arXiv subjects

Yubo Wang

Publications and source records attributed to Yubo Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

MixLM: High-Throughput and Effective LLM Ranking via Text-Embedding Mix-Interaction

Large language models (LLMs) excel at capturing semantic nuances and therefore show impressive relevance ranking performance in modern recommendation and search systems. However, they suffer from high computational overhead under industrial latency and throughput requirements. In particular, cross-encoder ranking systems often create long context prefill-heavy workloads, as the model has to be presented with the user, query and item information. To this end, we propose MixLM, a novel LLM-based ranking framework, which significantly improves the system throughput via reducing the input context length, while preserving the semantic strength of cross-encoder rankers. In contrast to a standard ranking system where the context is presented to the model as pure text, we propose to use mix-interaction, a mixture of text and embedding tokens to represent the input. Specifically, MixLM encodes all items in the catalog into a few embedding tokens and stores in a nearline cache. The encoded item descriptions are used during online inference, effectively reducing the item length from a few thousand text tokens to a few embedding tokens. We share insights from deploying our MixLM framework to a real-world search application at LinkedIn, including a detailed discussion of our training pipelines, as well as a thorough analysis of our online serving infrastructure optimization. With the same latency budget and on-par relevance metrics, MixLM increased throughput by 10.0x comparing with strong baselines, 75.9x over full-text LLM rerankers. The efficiency gains delivered by MixLM enabled full-traffic deployment of LLM-powered search, which resulted in a significant 0.47\% increase in Daily Active Users (DAU) in online A/B tests.

cs.IR↗

arXiv2026

GORAG: Graph-based Online Retrieval Augmented Generation for Dynamic Few-shot Social Media Text Classification

Text classification is vital for Web for Good applications like hate speech and misinformation detection. However, traditional models (e.g., BERT) often fail in dynamic few-shot settings where labeled data are scarce, and target labels frequently evolve. While Large Language Models (LLMs) show promise in few-shot settings, their performance is often hindered by increased input size in dynamic evolving scenarios. To address these issues, we propose GORAG, a Graph-based Online Retrieval-Augmented Generation framework for dynamic few-shot text classification. GORAG constructs and maintains a weighted graph of keywords and text labels, representing their correlations as edges. To model these correlations, GORAG employs an edge weighting mechanism to prioritize the importance and reliability of extracted information and dynamically retrieves relevant context using a tailored minimum-cost spanning tree for each input. Empirical evaluations show GORAG outperforms existing approaches by providing more comprehensive and precise contextual information. Our code is released at: https://github.com/Wyb0627/GORAG.

cs.CL↗

arXiv2026

Co-Channel Interference Mitigation Using Deep Learning for Drone-Based Large-Scale Antenna Measurements

Unmanned aerial vehicles (UAVs) enable efficient in-situ radiation characterization of large-aperture antennas directly in their deployment environments. In such measurements, a continuous-wave (CW) probe tone is commonly transmitted to characterize the antenna response. However, active co-channel emissions from neighboring antennas often introduce severe in-band interference, where classical FFT-based estimators fail to accurately estimate the CW tone amplitude when the signal-to-interference ratios (SIR) falls below -10 dB. This paper proposes a lightweight deep convolutional neural network (DC-CNN) that estimates the amplitude of the CW tone. The model is trained and evaluated on real 5~GHz measurement bursts spanning an effective SIR range of --33.3 dB to +46.7 dB. Despite its compact size (<20k parameters), the proposed DC-CNN achieves a mean absolute error (MAE) of 7% over the full range, with <1 dB error for SIR >= -30 dB. This robustness and efficiency make DC-CNN suitable for deployment on embedded UAV platforms for interference-resilient antenna pattern characterization.

eess.SP↗

arXiv2025

Holistic Evaluation of Multimodal LLMs on Spatial Intelligence

Multimodal models have achieved remarkable progress in recent years. Nevertheless, they continue to exhibit notable limitations in spatial understanding and reasoning, the very capability that anchors artificial general intelligence in the physical world. With the recent release of GPT-5, allegedly the most powerful AI model to date, it is timely to examine where the leading models (GPT, Gemini, Grok, Seed, Qwen, and Intern) stand on the path toward spatial intelligence (SI). We thus propose EASI for holistic Evaluation of multimodAl LLMs on Spatial Intelligence. EASI conceptualizes a comprehensive taxonomy of spatial tasks that unifies existing benchmarks and a growing collection of newly curated ones, enabling systematic evaluation of state-of-the-art models. In this report, we conduct the study across eight key benchmarks, at a cost exceeding ten billion total tokens. Our empirical study then reveals that (1) GPT-5 demonstrates unprecedented strength in SI, yet (2) still falls short of human performance significantly across a broad spectrum of SI-tasks. Moreover, we (3) show that SI-tasks expose greater model capability deficiency than non-SI tasks, to the extent that (4) proprietary models do not exhibit a decisive advantage when facing the most difficult ones. In addition, we conduct a qualitative evaluation across a diverse set of scenarios that are intuitive for humans, yet fail the most advanced multimodal models. EASI is an ongoing community effort: we have open-sourced the EASI codebase that provides a one-stop and reproducible solution with standardized interfaces, integrated protocols and prompts that significantly reduce the friction of configuring and running multiple benchmarks; we have also launched an accompanying EASI leaderboard to provide a continually updated snapshot of model performance across the full SI spectrum, accelerating collective progress toward robust SI.

cs.CV↗

arXiv2025

Survey and Experiments on Mental Disorder Detection via Social Media: From Large Language Models and RAG to Agents

Mental disorders represent a critical global health challenge, and social media is increasingly viewed as a vital resource for real-time digital phenotyping and intervention. To leverage this data, large language models (LLMs) have been introduced, offering stronger semantic understanding and reasoning than traditional deep learning, thereby enhancing the explainability of detection results. Despite the growing prominence of LLMs in this field, there is a scarcity of scholarly works that systematically synthesize how advanced enhancement techniques, specifically Retrieval-Augmented Generation (RAG) and Agentic systems, can be utilized to address these reliability and reasoning limitations. Here, we systematically survey the evolving landscape of LLM-based methods for social media mental disorder analysis, spanning standard pre-trained language models, RAG to mitigate hallucinations and contextual gaps, and agentic systems for autonomous reasoning and multi-step intervention. We organize existing work by technical paradigm and clinical target, extending beyond common internalizing disorders to include psychotic disorders and externalizing behaviors. Additionally, the paper comprehensively evaluates the performance of LLMs, including the impact of RAG, across various tasks. This work establishes a unified benchmark for the field, paving the way for the development of trustworthy, autonomous AI systems that can deliver precise and explainable mental health support.

cs.CL↗

arXiv2025

WaveTuner: Comprehensive Wavelet Subband Tuning for Time Series Forecasting

Due to the inherent complexity, temporal patterns in real-world time series often evolve across multiple intertwined scales, including long-term periodicity, short-term fluctuations, and abrupt regime shifts. While existing literature has designed many sophisticated decomposition approaches based on the time or frequency domain to partition trend-seasonality components and high-low frequency components, an alternative line of approaches based on the wavelet domain has been proposed to provide a unified multi-resolution representation with precise time-frequency localization. However, most wavelet-based methods suffer from a persistent bias toward recursively decomposing only low-frequency components, severely underutilizing subtle yet informative high-frequency components that are pivotal for precise time series forecasting. To address this problem, we propose WaveTuner, a Wavelet decomposition framework empowered by full-spectrum subband Tuning for time series forecasting. Concretely, WaveTuner comprises two key modules: (i) Adaptive Wavelet Refinement module, that transforms time series into time-frequency coefficients, utilizes an adaptive router to dynamically assign subband weights, and generates subband-specific embeddings to support refinement; and (ii) Multi-Branch Specialization module, that employs multiple functional branches, each instantiated as a flexible Kolmogorov-Arnold Network (KAN) with a distinct functional order to model a specific spectral subband. Equipped with these modules, WaveTuner comprehensively tunes global trends and local variations within a unified time-frequency framework. Extensive experiments on eight real-world datasets demonstrate WaveTuner achieves state-of-the-art forecasting performance in time series forecasting.

cs.LG↗

arXiv2025

EPARA: Parallelizing Categorized AI Inference in Edge Clouds

With the increasing adoption of AI applications such as large language models and computer vision AI, the computational demands on AI inference systems are continuously rising, making the enhancement of task processing capacity using existing hardware a primary objective in edge clouds. We propose EPARA, an end-to-end AI parallel inference framework in edge, aimed at enhancing the edge AI serving capability. Our key idea is to categorize tasks based on their sensitivity to latency/frequency and requirement for GPU resources, thereby achieving both request-level and service-level task-resource allocation. EPARA consists of three core components: 1) a task-categorized parallelism allocator that decides the parallel mode of each task, 2) a distributed request handler that performs the calculation for the specific request, and 3) a state-aware scheduler that periodically updates service placement in edge clouds. We implement a EPARA prototype and conduct a case study on the EPARA operation for LLMs and segmentation tasks. Evaluation through testbed experiments involving edge servers, embedded devices, and microcomputers shows that EPARA achieves up to 2.1$\times$ higher goodput in production workloads compared to prior frameworks, while adapting to various edge AI inference tasks.

cs.DC↗

arXiv2025

Shifting AI Efficiency From Model-Centric to Data-Centric Compression

The advancement of large language models (LLMs) and multi-modal LLMs (MLLMs) has historically relied on scaling model parameters. However, as hardware limits constrain further model growth, the primary computational bottleneck has shifted to the quadratic cost of self-attention over increasingly long sequences by ultra-long text contexts, high-resolution images, and extended videos. In this position paper, \textbf{we argue that the focus of research for efficient artificial intelligence (AI) is shifting from model-centric compression to data-centric compression}. We position data-centric compression as the emerging paradigm, which improves AI efficiency by directly compressing the volume of data processed during model training or inference. To formalize this shift, we establish a unified framework for existing efficiency strategies and demonstrate why it constitutes a crucial paradigm change for long-context AI. We then systematically review the landscape of data-centric compression methods, analyzing their benefits across diverse scenarios. Finally, we outline key challenges and promising future research directions. Our work aims to provide a novel perspective on AI efficiency, synthesize existing efforts, and catalyze innovation to address the challenges posed by ever-increasing context lengths.

cs.CL↗

arXiv2025

AudioMarathon: A Comprehensive Benchmark for Long-Context Audio Understanding and Efficiency in Audio LLMs

Processing long-form audio is a major challenge for Large Audio Language models (LALMs). These models struggle with the quadratic cost of attention ($O(N^2)$) and with modeling long-range temporal dependencies. Existing audio benchmarks are built mostly from short clips and do not evaluate models in realistic long context settings. To address this gap, we introduce AudioMarathon, a benchmark designed to evaluate both understanding and inference efficiency on long-form audio. AudioMarathon provides a diverse set of tasks built upon three pillars: long-context audio inputs with durations ranging from 90.0 to 300.0 seconds, which correspond to encoded sequences of 2,250 to 7,500 audio tokens, respectively, full domain coverage across speech, sound, and music, and complex reasoning that requires multi-hop inference. We evaluate state-of-the-art LALMs and observe clear performance drops as audio length grows. We also study acceleration techniques and analyze the trade-offs of token pruning and KV cache eviction. The results show large gaps across current LALMs and highlight the need for better temporal reasoning and memory-efficient architectures. We believe AudioMarathon will drive the audio and multimodal research community to develop more advanced audio understanding models capable of solving complex audio tasks.

cs.SD↗

arXiv2025

Coercive Field Reduction in Ultra-thin Al1-XScXN via Interfacial Engineering with a Scandium Electrode

Aluminum scandium nitride (AlScN) ferroelectrics are promising for next-generation non-volatile memory applications due to high remnant polarization compared with Pb(ZrxTi1-x)O3 and doped-HfO2 material systems, as well as their fast switching and scalability to nanometer thicknesses. As AlScN films are thinned to 10 nm thickness, coercive field has been shown to substantially increase, which hinders low voltage operation. We demonstrate that interfacial engineering through bottom electrode selection and strain management reduces this coercive field increase with scaling and improves ferroelectric performance. Here, we demonstrate robust ferroelectricity in ultra-thin AlScN capacitors deposited on a Sc bottom electrode under both alternating current and direct current conditions. The coercive field is reduced by over 20 percent compared to capacitors with an Al bottom electrode. Furthermore, the difference in dynamic switching behavior over a decade of frequency was evaluated by applying the KAI model. At frequencies lower than 16.7 kHz, the capacitors with Sc and Al bottom electrodes exhibit comparable KAI exponents of 0.030 and 0.028, indicating similar switching kinetics. However, at higher frequencies, the capacitor with an Al bottom electrode shows a significantly higher exponent of 0.063, indicating a stronger frequency dependence, whereas the capacitor with a Sc bottom electrode maintains a stable exponent of 0.030, suggesting a lower frequency dependence during faster switching scenarios. The Scanning Electron Nanobeam Diffraction technique was selected to measure the strain difference in AlScN thin films grown on templates with different lattice mismatch, providing a correlation between lattice mismatch, film strain and switching behavior in ultra-thin film systems.

cond-mat.mtrl-sci↗

arXiv2025

Privacy Risks of LLM-Empowered Recommender Systems: An Inversion Attack Perspective

The large language model (LLM) powered recommendation paradigm has been proposed to address the limitations of traditional recommender systems, which often struggle to handle cold start users or items with new IDs. Despite its effectiveness, this study uncovers that LLM empowered recommender systems are vulnerable to reconstruction attacks that can expose both system and user privacy. To examine this threat, we present the first systematic study on inversion attacks targeting LLM empowered recommender systems, where adversaries attempt to reconstruct original prompts that contain personal preferences, interaction histories, and demographic attributes by exploiting the output logits of recommendation models. We reproduce the vec2text framework and optimize it using our proposed method called Similarity Guided Refinement, enabling more accurate reconstruction of textual prompts from model generated logits. Extensive experiments across two domains (movies and books) and two representative LLM based recommendation models demonstrate that our method achieves high fidelity reconstructions. Specifically, we can recover nearly 65 percent of the user interacted items and correctly infer age and gender in 87 percent of the cases. The experiments also reveal that privacy leakage is largely insensitive to the victim model's performance but highly dependent on domain consistency and prompt complexity. These findings expose critical privacy vulnerabilities in LLM empowered recommender systems.

cs.IR↗

arXiv2025

BASIC: Boosting Visual Alignment with Intrinsic Refined Embeddings in Multimodal Large Language Models

Mainstream Multimodal Large Language Models (MLLMs) achieve visual understanding by using a vision projector to bridge well-pretrained vision encoders and large language models (LLMs). The inherent gap between visual and textual modalities makes the embeddings from the vision projector critical for visual comprehension. However, current alignment approaches treat visual embeddings as contextual cues and merely apply auto-regressive supervision to textual outputs, neglecting the necessity of introducing equivalent direct visual supervision, which hinders the potential finer alignment of visual embeddings. In this paper, based on our analysis of the refinement process of visual embeddings in the LLM's shallow layers, we propose BASIC, a method that utilizes refined visual embeddings within the LLM as supervision to directly guide the projector in generating initial visual embeddings. Specifically, the guidance is conducted from two perspectives: (i) optimizing embedding directions by reducing angles between initial and supervisory embeddings in semantic space; (ii) improving semantic matching by minimizing disparities between the logit distributions of both visual embeddings. Without additional supervisory models or artificial annotations, BASIC significantly improves the performance of MLLMs across a wide range of benchmarks, demonstrating the effectiveness of our introduced direct visual supervision.

cs.CV↗

arXiv2025

CROP: Integrating Topological and Spatial Structures via Cross-View Prefixes for Molecular LLMs

Recent advances in molecular science have been propelled significantly by large language models (LLMs). However, their effectiveness is limited when relying solely on molecular sequences, which fail to capture the complex structures of molecules. Beyond sequence representation, molecules exhibit two complementary structural views: the first focuses on the topological relationships between atoms, as exemplified by the graph view; and the second emphasizes the spatial configuration of molecules, as represented by the image view. The two types of views provide unique insights into molecular structures. To leverage these views collaboratively, we propose the CROss-view Prefixes (CROP) to enhance LLMs' molecular understanding through efficient multi-view integration. CROP possesses two advantages: (i) efficiency: by jointly resampling multiple structural views into fixed-length prefixes, it avoids excessive consumption of the LLM's limited context length and allows easy expansion to more views; (ii) effectiveness: by utilizing the LLM's self-encoded molecular sequences to guide the resampling process, it boosts the quality of the generated prefixes. Specifically, our framework features a carefully designed SMILES Guided Resampler for view resampling, and a Structural Embedding Gate for converting the resulting embeddings into LLM's prefixes. Extensive experiments demonstrate the superiority of CROP in tasks including molecule captioning, IUPAC name prediction and molecule property prediction.

q-bio.QM↗

arXiv2025

Understanding the Embedding Models on Hyper-relational Knowledge Graph

Recently, Hyper-relational Knowledge Graphs (HKGs) have been proposed as an extension of traditional Knowledge Graphs (KGs) to better represent real-world facts with additional qualifiers. As a result, researchers have attempted to adapt classical Knowledge Graph Embedding (KGE) models for HKGs by designing extra qualifier processing modules. However, it remains unclear whether the superior performance of Hyper-relational KGE (HKGE) models arises from their base KGE model or the specially designed extension module. Hence, in this paper, we data-wise convert HKGs to KG format using three decomposition methods and then evaluate the performance of several classical KGE models on HKGs. Our results show that some KGE models achieve performance comparable to that of HKGE models. Upon further analysis, we find that the decomposition methods alter the original HKG topology and fail to fully preserve HKG information. Moreover, we observe that current HKGE models are either insufficient in capturing the graph's long-range dependency or struggle to integrate main-triple and qualifier information due to the information compression issue. To further justify our findings and offer a potential direction for future HKGE research, we propose the FormerGNN framework. This framework employs a qualifier integrator to preserve the original HKG topology, and a GNN-based graph encoder to capture the graph's long-range dependencies, followed by an improved approach for integrating main-triple and qualifier information to mitigate compression issues. Our experimental results demonstrate that FormerGNN outperforms existing HKGE models.

cs.LG↗

arXiv2025

MEGA-Bench: Scaling Multimodal Evaluation to over 500 Real-World Tasks

We present MEGA-Bench, an evaluation suite that scales multimodal evaluation to over 500 real-world tasks, to address the highly heterogeneous daily use cases of end users. Our objective is to optimize for a set of high-quality data samples that cover a highly diverse and rich set of multimodal tasks, while enabling cost-effective and accurate model evaluation. In particular, we collected 505 realistic tasks encompassing over 8,000 samples from 16 expert annotators to extensively cover the multimodal task space. Instead of unifying these problems into standard multi-choice questions (like MMMU, MMBench, and MMT-Bench), we embrace a wide range of output formats like numbers, phrases, code, \LaTeX, coordinates, JSON, free-form, etc. To accommodate these formats, we developed over 40 metrics to evaluate these tasks. Unlike existing benchmarks, MEGA-Bench offers a fine-grained capability report across multiple dimensions (e.g., application, input type, output format, skill), allowing users to interact with and visualize model capabilities in depth. We evaluate a wide variety of frontier vision-language models on MEGA-Bench to understand their capabilities across these dimensions.

cs.CV↗

arXiv2025

Balancing Semantic Relevance and Engagement in Related Video Recommendations

Related video recommendations commonly use collaborative filtering (CF) driven by co-engagement signals, often resulting in recommendations lacking semantic coherence and exhibiting strong popularity bias. This paper introduces a novel multi-objective retrieval framework, enhancing standard two-tower models to explicitly balance semantic relevance and user engagement. Our approach uniquely combines: (a) multi-task learning (MTL) to jointly optimize co-engagement and semantic relevance, explicitly prioritizing topical coherence; (b) fusion of multimodal content features (textual and visual embeddings) for richer semantic understanding; and (c) off-policy correction (OPC) via inverse propensity weighting to effectively mitigate popularity bias. Evaluation on industrial-scale data and a two-week live A/B test reveals our framework's efficacy. We observed significant improvements in semantic relevance (from 51% to 63% topic match rate), a reduction in popular item distribution (-13.8% popular video recommendations), and a +0.04% improvement in our topline user engagement metric. Our method successfully achieves better semantic coherence, balanced engagement, and practical scalability for real-world deployment.

cs.IR↗

arXiv2025

FIXME: Towards End-to-End Benchmarking of LLM-Aided Design Verification

Despite the transformative potential of Large Language Models (LLMs) in hardware design, a comprehensive evaluation of their capabilities in design verification remains underexplored. Current efforts predominantly focus on RTL generation and basic debugging, overlooking the critical domain of functional verification, which is the primary bottleneck in modern design methodologies due to the rapid escalation of hardware complexity. We present FIXME, the first end-to-end, multi-model, and open-source evaluation framework for assessing LLM performance in hardware functional verification (FV) to address this crucial gap. FIXME introduces a structured three-level difficulty hierarchy spanning six verification sub-domains and 180 diverse tasks, enabling in-depth analysis across the design lifecycle. Leveraging a collaborative AI-human approach, we construct a high-quality dataset using 100% silicon-proven designs, ensuring comprehensive coverage of real-world challenges. Furthermore, we enhance the functional coverage by 45.57% through expert-guided optimization. By rigorously evaluating state-of-the-art LLMs such as GPT-4, Claude3, and LlaMA3, we identify key areas for improvement and outline promising research directions to unlock the full potential of LLM-driven automation in hardware design verification. The benchmark is available at https://github.com/ChatDesignVerification/FIXME.

cs.AR↗

arXiv2025

C-TLSAN: Content-Enhanced Time-Aware Long- and Short-Term Attention Network for Personalized Recommendation

Sequential recommender systems aim to model users' evolving preferences by capturing patterns in their historical interactions. Recent advances in this area have leveraged deep neural networks and attention mechanisms to effectively represent sequential behaviors and time-sensitive interests. In this work, we propose C-TLSAN (Content-Enhanced Time-Aware Long- and Short-Term Attention Network), an extension of the TLSAN architecture that jointly models long- and short-term user preferences while incorporating semantic content associated with items, such as product descriptions. C-TLSAN enriches the recommendation pipeline by embedding textual content linked to users' historical interactions directly into both long-term and short-term attention layers. This allows the model to learn from both behavioral patterns and rich item content, enhancing user and item representations across temporal dimensions. By fusing sequential signals with textual semantics, our approach improves the expressiveness and personalization capacity of recommendation systems. We conduct extensive experiments on large-scale Amazon datasets, benchmarking C-TLSAN against state-of-the-art baselines, including recent sequential recommenders based on Large Language Models (LLMs), which represent interaction history and predictions in text form. Empirical results demonstrate that C-TLSAN consistently outperforms strong baselines in next-item prediction tasks. Notably, it improves AUC by 1.66%, Recall@10 by 93.99%, and Precision@10 by 94.80% on average over the best-performing baseline (TLSAN) across 10 Amazon product categories. These results highlight the value of integrating content-aware enhancements into temporal modeling frameworks for sequential recommendation. Our code is available at https://github.com/booml247/cTLSAN.

cs.LG↗