arXiv subjects

Xinyu Wei

Publications and source records attributed to Xinyu Wei.

33 records · Page 2Linked to original sources

arXiv2025

Separation and Collaboration: Two-Level Routing Grouped Mixture-of-Experts for Multi-Domain Continual Learning

Multi-Domain Continual Learning (MDCL) acquires knowledge from sequential tasks with shifting class sets and distribution. Despite the Parameter-Efficient Fine-Tuning (PEFT) methods can adapt for this dual heterogeneity, they still suffer from catastrophic forgetting and forward forgetting. To address these challenges, we propose a Two-Level Routing Grouped Mixture-of-Experts (TRGE) method. Firstly, TRGE dynamically expands the pre-trained CLIP model, assigning specific expert group for each task to mitigate catastrophic forgetting. With the number of experts continually grows in this process, TRGE maintains the static experts count within the group and introduces the intra-group router to alleviate routing overfitting caused by the increasing routing complexity. Meanwhile, we design an inter-group routing policy based on task identifiers and task prototype distance, which dynamically selects relevant expert groups and combines their outputs to enhance inter-task collaboration. Secondly, to get the correct task identifiers, we leverage Multimodal Large Language Models (MLLMs) which own powerful multimodal comprehension capabilities to generate semantic task descriptions and recognize the correct task identifier. Finally, to mitigate forward forgetting, we dynamically fuse outputs for unseen samples from the frozen CLIP model and TRGE adapter based on training progress, leveraging both pre-trained and learned knowledge. Through extensive experiments across various settings, our method outperforms other advanced methods with fewer trainable parameters.

cs.LG↗

arXiv2025

Delving into RL for Image Generation with CoT: A Study on DPO vs. GRPO

Recent advancements underscore the significant role of Reinforcement Learning (RL) in enhancing the Chain-of-Thought (CoT) reasoning capabilities of large language models (LLMs). Two prominent RL algorithms, Direct Preference Optimization (DPO) and Group Relative Policy Optimization (GRPO), are central to these developments, showcasing different pros and cons. Autoregressive image generation, also interpretable as a sequential CoT reasoning process, presents unique challenges distinct from LLM-based CoT reasoning. These encompass ensuring text-image consistency, improving image aesthetic quality, and designing sophisticated reward models, rather than relying on simpler rule-based rewards. While recent efforts have extended RL to this domain, these explorations typically lack an in-depth analysis of the domain-specific challenges and the characteristics of different RL strategies. To bridge this gap, we provide the first comprehensive investigation of the GRPO and DPO algorithms in autoregressive image generation, evaluating their in-domain performance and out-of-domain generalization, while scrutinizing the impact of different reward models on their respective capabilities. Our findings reveal that GRPO and DPO exhibit distinct advantages, and crucially, that reward models possessing stronger intrinsic generalization capabilities potentially enhance the generalization potential of the applied RL algorithms. Furthermore, we systematically explore three prevalent scaling strategies to enhance both their in-domain and out-of-domain proficiency, deriving unique insights into efficiently scaling performance for each paradigm. We hope our study paves a new path for inspiring future work on developing more effective RL algorithms to achieve robust CoT reasoning in the realm of autoregressive image generation. Code is released at https://github.com/ZiyuGuo99/Image-Generation-CoT

cs.CV↗

arXiv2025

Perceive Anything: Recognize, Explain, Caption, and Segment Anything in Images and Videos

We present Perceive Anything Model (PAM), a conceptually straightforward and efficient framework for comprehensive region-level visual understanding in images and videos. Our approach extends the powerful segmentation model SAM 2 by integrating Large Language Models (LLMs), enabling simultaneous object segmentation with the generation of diverse, region-specific semantic outputs, including categories, label definition, functional explanations, and detailed captions. A key component, Semantic Perceiver, is introduced to efficiently transform SAM 2's rich visual features, which inherently carry general vision, localization, and semantic priors into multi-modal tokens for LLM comprehension. To support robust multi-granularity understanding, we also develop a dedicated data refinement and augmentation pipeline, yielding a high-quality dataset of 1.5M image and 0.6M video region-semantic annotations, including novel region-level streaming video caption data. PAM is designed for lightweightness and efficiency, while also demonstrates strong performance across a diverse range of region understanding tasks. It runs 1.2-2.4x faster and consumes less GPU memory than prior approaches, offering a practical solution for real-world applications. We believe that our effective approach will serve as a strong baseline for future research in region-level visual understanding.

cs.CV↗

arXiv2025

Are Large Language Models Good In-context Learners for Financial Sentiment Analysis?

Recently, large language models (LLMs) with hundreds of billions of parameters have demonstrated the emergent ability, surpassing traditional methods in various domains even without fine-tuning over domain-specific data. However, when it comes to financial sentiment analysis (FSA)$\unicode{x2013}$a fundamental task in financial AI$\unicode{x2013}$these models often encounter various challenges, such as complex financial terminology, subjective human emotions, and ambiguous inclination expressions. In this paper, we aim to answer the fundamental question: whether LLMs are good in-context learners for FSA? Unveiling this question can yield informative insights on whether LLMs can learn to address the challenges by generalizing in-context demonstrations of financial document-sentiment pairs to the sentiment analysis of new documents, given that finetuning these models on finance-specific data is difficult, if not impossible at all. To the best of our knowledge, this is the first paper exploring in-context learning for FSA that covers most modern LLMs (recently released DeepSeek V3 included) and multiple in-context sample selection methods. Comprehensive experiments validate the in-context learning capability of LLMs for FSA.

cs.CL↗

arXiv2025

PixWizard: Versatile Image-to-Image Visual Assistant with Open-Language Instructions

This paper presents a versatile image-to-image visual assistant, PixWizard, designed for image generation, manipulation, and translation based on free-from language instructions. To this end, we tackle a variety of vision tasks into a unified image-text-to-image generation framework and curate an Omni Pixel-to-Pixel Instruction-Tuning Dataset. By constructing detailed instruction templates in natural language, we comprehensively include a large set of diverse vision tasks such as text-to-image generation, image restoration, image grounding, dense image prediction, image editing, controllable generation, inpainting/outpainting, and more. Furthermore, we adopt Diffusion Transformers (DiT) as our foundation model and extend its capabilities with a flexible any resolution mechanism, enabling the model to dynamically process images based on the aspect ratio of the input, closely aligning with human perceptual processes. The model also incorporates structure-aware and semantic-aware guidance to facilitate effective fusion of information from the input image. Our experiments demonstrate that PixWizard not only shows impressive generative and understanding abilities for images with diverse resolutions but also exhibits promising generalization capabilities with unseen tasks and human instructions. The code and related resources are available at https://github.com/AFeng-x/PixWizard

cs.CV↗

arXiv2025

Draw-and-Understand: Leveraging Visual Prompts to Enable MLLMs to Comprehend What You Want

In this paper, we present the Draw-and-Understand framework, exploring how to integrate visual prompting understanding capabilities into Multimodal Large Language Models (MLLMs). Visual prompts allow users to interact through multi-modal instructions, enhancing the models' interactivity and fine-grained image comprehension. In this framework, we propose a general architecture adaptable to different pre-trained MLLMs, enabling it to recognize various types of visual prompts (such as points, bounding boxes, and free-form shapes) alongside language understanding. Additionally, we introduce MDVP-Instruct-Data, a multi-domain dataset featuring 1.2 million image-visual prompt-text triplets, including natural images, document images, scene text images, mobile/web screenshots, and remote sensing images. Building on this dataset, we introduce MDVP-Bench, a challenging benchmark designed to evaluate a model's ability to understand visual prompting instructions. The experimental results demonstrate that our framework can be easily and effectively applied to various MLLMs, such as SPHINX-X and LLaVA. After training with MDVP-Instruct-Data and image-level instruction datasets, our models exhibit impressive multimodal interaction capabilities and pixel-level understanding, while maintaining their image-level visual perception performance.

cs.CV↗

arXiv2024

MAVIS: Mathematical Visual Instruction Tuning with an Automatic Data Engine

The mathematical capabilities of Multi-modal Large Language Models (MLLMs) remain under-explored with three areas to be improved: visual encoding of math diagrams, diagram-language alignment, and chain-of-thought (CoT) reasoning. This draws forth an urgent demand for an effective training paradigm and a large-scale, comprehensive dataset with detailed CoT rationales, which is challenging to collect and costly to annotate manually. To tackle this issue, we propose MAVIS, a MAthematical VISual instruction tuning pipeline for MLLMs, featuring an automatic data engine to efficiently create mathematical visual datasets. We design the data generation process to be entirely independent of human intervention or GPT API usage, while ensuring the diagram-caption correspondence, question-answer correctness, and CoT reasoning quality. With this approach, we curate two datasets, MAVIS-Caption (558K diagram-caption pairs) and MAVIS-Instruct (834K visual math problems with CoT rationales), and propose four progressive stages for training MLLMs from scratch. First, we utilize MAVIS-Caption to fine-tune a math-specific vision encoder (CLIP-Math) through contrastive learning, tailored for improved diagram visual encoding. Second, we also leverage MAVIS-Caption to align the CLIP-Math with a large language model (LLM) by a projection layer, enhancing vision-language alignment in mathematical domains. Third, we adopt MAVIS-Instruct to perform the instruction tuning for robust problem-solving skills, and term the resulting model as MAVIS-7B. Fourth, we apply Direct Preference Optimization (DPO) to enhance the CoT capabilities of our model, further refining its step-wise reasoning performance. Code and data will be released at https://github.com/ZrrSkywalker/MAVIS

cs.CV↗

arXiv2024

MR-MLLM: Mutual Reinforcement of Multimodal Comprehension and Vision Perception

In recent years, multimodal large language models (MLLMs) have shown remarkable capabilities in tasks like visual question answering and common sense reasoning, while visual perception models have made significant strides in perception tasks, such as detection and segmentation. However, MLLMs mainly focus on high-level image-text interpretations and struggle with fine-grained visual understanding, and vision perception models usually suffer from open-world distribution shifts due to their limited model capacity. To overcome these challenges, we propose the Mutually Reinforced Multimodal Large Language Model (MR-MLLM), a novel framework that synergistically enhances visual perception and multimodal comprehension. First, a shared query fusion mechanism is proposed to harmonize detailed visual inputs from vision models with the linguistic depth of language models, enhancing multimodal comprehension and vision perception synergistically. Second, we propose the perception-enhanced cross-modal integration method, incorporating novel modalities from vision perception outputs, like object detection bounding boxes, to capture subtle visual elements, thus enriching the understanding of both visual and textual data. In addition, an innovative perception-embedded prompt generation mechanism is proposed to embed perceptual information into the language model's prompts, aligning the responses contextually and perceptually for a more accurate multimodal interpretation. Extensive experiments demonstrate MR-MLLM's superior performance in various multimodal comprehension and vision perception tasks, particularly those requiring corner case vision perception and fine-grained language comprehension.

cs.CV↗

arXiv2024

IDAdapter: Learning Mixed Features for Tuning-Free Personalization of Text-to-Image Models

Leveraging Stable Diffusion for the generation of personalized portraits has emerged as a powerful and noteworthy tool, enabling users to create high-fidelity, custom character avatars based on their specific prompts. However, existing personalization methods face challenges, including test-time fine-tuning, the requirement of multiple input images, low preservation of identity, and limited diversity in generated outcomes. To overcome these challenges, we introduce IDAdapter, a tuning-free approach that enhances the diversity and identity preservation in personalized image generation from a single face image. IDAdapter integrates a personalized concept into the generation process through a combination of textual and visual injections and a face identity loss. During the training phase, we incorporate mixed features from multiple reference images of a specific identity to enrich identity-related content details, guiding the model to generate images with more diverse styles, expressions, and angles compared to previous works. Extensive evaluations demonstrate the effectiveness of our method, achieving both diversity and identity fidelity in generated images.

cs.CV↗

arXiv2023

Cloud-Device Collaborative Learning for Multimodal Large Language Models

The burgeoning field of Multimodal Large Language Models (MLLMs) has exhibited remarkable performance in diverse tasks such as captioning, commonsense reasoning, and visual scene understanding. However, the deployment of these large-scale MLLMs on client devices is hindered by their extensive model parameters, leading to a notable decline in generalization capabilities when these models are compressed for device deployment. Addressing this challenge, we introduce a Cloud-Device Collaborative Continual Adaptation framework, designed to enhance the performance of compressed, device-deployed MLLMs by leveraging the robust capabilities of cloud-based, larger-scale MLLMs. Our framework is structured into three key components: a device-to-cloud uplink for efficient data transmission, cloud-based knowledge adaptation, and an optimized cloud-to-device downlink for model deployment. In the uplink phase, we employ an Uncertainty-guided Token Sampling (UTS) strategy to effectively filter out-of-distribution tokens, thereby reducing transmission costs and improving training efficiency. On the cloud side, we propose Adapter-based Knowledge Distillation (AKD) method to transfer refined knowledge from large-scale to compressed, pocket-size MLLMs. Furthermore, we propose a Dynamic Weight update Compression (DWC) strategy for the downlink, which adaptively selects and quantizes updated weight parameters, enhancing transmission efficiency and reducing the representational disparity between cloud and device models. Extensive experiments on several multimodal benchmarks demonstrate the superiority of our proposed framework over prior Knowledge Distillation and device-cloud collaboration methods. Notably, we also validate the feasibility of our approach to real-world experiments.

cs.CV↗

arXiv2021

Accretionary Learning with Deep Neural Networks

One of the fundamental limitations of Deep Neural Networks (DNN) is its inability to acquire and accumulate new cognitive capabilities. When some new data appears, such as new object classes that are not in the prescribed set of objects being recognized, a conventional DNN would not be able to recognize them due to the fundamental formulation that it takes. The current solution is typically to re-design and re-learn the entire network, perhaps with a new configuration, from a newly expanded dataset to accommodate new knowledge. This process is quite different from that of a human learner. In this paper, we propose a new learning method named Accretionary Learning (AL) to emulate human learning, in that the set of objects to be recognized may not be pre-specified. The corresponding learning structure is modularized, which can dynamically expand to register and use new knowledge. During accretionary learning, the learning process does not require the system to be totally re-designed and re-trained as the set of objects grows in size. The proposed DNN structure does not forget previous knowledge when learning to recognize new data classes. We show that the new structure and the design methodology lead to a system that can grow to cope with increased cognitive complexity while providing stable and superior overall performance.

cs.LG↗

arXiv2020

Analytical Model for the Transient Permittivity of Uncured TiO2 Whisker/Liquid Silicone Rubber Composites Under an AC Electric Field

The electric field grading of dielectric permittivity gradient devices is an effective way of enhancing their insulation properties. The in-situ electric field-driven assembly is an advanced method for the fabrication of insulation devices with adaptive permittivity gradients, however, there is no theoretical guidance for design. In this work, an analytical model with a time constant is developed to determine the transient permittivity of uncured composites under an applied AC electric field. This model is based on optical image and dielectric permittivity monitoring, which avoids the direct processing of complex electrodynamics. For a composite with given components, the increased filler content and electric field strength can accelerate the transient process. Compared with the finite element method (FEM) based on differential equations, this statistical model is simple but efficient, and can be applied to any low-viscosity uncured composites, which may contain multiple fillers. More importantly, when a voltage is applied to an uncured composite insulating device, the proposed model can be used to analyse the spatiotemporal permittivity characteristics of this device and optimise its permittivity gradient for electric field grading.

cond-mat.mtrl-sci↗

arXiv2019

Spatiotemporal Attention Networks for Wind Power Forecasting

Wind power is one of the most important renewable energy sources and accurate wind power forecasting is very significant for reliable and economic power system operation and control strategies. This paper proposes a novel framework with spatiotemporal attention networks (STAN) for wind power forecasting. This model captures spatial correlations among wind farms and temporal dependencies of wind power time series. First of all, we employ a multi-head self-attention mechanism to extract spatial correlations among wind farms. Then, temporal dependencies are captured by the Sequence-to-Sequence (Seq2Seq) model with a global attention mechanism. Finally, experimental results demonstrate that our model achieves better performance than other baseline approaches. Our work provides useful insights to capture non-Euclidean spatial correlations.

cs.LG↗

arXiv2019

Camouflage Design of Analysis Based on HSV Color Statistics and K-means Clustering

Since ancient times, it has been essential to adopting camouflage on the battlefield, whether it is in the forefront, in-depth or the rear. The traditional evaluation method is made up of people opinion. By watching target or looking at the pictures, and determine the effect of camouflage, so it can be more influenced by man's subjective factors. And now, in order to objectively reflect the camouflage effect, we set up a model through using images similarity to evaluate camouflage effect. Image similarity comparison is divided into two main image feature comparison: image color features and texture features of images. We now using computer design camouflage, camouflage pattern design is divided into two aspects of design color and design plaques. For the design of the color, we based on HSV color model, and as for the design of plague, the key steps are the background color edge extraction, we adopt algorithm based on k-means clustering analysis of the method of background color edge extraction.

cs.GR↗

arXiv2019

A Books Recommendation Approach Based on Online Bookstore Data

In the era of information explosion, facing complex information, it is difficult for users to choose the information of interest, and businesses also need detailed information on ways to let the ad stand out. By this time, it is recommended that a good way. We firstly by using random interviews, simulations, asking experts, summarizes methods outlined the main factors affecting the scores of books that users drew. In order to further illustrate the impact of these factors, we also by combining the AHP consistency test, then fuzzy evaluation method, empowered each factor, influencing factors and the degree of influence come. For the second question, predict user evaluation of the listed books from the predict annex. First, given the books Annex labels, user data extraction scorebooks and mathematical analysis of data obtained from SPSS user preferences and then use software to nearest neighbor analysis to result in predicted value.

cs.DB↗