arXiv subjects

Yunfan Wu

Publications and source records attributed to Yunfan Wu.

27 records · Page 2Linked to original sources

arXiv2023

Graph Adversarial Immunization for Certifiable Robustness

Despite achieving great success, graph neural networks (GNNs) are vulnerable to adversarial attacks. Existing defenses focus on developing adversarial training or model modification. In this paper, we propose and formulate graph adversarial immunization, i.e., vaccinating part of graph structure to improve certifiable robustness of graph against any admissible adversarial attack. We first propose edge-level immunization to vaccinate node pairs. Unfortunately, such edge-level immunization cannot defend against emerging node injection attacks, since it only immunizes existing node pairs. To this end, we further propose node-level immunization. To avoid computationally intensive combinatorial optimization associated with adversarial immunization, we develop AdvImmune-Edge and AdvImmune-Node algorithms to effectively obtain the immune node pairs or nodes. Extensive experiments demonstrate the superiority of AdvImmune methods. In particular, AdvImmune-Node remarkably improves the ratio of robust nodes by 79%, 294%, and 100%, after immunizing only 5% of nodes. Furthermore, AdvImmune methods show excellent defensive performance against various attacks, outperforming state-of-the-art defenses. To the best of our knowledge, this is the first attempt to improve certifiable robustness from graph data perspective without losing performance on clean graphs, providing new insights into graph adversarial learning.

cs.LG↗

arXiv2023

Popularity Debiasing from Exposure to Interaction in Collaborative Filtering

Recommender systems often suffer from popularity bias, where popular items are overly recommended while sacrificing unpopular items. Existing researches generally focus on ensuring the number of recommendations exposure of each item is equal or proportional, using inverse propensity weighting, causal intervention, or adversarial training. However, increasing the exposure of unpopular items may not bring more clicks or interactions, resulting in skewed benefits and failing in achieving real reasonable popularity debiasing. In this paper, we propose a new criterion for popularity debiasing, i.e., in an unbiased recommender system, both popular and unpopular items should receive Interactions Proportional to the number of users who Like it, namely IPL criterion. Under the guidance of the criterion, we then propose a debiasing framework with IPL regularization term which is theoretically shown to achieve a win-win situation of both popularity debiasing and recommendation performance. Experiments conducted on four public datasets demonstrate that when equipping two representative collaborative filtering models with our framework, the popularity bias is effectively alleviated while maintaining the recommendation performance.

cs.IR↗

arXiv2022

Distributed quantum sensing with a mode-entangled network of spin-squeezed atomic states

Quantum sensors are used for precision timekeeping, field sensing, and quantum communication. Comparisons among a distributed network of these sensors are capable of, for example, synchronizing clocks at different locations. The performance of a sensor network is limited by technical challenges as well as the inherent noise associated with the quantum states used to realize the network. For networks with only local entanglement at each node, the noise performance of the network improves at best with square root of the number of nodes. Here, we demonstrate that nonlocal entanglement between network nodes offers better scaling with network size. A shared quantum nondemolition measurement entangles a clock network with up to four nodes. This network provides up to 4.5 dB better precision than one without nonlocal entanglement, and 11.6 dB improvement as compared to a network of sensors operating at the quantum projection noise limit. We demonstrate the generality of the approach with atomic clock and atomic interferometer protocols, in scientific and technologically relevant configurations optimized for intrinsically differential comparisons of sensor outputs.

quant-ph↗

arXiv2022

INMO: A Model-Agnostic and Scalable Module for Inductive Collaborative Filtering

Collaborative filtering is one of the most common scenarios and popular research topics in recommender systems. Among existing methods, latent factor models, i.e., learning a specific embedding for each user/item by reconstructing the observed interaction matrix, have shown excellent performances. However, such user-specific and item-specific embeddings are intrinsically transductive, making it difficult to deal with new users and new items unseen during training. Besides, the number of model parameters heavily depends on the number of all users and items, restricting its scalability to real-world applications. To solve the above challenges, in this paper, we propose a novel model-agnostic and scalable Inductive Embedding Module for collaborative filtering, namely INMO. INMO generates the inductive embeddings for users (items) by characterizing their interactions with some template items (template users), instead of employing an embedding lookup table. Under the theoretical analysis, we further propose an effective indicator for the selection of template users/items. Our proposed INMO can be attached to existing latent factor models as a pre-module, inheriting the expressiveness of backbone models, while bringing the inductive ability and reducing model parameters. We validate the generality of INMO by attaching it to both Matrix Factorization (MF) and LightGCN, which are two representative latent factor models for collaborative filtering. Extensive experiments on three public benchmarks demonstrate the effectiveness and efficiency of INMO in both transductive and inductive recommendation scenarios.

cs.IR↗

arXiv2021

Investigating Attention Mechanism in 3D Point Cloud Object Detection

Object detection in three-dimensional (3D) space attracts much interest from academia and industry since it is an essential task in AI-driven applications such as robotics, autonomous driving, and augmented reality. As the basic format of 3D data, the point cloud can provide detailed geometric information about the objects in the original 3D space. However, due to 3D data's sparsity and unorderedness, specially designed networks and modules are needed to process this type of data. Attention mechanism has achieved impressive performance in diverse computer vision tasks; however, it is unclear how attention modules would affect the performance of 3D point cloud object detection and what sort of attention modules could fit with the inherent properties of 3D data. This work investigates the role of the attention mechanism in 3D point cloud object detection and provides insights into the potential of different attention modules. To achieve that, we comprehensively investigate classical 2D attentions, novel 3D attentions, including the latest point cloud transformers on SUN RGB-D and ScanNetV2 datasets. Based on the detailed experiments and analysis, we conclude the effects of different attention modules. This paper is expected to serve as a reference source for benefiting attention-embedded 3D point cloud object detection. The code and trained models are available at: https://github.com/ShiQiu0419/attentions_in_3D_detection.

cs.CV↗

arXiv2021

Single Node Injection Attack against Graph Neural Networks

Node injection attack on Graph Neural Networks (GNNs) is an emerging and practical attack scenario that the attacker injects malicious nodes rather than modifying original nodes or edges to affect the performance of GNNs. However, existing node injection attacks ignore extremely limited scenarios, namely the injected nodes might be excessive such that they may be perceptible to the target GNN. In this paper, we focus on an extremely limited scenario of single node injection evasion attack, i.e., the attacker is only allowed to inject one single node during the test phase to hurt GNN's performance. The discreteness of network structure and the coupling effect between network structure and node features bring great challenges to this extremely limited scenario. We first propose an optimization-based method to explore the performance upper bound of single node injection evasion attack. Experimental results show that 100%, 98.60%, and 94.98% nodes on three public datasets are successfully attacked even when only injecting one node with one edge, confirming the feasibility of single node injection evasion attack. However, such an optimization-based method needs to be re-optimized for each attack, which is computationally unbearable. To solve the dilemma, we further propose a Generalizable Node Injection Attack model, namely G-NIA, to improve the attack efficiency while ensuring the attack performance. Experiments are conducted across three well-known GNNs. Our proposed G-NIA significantly outperforms state-of-the-art baselines and is 500 times faster than the optimization-based method when inferring.

cs.LG↗

arXiv2021

Utilizing machine learning to improve the precision of fluorescence imaging of cavity-generated spin squeezed states

We present a supervised learning model to calibrate the photon collection rate during the fluorescence imaging of cold atoms. The linear regression model finds the collection rate at each location on the sensor such that the atomic population difference equals that of a highly precise optical cavity measurement. This 192 variable regression results in a measurement variance 27% smaller than our previous single variable regression calibration. The measurement variance is now in agreement with the theoretical limit due to other known noise sources. This model efficiently trains in less than a minute on a standard personal computer's CPU, and requires less than 10 minutes of data collection. Furthermore, the model is applicable across a large changes in population difference and across data collected on different days.

quant-ph↗

arXiv2019

Demonstration of a free space rubidium atomic clock with noise below the quantum projection limit

A technique is demonstrated that allows free space atomic fountain clocks and interferometers to utilize optical cavity generated spin-squeezed states with over $390\,000$ ${}^{87}\text{Rb}$ atoms. Fluorescence imaging is used for population spectroscopy, after a free fall time of 4 milliseconds, to resolve a single-shot phase sensitivity of $814 (61)$ microradians, which is $5.8(0.6)$ decibels (dB) below the quantum projection limit. The dynamic range is observed to be 100 milliradians. When operating as a microwave atomic clock with $240\,000$ atoms at a 3.6 ms Ramsey time, a single-shot fractional frequency stability of $8.4(0.2)\times10^{-12}$ is reported, $3.8(0.2)$ dB below the quantum projection limit.

quant-ph↗

arXiv2019

Retrieval of cavity-generated atomic spin-squeezing after free-space release

The compatibility of cavity-generated spin-squeezed atomic states with atom-interferometric sensors that require freely falling atoms is demonstrated. An ensemble of $500,000$ spin-squeezed atoms in a high-finesse optical cavity with near-uniform atom-cavity coupling is prepared, released into free space, recaptured in the cavity, and probed. Up to $\sim$10 dB of metrologically-relevant squeezing is retrieved for 700 microsecond free-fall times, and decaying levels of squeezing are realized for up to 3 millisecond free-fall times. The degradation of squeezing results from loss of atom-cavity coupling homogeneity between the initial squeezed state generation and final collective state read-out. A theoretical model is developed to quantify this degradation and this model is experimentally validated.

quant-ph↗