arXiv subjects

Ran Zhou

Publications and source records attributed to Ran Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Power from Potential: A Survey of Electrostatic Actuators for Haptics

As haptic interfaces integrate more seamlessly into wearables and everyday environments, they increasingly require actuators that are soft, thin, silent, and energy efficient. However, conventional motors and temperature-responsive polymers often struggle to deliver these properties due to their bulky form factors and high power consumption. High-Voltage Electrostatic Actuators (HVEAs), which generate force by applying an electric field to localized charge concentrations using high voltages and ultra-low currents, have recently emerged as a compelling alternative due to their fast, silent, and low-power operation within highly customizable and compliant form factors. This paper presents a focused review of HVEAs for haptics, examining four major classes: electrostatic switchable adhesives, dielectric elastomer actuators, soft electrohydraulic actuators, and electrokinetic pumps. For each class, we describe their mechanisms that enable haptic output; characterize their bandwidths, force densities, and spatial scalability; and evaluate their versatility for rendering cutaneous and kinesthetic feedback across wearable and world-grounded interfaces. Through this cross-technology analysis, we identify common design constraints and emerging strategies for improving ergonomics, streamlining fabrication, and integrating self-sensing. We conclude by outlining where HVEAs are uniquely positioned to advance haptic interaction and highlighting key research directions needed to translate these technologies into practical systems.

cs.RO↗

arXiv2026

Unified Semantic Modeling Framework for Large-Scale Job Understanding at LinkedIn

Job understanding is critical to LinkedIn's mission of connecting talent with opportunity. This task involves transforming unstructured and noisy job postings into standardized or derived job attributes that power numerous LinkedIn products. However, building a scalable, cost-efficient, and high-performing job understanding system remains challenging. In this paper, we present a unified semantic modeling framework powered by a small language model (SLM) to address the challenges. We begin by fine-tuning an open-source SLM using a suite of carefully curated synthetic tasks augmented with reasoning traces. These tasks jointly target taxonomy-guided classification and taxonomy-agnostic entity extraction. This allows the resulting model to acquire robust zero-shot generalization for job understanding in structured and unstructured contexts. Building upon this foundation, we introduce a multi-adapter architecture with attribute grouping to facilitate efficient task-specific adaptation while streamlining model management across diverse downstream attributes. Offline evaluations and online A/B tests demonstrate significant performance improvement while reducing operational complexity. Our work provides practical insights into building industry-scale text understanding systems.

cs.AI↗

arXiv2026

Masked Generative Transformer Is What You Need for Image Editing

Diffusion models dominate image editing, yet their global denoising mechanism entangles edited regions with surrounding context, causing modifications to propagate into areas that should remain intact. We propose a fundamentally different approach by leveraging Masked Generative Transformers (MGTs), whose localized token-prediction paradigm naturally confines changes to intended regions. We present EditMGT, an MGT-based editing framework that is the first of its kind. Our approach employs multi-layer attention consolidation to aggregate cross-attention maps into precise edit localization signals, and region-hold sampling to explicitly prevent token flipping in non-target areas. To support training, we construct CrispEdit-2M, a 2M-sample high-resolution (>1024) editing dataset spanning seven categories. With only 960M parameters, EditMGT achieves state-of-the-art image similarity on multiple benchmarks while delivering 6x faster editing, demonstrating that MGTs offer a compelling alternative to diffusion-based editing.

cs.CV↗

arXiv2026

Otherness as a Quality in Designing Expressive Robotic Touch

Haptic technologies have advanced rapidly, yet exploration of robotic touch remains dominated by replicating realistic environmental cues or hand gestures, which narrows the design space and risks social resistance. This paper argues for alternatives: grounded in the notion of "otherness" from human-robot interaction (HRI), we propose treating robotic touch's inherent otherness as a design quality. Instead of being a limitation in pursuing realism, otherness can be embraced to elicit ambiguity and provoke alternative interpretations, fostering expressive and evocative robotic touch design. To develop this perspective, we analyze inspirational art and design precedents and four design research cases through a reflective Research through Design (RtD) approach. Through this analysis, we articulate a set of design languages structured around why otherness matters for touch meaning-making, how it can be shaped through design strategies, and where it can be embedded within robotic touch systems. We conclude by reflecting on the tensions and risks involved in designing robotic touch with otherness in mind.

cs.HC↗

arXiv2026

Grounded Token Initialization for New Vocabulary in LMs for Generative Recommendation

Language models (LMs) are increasingly extended with new learnable vocabulary tokens for domain-specific tasks, such as Semantic-ID tokens in generative recommendation. The standard practice initializes these new tokens as the mean of existing vocabulary embeddings, then relies on supervised fine-tuning to learn their representations. We present a systematic analysis of this strategy: through spectral and geometric diagnostics, we show that mean initialization collapses all new tokens into a degenerate subspace, erasing inter-token distinctions that subsequent fine-tuning struggles to fully recover. These findings suggest that \emph{token initialization} is a key bottleneck when extending LMs with new vocabularies. Motivated by this diagnosis, we propose the \emph{Grounded Token Initialization Hypothesis}: linguistically grounding novel tokens in the pretrained embedding space before fine-tuning better enables the model to leverage its general-purpose knowledge for novel-token domains. We operationalize this hypothesis as GTI (Grounded Token Initialization), a lightweight grounding stage that, prior to fine-tuning, maps new tokens to distinct, semantically meaningful locations in the pretrained embedding space using only paired linguistic supervision. Despite its simplicity, GTI outperforms both mean initialization and existing auxiliary-task adaptation methods in the majority of evaluation settings across multiple generative recommendation benchmarks, including industry-scale and public datasets. Further analyses show that grounded embeddings produce richer inter-token structure that persists through fine-tuning, corroborating the hypothesis that initialization quality is a key bottleneck in vocabulary extension.

cs.CL↗

arXiv2025

EditMGT: Unleashing Potentials of Masked Generative Transformers in Image Editing

Recent advances in diffusion models (DMs) have achieved exceptional visual quality in image editing tasks. However, the global denoising dynamics of DMs inherently conflate local editing targets with the full-image context, leading to unintended modifications in non-target regions. In this paper, we shift our attention beyond DMs and turn to Masked Generative Transformers (MGTs) as an alternative approach to tackle this challenge. By predicting multiple masked tokens rather than holistic refinement, MGTs exhibit a localized decoding paradigm that endows them with the inherent capacity to explicitly preserve non-relevant regions during the editing process. Building upon this insight, we introduce the first MGT-based image editing framework, termed EditMGT. We first demonstrate that MGT's cross-attention maps provide informative localization signals for localizing edit-relevant regions and devise a multi-layer attention consolidation scheme that refines these maps to achieve fine-grained and precise localization. On top of these adaptive localization results, we introduce region-hold sampling, which restricts token flipping within low-attention areas to suppress spurious edits, thereby confining modifications to the intended target regions and preserving the integrity of surrounding non-target areas. To train EditMGT, we construct CrispEdit-2M, a high-resolution dataset spanning seven diverse editing categories. Without introducing additional parameters, we adapt a pre-trained text-to-image MGT into an image editing model through attention injection. Extensive experiments across four standard benchmarks demonstrate that, with fewer than 1B parameters, our model achieves similarity performance while enabling 6 times faster editing. Moreover, it delivers comparable or superior editing quality, with improvements of 3.6% and 17.6% on style change and style transfer tasks, respectively.

cs.CV↗

arXiv2025

Theoretical investigations of the origin of persistent luminescence in spinel oxides MgGa2O4 and MgAl2O4

MgGa2O4 and MgAl2O4 have attracted significant interest due to their unique intrinsic persistent luminescence, offering promising potential for various applications. In this paper, from the perspective of defect physics, we systemically investigate the origin of persistent luminescence phenomena in pristine MgGa2O4 and MgAl2O4, employing accurate hybrid functional calculations. Our results show that vacancies and antisite defects involving the two cations are the dominant point defects in both materials. Our calculated optical excitation and emission peaks associated with the MgGa defect agree well with the experimentally observed blue luminescence peak at about 2.9 eV in MgGa2O4. In MgAl2O4, the intradefect optical transition within the VO-MgAl donor-acceptor defect complex is identified as a likely origin for the observed 2.7 eV emission peak. Furthermore, the calculated radiative recombination coefficients of MgGa and VO-MgAl are significantly higher than their nonradiative counterparts, supporting their roles as efficient luminescent centers. Our results regarding the optical processes of oxygen vacancy VO in MgGa2O4 and MgAl2O4 are also in good agreement with experimental results. Based on the calculated defect thermodynamic transition levels, the intrinsic persistent luminescence in MgGa2O4 and MgAl2O4 may be attributed to electron traps, GaMg and VO, in the former and a hole trap, MgAl, in the latter. Donor-acceptor defect complexes (VO+VMg and VO+VGa) are also found to serve as effective carrier trapping centers in MgGa2O4. The calculated trap depths are also consistent with thermoluminescence spectroscopy measurements.

cond-mat.mtrl-sci↗

arXiv2025

MERIT: Multilingual Semantic Retrieval with Interleaved Multi-Condition Query

Semantic retrieval is crucial for modern applications yet remains underexplored in current research. Existing datasets are limited to single languages, single images, or singular retrieval conditions, often failing to fully exploit the expressive capacity of visual information as evidenced by maintained performance when images are replaced with captions. However, practical retrieval scenarios frequently involve interleaved multi-condition queries with multiple images. Hence, this paper introduces MERIT, the first multilingual dataset for interleaved multi-condition semantic retrieval, comprising 320,000 queries with 135,000 products in 5 languages, covering 7 distinct product categories. Extensive experiments on MERIT identify existing models's limitation: focusing solely on global semantic information while neglecting specific conditional elements in queries. Consequently, we propose Coral, a novel fine-tuning framework that adapts pre-trained MLLMs by integrating embedding reconstruction to preserve fine-grained conditional elements and contrastive learning to extract comprehensive global semantics. Experiments demonstrate that Coral achieves a 45.9% performance improvement over conventional approaches on MERIT, with strong generalization capabilities validated across 8 established retrieval benchmarks. Collectively, our contributions - a novel dataset, identification of critical limitations in existing approaches, and an innovative fine-tuning framework - establish a foundation for future research in interleaved multi-condition semantic retrieval.

cs.CV↗

arXiv2025

Shape-Kit: A Design Toolkit for Crafting On-Body Expressive Haptics

Driven by the vision of everyday haptics, the HCI community is advocating for "design touch first" and investigating "how to touch well." However, a gap remains between the exploratory nature of haptic design and technical reproducibility. We present Shape-Kit, a hybrid design toolkit embodying our "crafting haptics" metaphor, where hand touch is transduced into dynamic pin-based sensations that can be freely explored across the body. An ad-hoc tracking module captures and digitizes these patterns. Our study with 14 designers and artists demonstrates how Shape-Kit facilitates sensorial exploration for expressive haptic design. We analyze how designers collaboratively ideate, prototype, iterate, and compose touch experiences and show the subtlety and richness of touch that can be achieved through diverse crafting methods with Shape-Kit. Reflecting on the findings, our work contributes key insights into haptic toolkit design and touch design practices centered on the "crafting haptics" metaphor. We discuss in-depth how Shape-Kit's simplicity, though remaining constrained, enables focused crafting for deeper exploration, while its collaborative nature fosters shared sense-making of touch experiences.

cs.HC↗

arXiv2024

QCRD: Quality-guided Contrastive Rationale Distillation for Large Language Models

The deployment of large language models (LLMs) faces considerable challenges concerning resource constraints and inference efficiency. Recent research has increasingly focused on smaller, task-specific models enhanced by distilling knowledge from LLMs. However, prior studies have often overlooked the diversity and quality of knowledge, especially the untapped potential of negative knowledge. Constructing effective negative knowledge remains severely understudied. In this paper, we introduce a novel framework called quality-guided contrastive rationale distillation aimed at enhancing reasoning capabilities through contrastive knowledge learning. For positive knowledge, we enrich its diversity through temperature sampling and employ self-consistency for further denoising and refinement. For negative knowledge, we propose an innovative self-adversarial approach that generates low-quality rationales by sampling previous iterations of smaller language models, embracing the idea that one can learn from one's own weaknesses. A contrastive loss is developed to distill both positive and negative knowledge into smaller language models, where an online-updating discriminator is integrated to assess qualities of rationales and assign them appropriate weights, optimizing the training process. Through extensive experiments across multiple reasoning tasks, we demonstrate that our method consistently outperforms existing distillation techniques, yielding higher-quality rationales.

cs.CL↗

arXiv2024

Ultra-broadband Optical Switching Plasmons Waveguide in Ge Nanowires

Plasmonic devices, with their ultra-high integration density and data-carrying capacity comparable to optical devices, are currently a hot topic in the field of nanophotonic devices. Photodetectors, non-volatile memories, and ultra-compact lasers based on plasmons in low-dimensional materials are emerging at a rapid pace. However, the narrow optical response band and limited of convenient tunable methods currently available have hindered the development of these plasmonic materials. Here, we report a ultrabroadband non-equilibrium plasmonic responses based on Ge nanowires tuned by optical method. We tracked the blue shift of the plasmonic response of Ge nanowires due to photo-induced carriers over an ultra-broad spectral range of 800-2000 $cm^{-1}$. For the first time, we have achieved the imaging of propagating surface plasmon polaritons (SPPs) in semiconductor nanowires, which were tuned by photo-induced carriers. The ultrafast and ultrabroadband response of semiconductor nanowire plasmons is of great significance for future ultrafast all-optical devices.

physics.optics↗

arXiv2024

WAL-Net: Weakly supervised auxiliary task learning network for carotid plaques classification

The classification of carotid artery ultrasound images is a crucial means for diagnosing carotid plaques, holding significant clinical relevance for predicting the risk of stroke. Recent research suggests that utilizing plaque segmentation as an auxiliary task for classification can enhance performance by leveraging the correlation between segmentation and classification tasks. However, this approach relies on obtaining a substantial amount of challenging-to-acquire segmentation annotations. This paper proposes a novel weakly supervised auxiliary task learning network model (WAL-Net) to explore the interdependence between carotid plaque classification and segmentation tasks. The plaque classification task is primary task, while the plaque segmentation task serves as an auxiliary task, providing valuable information to enhance the performance of the primary task. Weakly supervised learning is adopted in the auxiliary task to completely break away from the dependence on segmentation annotations. Experiments and evaluations are conducted on a dataset comprising 1270 carotid plaque ultrasound images from Wuhan University Zhongnan Hospital. Results indicate that the proposed method achieved an approximately 1.3% improvement in carotid plaque classification accuracy compared to the baseline network. Specifically, the accuracy of mixed-echoic plaques classification increased by approximately 3.3%, demonstrating the effectiveness of our approach.

eess.IV↗

arXiv2024

GroundingGPT:Language Enhanced Multi-modal Grounding Model

Multi-modal large language models have demonstrated impressive performance across various tasks in different modalities. However, existing multi-modal models primarily emphasize capturing global information within each modality while neglecting the importance of perceiving local information across modalities. Consequently, these models lack the ability to effectively understand the fine-grained details of input data, limiting their performance in tasks that require a more nuanced understanding. To address this limitation, there is a compelling need to develop models that enable fine-grained understanding across multiple modalities, thereby enhancing their applicability to a wide range of tasks. In this paper, we propose GroundingGPT, a language enhanced multi-modal grounding model. Beyond capturing global information like other multi-modal models, our proposed model excels at tasks demanding a detailed understanding of local information within the input. It demonstrates precise identification and localization of specific regions in images or moments in videos. To achieve this objective, we design a diversified dataset construction pipeline, resulting in a multi-modal, multi-granularity dataset for model training. The code, dataset, and demo of our model can be found at https: //github.com/lzw-lzw/GroundingGPT.

cs.CV↗

arXiv2023

A region and category confidence-based multi-task network for carotid ultrasound image segmentation and classification

The segmentation and classification of carotid plaques in ultrasound images play important roles in the treatment of atherosclerosis and assessment for the risk of stroke. Although deep learning methods have been used for carotid plaque segmentation and classification, two-stage methods will increase the complexity of the overall analysis and the existing multi-task methods ignored the relationship between the segmentation and classification. These will lead to suboptimal performance as valuable information might not be fully leveraged across all tasks. Therefore, we propose a multi-task learning framework (RCCM-Net) for ultrasound carotid plaque segmentation and classification, which utilizes a region confidence module (RCM) and a sample category confidence module (CCM) to exploit the correlation between these two tasks. The RCM provides knowledge from the probability of plaque regions to the classification task, while the CCM is designed to learn the categorical sample weight for the segmentation task. A total of 1270 2D ultrasound images of carotid plaques were collected from Zhongnan Hospital (Wuhan, China) for our experiments. The results showed that the proposed method can improve both segmentation and classification performance compared to existing single-task networks (i.e., SegNet, Deeplabv3+, UNet++, EfficientNet, Res2Net, RepVGG, DPN) and multi-task algorithms (i.e., HRNet, MTANet), with an accuracy of 85.82% for classification and a Dice-similarity-coefficient of 84.92% for segmentation. In the ablation study, the results demonstrated that both the designed RCM and CCM were beneficial in improving the network's performance. Therefore, we believe that the proposed method could be useful for carotid plaque analysis in clinical trials and practice.

eess.IV↗

arXiv2023

Improving Self-training for Cross-lingual Named Entity Recognition with Contrastive and Prototype Learning

In cross-lingual named entity recognition (NER), self-training is commonly used to bridge the linguistic gap by training on pseudo-labeled target-language data. However, due to sub-optimal performance on target languages, the pseudo labels are often noisy and limit the overall performance. In this work, we aim to improve self-training for cross-lingual NER by combining representation learning and pseudo label refinement in one coherent framework. Our proposed method, namely ContProto mainly comprises two components: (1) contrastive self-training and (2) prototype-based pseudo-labeling. Our contrastive self-training facilitates span classification by separating clusters of different classes, and enhances cross-lingual transferability by producing closely-aligned representations between the source and target language. Meanwhile, prototype-based pseudo-labeling effectively improves the accuracy of pseudo labels during training. We evaluate ContProto on multiple transfer pairs, and experimental results show our method brings in substantial improvements over current state-of-the-art methods.

cs.CL↗

arXiv2022

ConNER: Consistency Training for Cross-lingual Named Entity Recognition

Cross-lingual named entity recognition (NER) suffers from data scarcity in the target languages, especially under zero-shot settings. Existing translate-train or knowledge distillation methods attempt to bridge the language gap, but often introduce a high level of noise. To solve this problem, consistency training methods regularize the model to be robust towards perturbations on data or hidden states. However, such methods are likely to violate the consistency hypothesis, or mainly focus on coarse-grain consistency. We propose ConNER as a novel consistency training framework for cross-lingual NER, which comprises of: (1) translation-based consistency training on unlabeled target-language data, and (2) dropoutbased consistency training on labeled source-language data. ConNER effectively leverages unlabeled target-language data and alleviates overfitting on the source language to enhance the cross-lingual adaptability. Experimental results show our ConNER achieves consistent improvement over various baseline methods.

cs.CL↗

arXiv2022

Cultivating Visualization Literacy for Children Through Curiosity and Play

Fostering data visualization literacy (DVL) as part of childhood education could lead to a more data literate society. However, most work in DVL for children relies on a more formal educational context (i.e., a teacher-led approach) that limits children's engagement with data to classroom-based environments and, consequently, children's ability to ask questions about and explore data on topics they find personally meaningful. We explore how a curiosity-driven, child-led approach can provide more agency to children when they are authoring data visualizations. This paper explores how informal learning with crafting physicalizations through play and curiosity may foster increased literacy and engagement with data. Employing a constructionist approach, we designed a do-it-yourself toolkit made out of everyday materials (e.g., paper, cardboard, mirrors) that enables children to create, customize, and personalize three different interactive visualizations (bar, line, pie). We used the toolkit as a design probe in a series of in-person workshops with 5 children (6 to 11-year-olds) and interviews with 5 educators. Our observations reveal that the toolkit helped children creatively engage and interact with visualizations. Children with prior knowledge of data visualization reported the toolkit serving as more of an authoring tool that they envision using in their daily lives, while children with little to no experience found the toolkit as an engaging introduction to data visualization. Our study demonstrates the potential of using the constructionist approach to cultivate children's DVL through curiosity and play.

cs.HC↗

arXiv2022

MReD: A Meta-Review Dataset for Structure-Controllable Text Generation

When directly using existing text generation datasets for controllable generation, we are facing the problem of not having the domain knowledge and thus the aspects that could be controlled are limited. A typical example is when using CNN/Daily Mail dataset for controllable text summarization, there is no guided information on the emphasis of summary sentences. A more useful text generator should leverage both the input text and the control signal to guide the generation, which can only be built with a deep understanding of the domain knowledge. Motivated by this vision, our paper introduces a new text generation dataset, named MReD. Our new dataset consists of 7,089 meta-reviews and all its 45k meta-review sentences are manually annotated with one of the 9 carefully defined categories, including abstract, strength, decision, etc. We present experimental results on start-of-the-art summarization models, and propose methods for structure-controlled generation with both extractive and abstractive models using our annotated data. By exploring various settings and analyzing the model behavior with respect to the control signal, we demonstrate the challenges of our proposed task and the values of our dataset MReD. Meanwhile, MReD also allows us to have a better understanding of the meta-review domain.

cs.CL↗