arXiv subjects

Hao-Yu Wu

Publications and source records attributed to Hao-Yu Wu.

11 recordsLinked to original sources

arXiv2026

PinPoint: Evaluation of Composed Image Retrieval with Explicit Negatives, Multi-Image Queries, and Paraphrase Testing

Composed Image Retrieval (CIR) has made significant progress, yet current benchmarks are limited to single ground-truth answers and lack the annotations needed to evaluate false positive avoidance, robustness and multi-image reasoning. We present PinPoint, a comprehensive real world benchmark with 7,635 queries and 329K relevance judgments across 23 query categories. PinPoint advances the field by providing: (1) multiple correct answers (averaging 9.1 per query) (2) explicit hard negatives, (3) six instruction paraphrases per query for robustness testing, (4) multi-image composition support (13.4% of queries), and (5) demographic metadata for fairness evaluation. Based on our analysis of 20+ methods across 4 different major paradigms, we uncover three significant drawbacks: The best methods while achieving mAP@10 of 28.5%, still retrieves irrelevant results (hard negatives) 9% of the time. The best models also exhibit 25.1% performance variation across paraphrases, indicating significant potential for enhancing current CIR techniques. Multi-image queries performs 40 to 70% worse across different methods. To overcome these new issues uncovered by our evaluation framework, we propose a training-free reranking method based on an off-the-shelf MLLM that can be applied to any existing system to bridge the gap. We release the complete dataset, including all images, queries, annotations, retrieval index, and benchmarking code.

cs.CV↗

arXiv2025

Gaussian tripartite steering in Schwarzschild black hole

Multipartite steering is a fundamental quantum resource that is uniquely suited to tackling complex relativistic quantum information challenges, but its properties in the gravitational field context remain to be elucidated. We study the distribution of Gaussian tripartite steering in the background of a Schwarzschild black hole. Our results show that physically accessible $2\rightarrow 1$ steerability remains robust at any Hawking temperature, which contrasts with the behavior of $1\rightarrow 1$ steering that experiences ``sudden death" as the Hawking temperature increases, making $2\rightarrow 1$ steerability a more reliable candidate for relativistic quantum information tasks. Notably, we observe that the sudden death of quantum steering is accompanied by a peak in steering asymmetry, signifying a critical transition from two-way steering to one-way steering in the relativistic three-mode system. We find that the Hawking effect can generate inaccessible two-way steering of a bipartite system in both directions, as shown in [Phys. Rev. D 93, 125011 (2016)], while it can only generate physically inaccessible one-way $2\rightarrow 1$ steering in tripartite systems. These insights enhance our understanding of the redistribution of multipartite steering near the event horizon of the black hole.

gr-qc↗

arXiv2023

Does Hawking effect always degrade fidelity of quantum teleportation in Schwarzschild spacetime?

Previous studies have shown that the Hawking effect always destroys quantum correlations and the fidelity of quantum teleportation in the Schwarzschild black hole. Here, we investigate the fidelity of quantum teleportation of Dirac fields between users in Schwarzschild spacetime. We find that, with the increase of the Hawking temperature, the fidelity of quantum teleportation can monotonically increase, monotonically decrease, or non-monotonically increase, depending on the choice of the initial state, which means that the Hawking effect can create net fidelity of quantum teleportation. This striking result banishes the extended belief that the Hawking effect of the black hole can only destroy the fidelity of quantum teleportation. We also find that quantum steering cannot fully guarantee the fidelity of quantum teleportation in Schwarzschild spacetime. This new unexpected source may provide a new idea for the experimental evidence of the Hawking effect.

gr-qc↗

arXiv2023

Spatial-spectral mapping to prepare the frequency entangled qudits

Entangled qudits, the high-dimensional entangled states, play an important role in the study of quantum information. How to prepare entangled qudits in an efficient and easy-to-operate manner is still a challenge in quantum technology. Here, we demonstrate a method to engineer frequency entangled qudits in a spontaneous parametric downconversion process. The proposal employs an angle-dependent phase-matching condition in a nonlinear crystal, which forms a classical-quantum mapping between the spatial (pump) and spectral (biphotons) degrees of freedom. In particular, the pump profile is separated into several bins in the spatial domain, and thus shapes the down-converted biphotons into discrete frequency modes in the joint spectral space. Our approach provides a feasible and efficient method to prepare a high-dimensional frequency entangled state. As an experimental demonstration, we generate a three-dimensional entangled state by using a homemade variable slit mask.

quant-ph↗

arXiv2023

Quantum entanglement for continuous variables sharing in an expanding spacetime

Detecting the structure of spacetime with quantum technologies has always been one of the frontier topics of relativistic quantum information. Here, we analytically study the generation and redistribution of Gaussian entanglement of the scalar fields in an expanding spacetime. We consider a two-mode squeezed state via a Gaussian amplification channel that corresponds to the time-evolution of the state from the asymptotic past to the asymptotic future. Therefore, the dynamical entanglement of the Gaussian state in an expanding universe encodes historical information about the underlying spacetime structure, suggesting a promising application in observational cosmology. We find that quantum entanglement is more sensitive to the expansion rate than the expansion volume. According to the analysis of quantum entanglement, choosing the particles with the smaller momentum and the optimal mass is a better way to extract information about the expanding universe. These results can guide the simulation of the expanding universe in quantum systems.

quant-ph↗

arXiv2022

Large Scale Open-Set Deep Logo Detection

We present an open-set logo detection (OSLD) system, which can detect (localize and recognize) any number of unseen logo classes without re-training; it only requires a small set of canonical logo images for each logo class. We achieve this using a two-stage approach: (1) Generic logo detection to detect candidate logo regions in an image. (2) Logo matching for matching the detected logo regions to a set of canonical logo images to recognize them. We constructed an open-set logo detection dataset with 12.1k logo classes and released it for research purposes.We demonstrate the effectiveness of OSLD on our dataset and on the standard Flickr-32 logo dataset, outperforming the state-of-the-art open-set and closed-set logo detection methods by a large margin. OSLD is scalable to millions of logo classes.

cs.CV↗

arXiv2021

Billion-Scale Pretraining with Vision Transformers for Multi-Task Visual Representations

Large-scale pretraining of visual representations has led to state-of-the-art performance on a range of benchmark computer vision tasks, yet the benefits of these techniques at extreme scale in complex production systems has been relatively unexplored. We consider the case of a popular visual discovery product, where these representations are trained with multi-task learning, from use-case specific visual understanding (e.g. skin tone classification) to general representation learning for all visual content (e.g. embeddings for retrieval). In this work, we describe how we (1) generate a dataset with over a billion images via large weakly-supervised pretraining to improve the performance of these visual representations, and (2) leverage Transformers to replace the traditional convolutional backbone, with insights into both system and performance improvements, especially at 1B+ image scale. To support this backbone model, we detail a systematic approach to deriving weakly-supervised image annotations from heterogenous text signals, demonstrating the benefits of clustering techniques to handle the long-tail distribution of image labels. Through a comprehensive study of offline and online evaluation, we show that large-scale Transformer-based pretraining provides significant benefits to industry computer vision applications. The model is deployed in a production visual shopping system, with 36% improvement in top-1 relevance and 23% improvement in click-through volume. We conduct extensive experiments to better understand the empirical relationships between Transformer-based architectures, dataset scale, and the performance of production vision systems.

cs.CV↗

arXiv2020

Shop The Look: Building a Large Scale Visual Shopping System at Pinterest

As online content becomes ever more visual, the demand for searching by visual queries grows correspondingly stronger. Shop The Look is an online shopping discovery service at Pinterest, leveraging visual search to enable users to find and buy products within an image. In this work, we provide a holistic view of how we built Shop The Look, a shopping oriented visual search system, along with lessons learned from addressing shopping needs. We discuss topics including core technology across object detection and visual embeddings, serving infrastructure for realtime inference, and data labeling methodology for training/evaluation data collection and human evaluation. The user-facing impacts of our system design choices are measured through offline evaluations, human relevance judgements, and online A/B experiments. The collective improvements amount to cumulative relative gains of over 160% in end-to-end human relevance judgements and over 80% in engagement. Shop The Look is deployed in production at Pinterest.

cs.CV↗

arXiv2019

Learning a Unified Embedding for Visual Search at Pinterest

At Pinterest, we utilize image embeddings throughout our search and recommendation systems to help our users navigate through visual content by powering experiences like browsing of related content and searching for exact products for shopping. In this work we describe a multi-task deep metric learning system to learn a single unified image embedding which can be used to power our multiple visual search products. The solution we present not only allows us to train for multiple application objectives in a single deep neural network architecture, but takes advantage of correlated information in the combination of all training data from each application to generate a unified embedding that outperforms all specialized embeddings previously deployed for each product. We discuss the challenges of handling images from different domains such as camera photos, high quality web images, and clean product catalog images. We also detail how to jointly train for multiple product objectives and how to leverage both engagement data and human labeled data. In addition, our trained embeddings can also be binarized for efficient storage and retrieval without compromising precision and recall. Through comprehensive evaluations on offline metrics, user studies, and online A/B experiments, we demonstrate that our proposed unified embedding improves both relevance and engagement of our visual search products for both browsing and searching purposes when compared to existing specialized embeddings. Finally, the deployment of the unified embedding at Pinterest has drastically reduced the operation and engineering cost of maintaining multiple embeddings while improving quality.

cs.CV↗

arXiv2019

Classification is a Strong Baseline for Deep Metric Learning

Deep metric learning aims to learn a function mapping image pixels to embedding feature vectors that model the similarity between images. Two major applications of metric learning are content-based image retrieval and face verification. For the retrieval tasks, the majority of current state-of-the-art (SOTA) approaches are triplet-based non-parametric training. For the face verification tasks, however, recent SOTA approaches have adopted classification-based parametric training. In this paper, we look into the effectiveness of classification based approaches on image retrieval datasets. We evaluate on several standard retrieval datasets such as CAR-196, CUB-200-2011, Stanford Online Product, and In-Shop datasets for image retrieval and clustering, and establish that our classification-based approach is competitive across different feature dimensions and base feature networks. We further provide insights into the performance effects of subsampling classes for scalable classification-based training, and the effects of binarization, enabling efficient storage and computation for practical applications.

cs.CV↗

arXiv2013

A complicated Duffing oscillator in the surface-electrode ion trap

The oscillation coupling and different nonlinear effects are observed in a single trapped $^{40}Ca^{+}$ ion confined in our home-built surface-electrode trap (SET). The coupling and the nonlinearity are originated from the high-order multipole potentials due to different layouts and the fabrication asymmetry of the SET. We solve a complicated Duffing equation with coupled oscillation terms by the multiple scale method, which fits the experimental values very well. Our investigation in the SET helps for exploring nonlinearity using currently available techniques and for suppressing instability of qubits in quantum information processing with trapped ions.

quant-ph↗