Author name cluster

Xinran Wang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

11 papers

2 author rows

YNIMG Journal 2026 Journal Article

The impact of downsampling on data quality, univariate measurement and multivariate pattern analysis in event-related potential research

Guanghui Zhang
Xinran Wang
Ying Xin
Fengyu Cong
Weiqi He
Wenbo Luo

The choice of sampling rate is a critical preprocessing step in event-related potential (ERP) research, yet its impact on different analytic approaches remains underexplored. In this study, we systematically evaluated how downsampling affects data quality measured via Standardized Measurement Error (SME), conventional univariate ERP metrics (mean amplitude, peak amplitude, peak latency, and 50% area latency), and multivariate pattern analysis (MVPA; decoding). We analyzed seven commonly studied ERP components: P3, N400, N170, N2pc, mismatch negativity, error-related negativity, and lateralized readiness potential collected from neurotypical young adults. Across omnibus analyses, sampling rate did not produce significant global effects on data quality, conventional ERP metrics, or decoding performance within the tested range (64-1024 Hz). However, exploratory pairwise comparisons revealed selective, measure-specific differences at lower sampling rates. In particular, latency-based measures such as 50% area latency showed increased SME at 64 Hz, suggesting reduced temporal precision under coarse sampling. Effect sizes for most ERP measures remained stable at 128 Hz and above, with noticeable attenuation primarily at 64 Hz. In contrast, multivariate decoding performance was highly robust across sampling rates, with both classification accuracy and effect sizes remaining stable even at 64 Hz. Together, these findings indicate that sampling rate does not exert a systematic influence on ERP or decoding metrics within the commonly used range, although very low sampling rates may selectively affect latency-sensitive measures. For studies focusing on conventional ERP analyses, moderate-to-high sampling rates are advisable when precise temporal estimates are required. In contrast, lower sampling rates may be sufficient for decoding analyses when fine-grained temporal precision is not essential. For researchers analyzing ERP data with similar components, intra-individual variability levels, and participant populations as in this study, following these recommendations should yield robust statistical power.

Details DOI

NeurIPS Conference 2025 Conference Paper

Beyond Expectations: Quantile-Guided Alignment for Risk-Calibrated Language Models

Xinran Wang
Jin Du
Azal Khan
qi le
Enmao Diao
Jiawei Zhou
Jie Ding
Ali Anwar

Large language models can generate rare but catastrophic outputs, such as harmful conversations or insecure code. Existing Reinforcement Learning from Human Feedback (RLHF) typically maximizes average reward, leaving high-risk tail events insufficiently controlled. We introduce Quantile‑Guided Alignment (QA), a framework that allows users to specify desired improvements at any quantile—individually or across multiple reward dimensions—thus shifting the distribution of outputs with finer control toward safer, more desirable outcomes. The method extends standard RLHF via an augmented reward formulation that enforces quantile constraints. Experiments on conversation and code‐generation tasks show that quantile alignment significantly enhances quality at targeted tails while maintaining overall performance. The results position QA as a principled route to risk‑calibrated language models with tail‑focused alignment.

PDF Details

NeurIPS Conference 2025 Conference Paper

CineTechBench: A Benchmark for Cinematographic Technique Understanding and Generation

Xinran Wang
Songyu Xu
Shan Xiangxuan
Yuxuan Zhang
Muxi Diao
Xueyan Duan
Yanhua huang
Kongming Liang

Cinematography is a cornerstone of film production and appreciation, shaping mood, emotion, and narrative through visual elements such as camera movement, shot composition, and lighting. Despite recent progress in multimodal large language models (MLLMs) and video generation models, the capacity of current models to grasp and reproduce cinematographic techniques remains largely uncharted, hindered by the scarcity of expert-annotated data. To bridge this gap, we present CineTechBench, a pioneering benchmark founded on precise, manual annotation by seasoned cinematography experts across key cinematography dimensions. Our benchmark covers seven essential aspects—shot scale, shot angle, composition, camera movement, lighting, color, and focal length—and includes over 600 annotated movie images and 120 movie clips with clear cinematographic techniques. For the understanding task, we design question–answer pairs and annotated descriptions to assess MLLMs’ ability to interpret and explain cinematographic techniques. For the generation task, we assess advanced video generation models on their capacity to reconstruct cinema-quality camera movements given conditions such as textual prompts or keyframes. We conduct a large-scale evaluation on 15+ MLLMs and 5+ video generation models. Our results offer insights into the limitations of current models and future directions for cinematography understanding and generation in automatical film production and appreciation. The code and benchmark can be accessed at \url{https: //github. com/PRIS-CV/CineTechBench}.

PDF Details

AAAI Conference 2025 Conference Paper

Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network

Xinran Wang
Guang Yang
Tian Ye
Yun Liu

Deep learning based dehazing networks trained on paired synthetic data have shown impressive performance, but they struggle with significant degradation in generalization ability on real-world hazy scenes. In this paper, we propose Dehaze-RetinexGAN, a lightweight Retinex-based Generative Adversarial Network for real-world image Dehazing using unpaired data. Our Dehaze-RetinexGAN consists of two stages: self-supervised pre-training and weakly-supervised fine-tuning. During the pre-training, we reduce the image dehazing task to an illumination-reflectance decomposition task based on the duality correlation between Retinex and dehazing. Specifically, a decomposition network named DecomNet is constructed to obtain an illumination and a reflectance, simultaneously. Moreover, a self-supervised learning strategy is developed to construct the connection between the preliminary dehazed result and the input hazy image, which constrains the solution space of DecomNet and accelerates training, leading to a more realistic dehazed result. In the fine-tuning stage, we develop a dual DTCWT-based attention module and embed it into the U-Net architecture to further improve the quality of preliminary result in the frequency domain. In addition, the adversarial learning is employed to constrain the relevance between the clean image and the final dehazed result in a weakly supervised manner, which can promote more natural performance. Extensive experiments on several real-world datasets demonstrate that our proposed framework performs favorably over state-of-the-art dehazing methods in visual quality and quantitative evaluation.

PDF Details DOI

ICLR Conference 2025 Conference Paper

MAP: Multi-Human-Value Alignment Palette

Xinran Wang
Qi Le
Ammar Ahmed
Enmao Diao
Yi Zhou 0015
Nathalie Baracaldo
Jie Ding 0002
Ali Anwar 0001

Ensuring that generative AI systems align with human values is essential but challenging, especially when considering multiple human values and their potential trade-offs. Since human values can be personalized and dynamically change over time, the desirable levels of value alignment vary across different ethnic groups, industry sectors, and user cohorts. Within existing frameworks, it is hard to define human values and align AI systems accordingly across different directions simultaneously, such as harmlessness, helpfulness, and positiveness. To address this, we develop a novel, first-principle approach called Multi-Human-Value Alignment Palette (MAP), which navigates the alignment across multiple human values in a structured and reliable way. MAP formulates the alignment problem as an optimization task with user-defined constraints, which define human value targets. It can be efficiently solved via a primal-dual approach, which determines whether a user-defined alignment target is achievable and how to achieve it. We conduct a detailed theoretical analysis of MAP by quantifying the trade-offs between values, the sensitivity to constraints, the fundamental connection between multi-value alignment and sequential alignment, and proving that linear weighted rewards are sufficient for multi-value alignment. Extensive experiments demonstrate MAP's ability to align multiple values in a principled manner while delivering strong empirical performance across various tasks.

Details

YNIMG Journal 2025 Journal Article

Olfactory sensation emotion regulation: The implicit emotion regulation function of positive olfactory stimuli during emotional picture processing

Jiaotao Cai
Xinran Wang
Jiayi Zhou
Ye di
Ziruo Shen
Shuo An
Bingyang Long
Yicheng Wang

Previous research has shown that olfactory stimuli can induce emotional states, physiological and neural responses related to emotions. It remains unclear whether positive olfactory stimuli could down-regulate individuals' negative emotions or up-regulate individuals' positive emotions in an unconscious way. The present study investigated the effects of emotional olfactory stimuli on the behavioral and electrophysiological responses to emotional pictures. Forty participants were exposed to different types of odor conditions (neutral, pleasant, unpleasant) and evaluated the emotional pictures' valence and arousal while electroencephalography was recorded. Behavioral results showed that participants reported more positive emotions in response to positive pictures in two types of pleasant odor conditions (especially the citrus) than in unpleasant or neutral odor conditions. However, the pleasant odor (lavender) increased the positive pictures' valence scores but decreased the positive pictures' arousal scores. The ERPs results showed that the pleasant odors reduced the amplitudes of N1 and EPN components in response to negative pictures, indicating that pleasant odors might down-regulate negative emotions through decreasing attentional capture for negative visual stimuli during the early stage. The pleasant odor (lavender) attenuated LPP amplitudes for emotional pictures, suggesting that the positive olfactory stimuli might be helpful to down-regulate the emotional arousal by reducing attentional deployment to negative visual stimuli during the late stage of emotional visual stimuli processing. These findings provided novel behavioral and neurophysiological evidence that positive olfactory stimuli modulated visual emotional processing across multiple stages, and suggested that olfactory sensation could function as a rapid and relatively effortless emotion regulation modality.

Details DOI

ICLR Conference 2025 Conference Paper

Probe Pruning: Accelerating LLMs through Dynamic Pruning via Model-Probing

Qi Le
Enmao Diao
Ziyan Wang
Xinran Wang
Jie Ding 0002
Li Yang
Ali Anwar 0001

We introduce Probe Pruning (PP), a novel framework for online, dynamic, structured pruning of Large Language Models (LLMs) applied in a batch-wise manner. PP leverages the insight that not all samples and tokens contribute equally to the model's output, and probing a small portion of each batch effectively identifies crucial weights, enabling tailored dynamic pruning for different batches. It comprises three main stages: probing, history-informed pruning, and full inference. In the probing stage, PP selects a small yet crucial set of hidden states, based on residual importance, to run a few model layers ahead. During the history-informed pruning stage, PP strategically integrates the probing states with historical states. Subsequently, it structurally prunes weights based on the integrated states and the PP importance score, a metric developed specifically to assess the importance of each weight channel in maintaining performance. In the final stage, full inference is conducted on the remaining weights. A major advantage of PP is its compatibility with existing models, as it operates without requiring additional neural network modules or fine-tuning. Comprehensive evaluations of PP on LLaMA-2/3 and OPT models reveal that even minimal probing—using just 1.5% of FLOPs—can substantially enhance the efficiency of structured pruning of LLMs. For instance, when evaluated on LLaMA-2-7B with WikiText2, PP achieves a 2.56 times lower ratio of performance degradation per unit of latency reduction compared to the state-of-the-art method at a 40\% pruning ratio.

Details

YNIMG Journal 2025 Journal Article

Temporal dynamics of perceptual integrity and semantic congruency during color-word processing: An ERP and decoding study

Guanghui Zhang
Ying Xin
Liting Song
Xinran Wang
Lihong Chen
Weiqi He
Wenbo Luo

Visual word recognition involves both perceptual and semantic processes, yet how both factors interact during early and late stages of neural processing remains unclear. In this study, we employed event-related potentials (ERPs) and multivariate pattern analysis (MVPA; decoding) to investigate the temporal dynamics of color-word congruency (congruent vs. incongruent) and font completeness (complete vs. incomplete) in a modified Stroop experiment. 26 participants (13 males; aged 19-28 years old; M = 21.8, SD = 2.5) viewed Chinese color words presented in either matching or mismatching font color, with font forms being either intact or degraded. The ERP results revealed that N170 amplitudes were significantly influenced by font integrity and marginally by color congruency, with a notable interaction between the two factors. Additionally, N2 amplitudes showed a significant main effect of font integrity only. P3 amplitudes were modulated by both factors independently, without interaction, while LPP responses were significantly affected only by color congruency. MVPA results further demonstrated that font integrity could be decoded from around 150 to 600 ms, while color congruency could be decoded reliably from approximately 360 to 800 ms. Moreover, the decoding analysis did not reveal an interaction similar to that observed in the ERP results between congruent and incongruent conditions across different perceptual contexts. These findings support a two-stage processing model, in which early perceptual features are processed prior to semantic congruency integration. The combination of ERP and MVPA highlights the distinct temporal profiles underlying perceptual and semantic processing in visual word recognition.

Details DOI

ICRA Conference 2023 Conference Paper

Mechanical Intelligence for Prehensile In-Hand Manipulation of Spatial Trajectories

Qiujie Lu
Zhongxue Gan
Xinran Wang
Guochao Bai
Zhuang Zhang
Nicolás Rojas 0002

The application of mechanical and other physical properties to the development of robotic systems that can easily adapt to changing external situations is known as mechanical intelligence. Following this concept, many robot hand designs can produce self-adaptive and versatile grasps with simple underactuated fingers and open-loop control, while mechanical- intelligent strategies for dexterous manipulation are still limited. This paper proposes a mechanical-intelligent technique to facilitate dexterous manipulation, in particular prehensile inhand manipulation. The proposed strategy is based on the generation of complex spatial trajectories of the hand-object system, controlled in open loop with the minimum number of actuators and using simple low-level non-position modes. This approach is exemplified by the rigorous analysis and testing of a three-fingered two-actuator underactuated robot hand, called the helical hand, which is capable of generating helical prehensile in-hand manipulation of diversiform objects under error tolerance controlled by constant speed algorithm.

Details

ICLR Conference 2021 Conference Paper

Information Laundering for Model Privacy

Xinran Wang
Yu Xiang 0004
Jun Gao
Jie Ding 0002

In this work, we propose information laundering, a novel framework for enhancing model privacy. Unlike data privacy that concerns the protection of raw data information, model privacy aims to protect an already-learned model that is to be deployed for public use. The private model can be obtained from general learning methods, and its deployment means that it will return a deterministic or random response for a given input query. An information-laundered model consists of probabilistic components that deliberately maneuver the intended input and output for queries of the model, so the model's adversarial acquisition is less likely. Under the proposed framework, we develop an information-theoretic principle to quantify the fundamental tradeoffs between model utility and privacy leakage and derive the optimal design.

Details

NeurIPS Conference 2020 Conference Paper

Assisted Learning: A Framework for Multi-Organization Learning

Xun Xian
Xinran Wang
Jie Ding
Reza Ghanadan

In an increasing number of AI scenarios, collaborations among different organizations or agents (e. g. , human and robots, mobile units) are often essential to accomplish an organization-specific mission. However, to avoid leaking useful and possibly proprietary information, organizations typically enforce stringent security constraints on sharing modeling algorithms and data, which significantly limits collaborations. In this work, we introduce the Assisted Learning framework for organizations to assist each other in supervised learning tasks without revealing any organization's algorithm, data, or even task. An organization seeks assistance by broadcasting task-specific but nonsensitive statistics and incorporating others' feedback in one or more iterations to eventually improve its predictive performance. Theoretical and experimental studies, including real-world medical benchmarks, show that Assisted Learning can often achieve near-oracle learning performance as if data and training processes were centralized.

PDF Details