehensive framework to explain and understand the fundamental factors behind CoT's success. To bridge this gap, we introduce a novel perspective grounded in the Hopfieldian view of cognition in cognitive neuroscience. We establish a connection between CoT reasoning and key cognitive elements such as stimuli, actions, neural populations, and representation spaces. From our view, we can understand the reasoning process as the movement between these representation spaces. Building on this insight, we develop a method for localizing reasoning errors in the response of CoTs. Moreover, we propose the Representation-of-Thought (RoT) framework, which leverages the robustness of low-dimensional representation spaces to enhance the robustness of the reasoning process in CoTs. Experimental results demonstrate that RoT improves the robustness and interpretability of CoT reasoning while offering fine-grained control over the reasoning process.

cs.AI↗

arXiv2024

A Personality-Guided Preference Aggregator for Ephemeral Group Recommendation

Ephemeral group recommendation (EGR) aims to suggest items for a group of users who come together for the first time. Existing work typically consider individual preferences as the sole factor in aggregating group preferences. However, they neglect to take into account the importance of the individual inherent factors, such as personality, and thus fail to accurately simulate the group decision-making process. Additionally, these methods often struggle due to insufficient interactive records. To tackle these issues, a Personality-Guided Preference Aggregator (PEGA) is proposed, which guides the preference aggregation of group members based on their personalities, rather than relying solely on their preferences. Specifically, implicit personalities are first extracted from user reviews. Hyper-rectangles are then used to aggregate individual personalities to obtain the "Group Personality", which allows for the learning of personality distributions within the group. Subsequently, a personality attention mechanism is employed to aggregate group preferences, and a preference-based fine-tuning module is used to balance the weights of personality and preferences. The role of personality in this approach is twofold: (1) To estimate the importance of individual users in a group and provide explainability; (2) To alleviate the data sparsity issue encountered in ephemeral groups. Experimental results demonstrate that, on four real-world datasets, the PEGA model significantly outperforms related baseline models in terms of classification accuracy and interpretability. Moreover, empirical evidence supports the idea that personality plays a pivotal role in enhancing the performance of EGR tasks.

cs.IR↗

arXiv2024

Off-Policy Evaluation with Irregularly-Spaced, Outcome-Dependent Observation Times

While the classic off-policy evaluation (OPE) literature commonly assumes decision time points to be evenly spaced for simplicity, in many real-world scenarios, such as those involving user-initiated visits, decisions are made at irregularly-spaced and potentially outcome-dependent time points. For a more principled evaluation of the dynamic policies, this paper constructs a novel OPE framework, which concerns not only the state-action process but also an observation process dictating the time points at which decisions are made. The framework is closely connected to the Markov decision process in computer science and with the renewal process in the statistical literature. Within the framework, two distinct value functions, derived from cumulative reward and integrated reward respectively, are considered, and statistical inference for each value function is developed under revised Markov and time-homogeneous assumptions. The validity of the proposed method is further supported by theoretical results, simulation studies, and a real-world application from electronic health records (EHR) evaluating periodontal disease treatments.

stat.ME↗

arXiv2024

Quantitative periodic homogenization for symmetric non-local stable-like operators

Homogenization for non-local operators in periodic environments has been studied intensively. So far, these works are mainly devoted to the qualitative results, that is, to determine explicitly the operators in the limit. To the best of authors' knowledge, there is no result concerning the convergence rates of the homogenization for stable-like operators in periodic environments. In this paper, we establish a quantitative homogenization result for symmetric $α$-stable-like operators on $\R^d$ with periodic coefficients. In particular, we show that the convergence rate for the solutions of associated Dirichlet problems on a bounded domain $D$ is of order $$ \varepsilon^{(2-α)/2}\I_{\{α\in (1,2)\}}+\varepsilon^{α/2}\I_{\{α\in (0,1)\}}+\varepsilon^{1/2}|\log \e|^2\I_{\{α=1\}}, $$ while, when the solution to the equation in the limit is in $C^2_c(D)$, the convergence rate becomes $$ \varepsilon^{2-α}\I_{\{α\in (1,2)\}}+\varepsilon^α\I_{\{α\in (0,1)\}}+\varepsilon |\log \e|^2\I_{\{α=1\}}. $$ This indicates that the boundary decay behaviors of the solution to the equation in the limit affects the convergence rate in the homogenization.

math.AP↗

arXiv2024

Spherical-like spectra for the description of the normal states of $^{108-120}$Cd in the SU3-IBM and the $Q_{2_{1}^{+}}$ anomaly

The Cd puzzle implies that the phonon excitation of a spherical nucleus should be questioned and refuted. For understanding this spherical-like $γ$-soft mode newly found experimentally, a possible answer was proposed recently in the SU3-IBM. In this paper, the evolutions of the normal states in $^{108-120}$Cd are investigated and compared with the experimental results. For better explaining the nearly zero B(E2) values between the $0_{2}^{+}$ state and the $2_{1}^{+}$ state, except for the SU(3) second-order and third-order Casimir operators, other SU(3) higher-order interactions are also considered in detail. It can be found that the results of theoretical fitting and experimental data agree well with simple parameter selection. The spherical-like spectra really exist. The deficiency may come from the lack of configuration mixing. The realistic spectra characteristics of the spherical-like spectra are found for $^{118,120}$Cd and the electric quadrupole moments of the $2_{1}^{+}$ state are predicted. The $Q_{2_{1}^{+}}$ anomaly in $^{108-116}$Cd are also discussed.

nucl-th↗