arXiv subjects

Siyuan Gao

Publications and source records attributed to Siyuan Gao.

8 recordsLinked to original sources

arXiv2026

Photonic Crystal Defect Nanocavities Based on Monocrystalline Yttrium Iron Garnet

Monocrystalline yttrium iron garnet (YIG) is a key material for magneto optics and quantum magnonics owing to its high optical transparency, large magneto-optical (MO) effects at room temperature, and exceptionally-long spin coherence. While rich MO phenomena have been demonstrated in the microwave regime, extending these concepts to technologically important telecommunication wavelengths remains challenging due to the difficulty of fabricating high-quality YIG nanostructures. Here, we demonstrate photonic crystal (PhC) defect nanocavities based on monocrystalline Bi-substitudted YIG by developing a YIG-on-insulator platform and high-precision YIG nanopatterning. The fabricated nanocavities exhibit cavity resonances around lambda = 1500 nm with Q factors up to 1,800 and a mode volume V of 1.1(lambda/n)^3, corresponding to Q/V reaching around 10^3. The measured Q factor is primarily governed by intentionally introduced lattice modulations for out-of-plane light coupling, suggesting that further optimization of the cavity geometry and measurement configuration could yield an order-of-magnitude improvement of the experimental Q factor. YIG-based PhC nanocavities provide a platform for strongly confined light-magnetism interactions in the optical regime, opening pathways toward downsized nonreciprocal photonic devices and enhanced photon-magnon coupling.

physics.optics↗

arXiv2026

HomeSafeBench: Benchmarking Embodied Vision-Language Models in Free-Exploration Home Safety Inspection

Safety hazards in the home are a leading cause of preventable domestic injuries, motivating an automated inspector that actively explores a home and reports hazards before they cause harm. We introduce HomeSafeBench, the first benchmark for free-exploration home safety inspection with egocentric visual feedback, in which an embodied agent navigates a fully interactive 3D home, adjusts its viewpoint, and reports hazards purely from rendered first-person views. Built on the VirtualHome simulator, it covers five categories of common household hazards and comprises 1,000 human-validated inspection tasks. Evaluating a broad range of state-of-the-art Vision-Language Models (VLMs) reveals a large gap, where the best model reaches only about 34.7% F1, far below the 98.0% of a human inspector. Moreover, precision far exceeds recall across models, revealing a systematic tendency to under-report hazards that reflects a shared deficiency in risk recognition. To close this gap at low cost, we propose CueBack, an offline data-construction method that exploits the clue-precedes-confirmation structure of inspection, backtracking a privileged trajectory to the earliest frame where a hazard cue becomes visible and rewriting it into executable supervision. Fine-tuning a 4B-size VLM on CueBack-constructed data raises the average F1 from 18.7% to 45.3% on an out-of-distribution test set, surpassing the strongest closed-source model performance 34.7%. The benchmark, training dataset, and code are available at https://github.com/BITHLP/HomeSafeBench.

cs.CV↗

arXiv2026

Arbitrary Polarization Generation in Magneto-optical Metasurfaces Enabled by Bound States in the Continuum

The generation of arbitrary polarization states of light is essential for optical communication and photonic information processing. Photonic crystal and metasurface platforms supporting bound states in the continuum (BICs) provide a powerful route for polarization engineering through tailoring the radiation from the resonant modes. However, existing approaches typically rely on static structural symmetry breaking or off-normal radiation, which limits continuous polarization tuning of vertical radiation. Here, we demonstrate a magnetooptical metasurface that generates arbitrary polarization states of light at normal radiation. By applying an external magnetic field with variable rientation, a symmetry-protected BIC is transformed into a quasi-BIC whose radiation polarization can be continuously tuned. The magneto-optical perturbation drives the controlled migration of polarization singularities in momentum space, allowing the emitted states to continuously span the entire Poincaré sphere without structural modification. This approach establishes a compact platform for actively tunable polarization sources and polarizationencoded photonic devices.

physics.optics↗

arXiv2025

Design of Ultrathin Faraday Rotators based on All-dielectric Magneto-optical Metasurfaces at the Telecommunication Band

Magneto-optical (MO) interactions offer a direct route to nonreciprocal optical devices but are intrinsically weak in the optical domain, posing a major challenge in downsizing MO functional devices. In this study, we present a design strategy for ultra-thin MO Faraday rotators based on all-dielectric metasurfaces supporting high-quality factor quasi-bound states in the continuum (QBIC) modes. Light trapping in QBIC modes induced by band folding significantly enhances MO interactions in a controllable manner, enabling a technologically relevant 45$^\circ$ Faraday rotation with a MO metasurface that is only a few hundred nanometers thick. The design also incorporates electromagnetically induced transparency via spectrally overlapping resonant modes to achieve high light transmittance reaching 80%. This approach not only enables compact yet practical MO Faraday rotator but also holds promises for advancing free-space magnetic sensors and MO modulators.

physics.optics↗

arXiv2025

Decoupled Entity Representation Learning for Pinterest Ads Ranking

In this paper, we introduce a novel framework following an upstream-downstream paradigm to construct user and item (Pin) embeddings from diverse data sources, which are essential for Pinterest to deliver personalized Pins and ads effectively. Our upstream models are trained on extensive data sources featuring varied signals, utilizing complex architectures to capture intricate relationships between users and Pins on Pinterest. To ensure scalability of the upstream models, entity embeddings are learned, and regularly refreshed, rather than real-time computation, allowing for asynchronous interaction between the upstream and downstream models. These embeddings are then integrated as input features in numerous downstream tasks, including ad retrieval and ranking models for CTR and CVR predictions. We demonstrate that our framework achieves notable performance improvements in both offline and online settings across various downstream tasks. This framework has been deployed in Pinterest's production ad ranking systems, resulting in significant gains in online metrics.

cs.IR↗

arXiv2025

ELITE: Embedding-Less retrieval with Iterative Text Exploration

Large Language Models (LLMs) have achieved impressive progress in natural language processing, but their limited ability to retain long-term context constrains performance on document-level or multi-turn tasks. Retrieval-Augmented Generation (RAG) mitigates this by retrieving relevant information from an external corpus. However, existing RAG systems often rely on embedding-based retrieval trained on corpus-level semantic similarity, which can lead to retrieving content that is semantically similar in form but misaligned with the question's true intent. Furthermore, recent RAG variants construct graph- or hierarchy-based structures to improve retrieval accuracy, resulting in significant computation and storage overhead. In this paper, we propose an embedding-free retrieval framework. Our method leverages the logical inferencing ability of LLMs in retrieval using iterative search space refinement guided by our novel importance measure and extend our retrieval results with logically related information without explicit graph construction. Experiments on long-context QA benchmarks, including NovelQA and Marathon, show that our approach outperforms strong baselines while reducing storage and runtime by over an order of magnitude.

cs.CL↗

arXiv2021

Data-driven Clustering in Ad-hoc Networks based on Community Detection

High demands for industrial networks lead to increasingly large sensor networks. However, the complexity of networks and demands for accurate data require better stability and communication quality. Conventional clustering methods for ad-hoc networks are based on topology and connectivity, leading to unstable clustering results and low communication quality. In this paper, we focus on two situations: time-evolving networks, and multi-channel ad-hoc networks. We model ad-hoc networks as graphs and introduce community detection methods to both situations. Particularly, in time-evolving networks, our method utilizes the results of community detection to ensure stability. By using similarity or human-in-the-loop measures, we construct a new weighted graph for final clustering. In multi-channel networks, we perform allocations from the results of multiplex community detection. Experiments on real-world datasets show that our method outperforms baselines in both stability and quality.

eess.SP↗

arXiv2020

Inference of Dynamic Graph Changes for Functional Connectome

Dynamic functional connectivity is an effective measure for the brain's responses to continuous stimuli. We propose an inferential method to detect the dynamic changes of brain networks based on time-varying graphical models. Whereas most existing methods focus on testing the existence of change points, the dynamics in the brain network offer more signals in many neuroscience studies. We propose a novel method to conduct hypothesis testing on changes in dynamic brain networks. We introduce a bootstrap statistic to approximate the supreme of the high-dimensional empirical processes over dynamically changing edges. Our simulations show that this framework can capture the change points with changed connectivity. Finally, we apply our method to a brain imaging dataset under a natural audio-video stimulus and illustrate that we are able to detect temporal changes in brain networks. The functions of the identified regions are consistent with specific emotional annotations, which are closely associated with changes inferred by our method.

stat.AP↗