← Search

Xiaofeng Zhang

39 accepted papers

2026

Context Tokens are Anchors: Understanding the Repetition Curse in Diffusion MLLMs from an Information Flow Perspective

ICLR 2026poster

Recent diffusion-based Multimodal Large Language Models (dMLLMs) suffer from high inference latency and therefore rely on caching techniques to accelerate decoding. However, the application of cache mechanisms often introduces undesirable repetitive text generation, a phenomenon we term the Repeat C…

Cited by 0SourcecodeScholar
2026

D3ToM: Decider-Guided Dynamic Token Merging for Accelerating Diffusion MLLMs

AAAI 2026technical

Diffusion-based multimodal large language models (Diffusion MLLMs) have recently demonstrated impressive non-autoregressive generative capabilities across vision-and-language tasks. However, Diffusion MLLMs exhibit substantially slower inference than autoregressive models: Each denoising step employ

Cited by 0SourcePDFScholar
2026

Differential Fine-Tuning Large Language Models Towards Better Diverse Reasoning Abilities

ICLR 2026poster

Reasoning abilities of large language models (LLMs) require explicit derivations compared to general question-answering, supervised fine-tuning (SFT) can empower multiple reasoning abilities in LLMs via learning from various datasets. However, neither training the datasets jointly (mix-up) nor conti…

Cited by 0SourcecodeScholar
2026

Diversity Recommendation via Causal Deconfounding of Co-purchase Relations and Counterfactual Exposure

AAAI 2026technical

Beyond user-item modeling, item-to-item relationships are increasingly used to enhance recommendation. However, common methods largely rely on co-occurrence, making them prone to item popularity bias and user attributes, which degrades embedding quality and performance. Meanwhile, although diversity

Cited by 0SourcePDFScholar
2026

Hallucination Begins Where Saliency Drops

ICLR 2026oral

Recent studies have investigated attention dynamics in large vision language models (LVLMs), yet existing methods remain limited in reliably distinguishing hallucinated from correct outputs — primarily because they rely solely on forward-pass attention, ignoring gradient-based signals that reveal ho…

Cited by 0SourcecodeScholar
2026

Inference-time Physics Alignment of Video Generative Models with Latent World Models

CVPR 2026

State-of-the-art video generative models produce promising visual content yet often violate basic physics principles, limiting their utility. While some attribute this deficiency to insufficient physics understanding from pre-training, we find that the shortfall in physics plausibility also stems fr

Cited by 0SourcecodeScholar
2026

Remember Me: Bridging the Long-Range Gap in LVLMs with Three-Step Inference-Only Decay Resilience Strategies

AAAI 2026technical

Large Vision-Language Models (LVLMs) have achieved impressive performance across a wide range of multimodal tasks. However, they still face critical challenges in modeling long-range dependencies under the usage of Rotary Positional Encoding (ROPE). Although it can facilitate precise modeling of tok

Cited by 0SourcePDFScholar
2026

Rethinking LLM Reasoning: From Explicit Trajectories to Latent Representations

ICLR 2026poster

Large Language Models (LLMs) have achieved impressive performance on complex tasks by generating human-like, step-by-step rationales, referred to as \textit{reasoning trajectory}, before arriving at final answers. However, the length of these reasoning trajectories often far exceeds that of the fina…

Cited by 0SourcecodeScholar
2026

SGMT: Social Generating with Multiview-Guided Tuning In Recommender Systems

AAAI 2026technical

The sparsity of user–item interactions remains a fundamental obstacle in collaborative filtering, limiting the ability of Graph Neural Network (GNN)-based recommender systems to capture high-order user relationships without incurring over-smoothing and computational overhead. Existing social recomme

Cited by 0SourcePDFScholar
2026

SoPE: Spherical Coordinate-Based Positional Embedding for Enhancing Spatial Perception of 3D LVLMs

CVPR 2026

3D Large Vision-Language Models (3D LVLMs) built upon Large Language Models (LLMs) have achieved remarkable progress across various multimodal tasks. However, their inherited position-dependent modeling mechanism, Rotary Position Embedding (RoPE), remains suboptimal for 3D multimodal understanding.

Cited by 0SourceScholar
2026

The Intricate Dance of Prompt Complexity, Quality, Diversity and Consistency in T2I Models

ICLR 2026poster

Text-to-image (T2I) models offer great potential for creating virtually limitless synthetic data, a valuable resource compared to fixed and finite real datasets. Previous works evaluate the utility of synthetic data from T2I models on three key desiderata: quality, diversity, and consistency. While…

Cited by 0SourceScholar
2025

CURV: Coherent Uncertainty-Aware Reasoning in Vision-Language Models for X-Ray Report Generation

NeurIPS 2025poster

Vision-language models have been explored for radiology report generation with promising results. Yet, uncertainty elaborated in findings and the reasoning process for reaching clinical impressions are seldom explicitly modeled, reducing the clinical accuracy and trustworthiness of the generated rep…

Cited by 0SourceScholar
2025

CoDeR: Counterfactual Demand Reasoning for Sequential Recommendation

AAAI 2025technical

Sequential recommendation systems aim to predict the next item based on users' historical interactions. While traditional methods focus on learning feature representations or user preferences, they often struggle with detecting subtle demand shifts in short sequences, especially when these shifts ar…

2025

CulturalFrames: Assessing Cultural Expectation Alignment in Text-to-Image Models and Evaluation Metrics

EMNLP 2025

The increasing ubiquity of text-to-image (T2I) models as tools for visual content generation raises concerns about their ability to accurately represent diverse cultural contexts - where missed cues can stereotype communities and undermine usability. In this work, we present the first study to syste

Cited by 0SourcePDFScholar
2025

DAPE-BR: Distance-Aware Positional Encoding for Mitigating Object Hallucination in LVLMs

EMNLP 2025

Large Vision–Language Models (LVLMs) have garnered substantial interest owing to their impressive ability to interpret visual inputs and converse with users.Nevertheless, LVLMs still suffer from object hallucination – generating descriptions for objects that are absent from the image, which undermin

Cited by 0SourcePDFScholar
2025

EFDTR: Learnable Elliptical Fourier Descriptor Transformer for Instance Segmentation

ICML 2025poster

Polygon-based object representations efficiently model object boundaries but are limited by high optimization complexity, which hinders their adoption compared to more flexible pixel-based methods. In this paper, we introduce a novel vertex regression loss grounded in Fourier elliptic descriptors,…

2025

Enhancing Multimodal Large Language Models Complex Reason via Similarity Computation

AAAI 2025technical

Multimodal large language models have experienced rapid growth, and numerous different models have emerged. The interpretability of LVLMs remains an under-explored area. Especially when faced with more complex tasks such as chain-of-thought reasoning, its internal mechanisms still resemble a black b…

2025

Fine-Grained Preference Optimization Improves Spatial Reasoning in VLMs

NeurIPS 2025poster

Current Vision-Language Models (VLMs) struggle with fine-grained spatial reasoning, particularly when multi-step logic and precise spatial alignment are required. In this work, we introduce SpatialReasoner-R1, a vision-language reasoning model designed to address these limitations. To construct high…

Cited by 0SourceScholar
2025

From Redundancy to Relevance: Information Flow in LVLMs Across Reasoning Tasks

NAACL 2025long

Large Vision Language Models (LVLMs) achieve great performance on visual-language reasoning tasks, however, the black-box nature of LVLMs hinders in-depth research on the reasoning mechanism. As all images need to be converted into image tokens to fit the input format of large language models (LLMs)…

2025

Improving Complex Reasoning with Dynamic Prompt Corruption: A Soft Prompt Optimization Approach

ICLR 2025poster

Prompt Tuning (PT) has emerged as a promising Parameter-Efficient Fine-Tuning (PEFT) approach by appending trainable continuous prompt vectors to the input, maintaining competitive performance with significantly fewer trainable parameters. While PT has shown effectiveness in enhancing task performan…

Cited by 0SourcePDFScholar
2025

Increasing the Utility of Synthetic Images through Chamfer Guidance

NeurIPS 2025poster

Conditional image generative models hold considerable promise to produce infinite amounts of synthetic training data. Yet, recent progress in generation quality has come at the expense of generation diversity, limiting the utility of these models as a source of synthetic training data. Although gui…

Cited by 0SourceScholar
2025

LensNet: An End-to-End Learning Framework for Empirical Point Spread Function Modeling and Lensless Imaging Reconstruction

IJCAI 2025

Lensless imaging stands out as a promising alternative to conventional lens-based systems, particularly in scenarios demanding ultracompact form factors and cost-effective architectures. However, such systems are fundamentally governed by the Point Spread Function (PSF), which dictates how a point s

2025

SalaMAnder: Shapley-based Mathematical Expression Attribution and Metric for Chain-of-Thought Reasoning

EMNLP 2025

Chain-of-Thought (CoT) prompting enhances the math reasoning capability of large language models (LLMs) to a large margin. However, the mechanism underlying such improvements remains unexplored. In this paper, we present SalaMAnder ( S h a p l ey-b a sed M athematical Expression A ttribution a nd M

Cited by 0SourcePDFScholar
2025

Shallow Focus, Deep Fixes: Enhancing Shallow Layers Vision Attention Sinks to Alleviate Hallucination in LVLMs

EMNLP 2025

Multimodal large language models (MLLMs) demonstrate excellent abilities for understanding visual information, while the hallucination remains. Albeit image tokens constitute the majority of the MLLMs input, the relation between image tokens and hallucinations is still unexplored. In this paper, we

Cited by 0SourcePDFScholar
2024

Beyond What If: Advancing Counterfactual Text Generation with Structural Causal Modeling

IJCAI 2024poster

Exploring the realms of counterfactuals, this paper introduces a versatile approach in text generation using structural causal models (SCM), broadening the scope beyond traditional singular causal studies to encompass complex, multi-layered relationships. To comprehensively explore these intricate,…

Cited by 2SourcePDFScholar
2024

Causal-Story: Local Causal Attention Utilizing Parameter-Efficient Tuning for Visual Story Synthesis

ICASSP 2024accepted

The excellent text-to-image synthesis capability of diffusion models has driven progress in synthesizing coherent visual stories. The current state-of-the-art method combines the features of historical captions, historical frames, and the current captions as conditions for generating the current fra…

Cited by 0SourceScholar
2024

Instance-adaptive Zero-shot Chain-of-Thought Prompting

NeurIPS 2024poster

Zero-shot Chain-of-Thought (CoT) prompting emerges as a simple and effective strategy for enhancing the performance of large language models (LLMs) in real-world reasoning tasks. Nonetheless, the efficacy of a singular, task-level prompt uniformly applied across the whole of instances is inherently…

Cited by 5SourcePDFScholar
2024

Wavelet-Decoupling Contrastive Enhancement Network for Fine-Grained Skeleton-Based Action Recognition

ICASSP 2024accepted

Skeleton-based action recognition has attracted much attention, benefiting from its succinctness and robustness. However, the minimal inter-class variation in similar action sequences often leads to confusion. The inherent spatiotemporal coupling characteristics make it challenging to mine the subtl…

Cited by 0SourceScholar
2023

Beyond Pure Text: Summarizing Financial Reports Based on Both Textual and Tabular Data

IJCAI 2023poster

Abstractive text summarization is to generate concise summaries that well preserve both salient information and the overall semantic meanings of the given documents. However, real-world documents, e.g., financial reports, generally contain rich data such as charts and tabular data which invalidates…

Cited by 1SourcePDFScholar
2023

How to Determine the Most Powerful Pre-trained Language Model without Brute Force Fine-tuning? An Empirical Survey

EMNLP 2023long findings

Transferability estimation has been attached to great attention in the computer vision fields. Researchers try to estimate with low computational cost the performance of a model when transferred from a source task to a given target task. Considering the effectiveness of such estimations, the commun…

Cited by 0SourcecodeScholar
2023

Transformer-Patcher: One Mistake Worth One Neuron

ICLR 2023poster

Large Transformer-based Pretrained Language Models (PLMs) dominate almost all Natural Language Processing (NLP) tasks. Nevertheless, they still make mistakes from time to time. For a model deployed in an industrial environment, fixing these mistakes quickly and robustly is vital to improve user expe…

2022

Mixture of Attention Heads: Selecting Attention Heads Per Token

EMNLP 2022main

Mixture-of-Experts (MoE) networks have been proposed as an efficient way to scale up model capacity and implement conditional computing. However, the study of MoE components mostly focused on the feedforward layer in Transformer architecture. This paper proposes the Mixture of Attention Heads (MoA),…

2022

Prompt-Based Meta-Learning For Few-shot Text Classification

EMNLP 2022main

Few-shot Text Classification predicts the semantic label of a given text with a handful of supporting instances. Current meta-learning methods have achieved satisfying results in various few-shot situations. Still, they often require a large amount of data to construct many few-shot tasks for meta-t…

2021

Demodalizing Face Recognition with Synthetic Samples

AAAI 2021technical

Using data generated by generative adversarial networks or three-dimensional (3D) technology for face recognition training is a theoretically reasonable solution to the problems of unbalanced data distributions and data scarcity. However, due to the modal difference between synthetic data and real d…

2019

DADA: Deep Adversarial Data Augmentation for Extremely Low Data Regime Classification

ICASSP 2019accepted

Deep learning has revolutionized the performance of classification, but meanwhile demands sufficient labeled data for training. Given insufficient data, while many techniques have been developed to help combat overfitting, the challenge remains if one tries to train deep networks, especially in the…

Cited by 0SourceScholar