arXiv subjects

Xin Yi

Publications and source records attributed to Xin Yi.

At least 73 records · Page 4Linked to original sources

arXiv2025

Adanonymizer: Interactively Navigating and Balancing the Duality of Privacy and Output Performance in Human-LLM Interaction

Current Large Language Models (LLMs) cannot support users to precisely balance privacy protection and output performance during individual consultations. We introduce Adanonymizer, an anonymization plug-in that allows users to control this balance by navigating a trade-off curve. A survey (N=221) revealed a privacy paradox, where users frequently disclosed sensitive information despite acknowledging privacy risks. The study further demonstrated that privacy risks were not significantly correlated with model output performance, highlighting the potential to navigate this trade-off. Adanonymizer normalizes privacy and utility ratings by type and automates the pseudonymization of sensitive terms based on user preferences, significantly reducing user effort. Its 2D color palette interface visualizes the privacy-utility trade-off, allowing users to adjust the balance by manipulating a point. An evaluation (N=36) compared Adanonymizer with ablation methods and differential privacy techniques, where Adanonymizer significantly reduced modification time, achieved better perceived model performance and overall user preference.

cs.HC↗

arXiv2024

NLSR: Neuron-Level Safety Realignment of Large Language Models Against Harmful Fine-Tuning

The emergence of finetuning-as-a-service has revealed a new vulnerability in large language models (LLMs). A mere handful of malicious data uploaded by users can subtly manipulate the finetuning process, resulting in an alignment-broken model. Existing methods to counteract fine-tuning attacks typically require substantial computational resources. Even with parameter-efficient techniques like LoRA, gradient updates remain essential. To address these challenges, we propose \textbf{N}euron-\textbf{L}evel \textbf{S}afety \textbf{R}ealignment (\textbf{NLSR}), a training-free framework that restores the safety of LLMs based on the similarity difference of safety-critical neurons before and after fine-tuning. The core of our framework is first to construct a safety reference model from an initially aligned model to amplify safety-related features in neurons. We then utilize this reference model to identify safety-critical neurons, which we prepare as patches. Finally, we selectively restore only those neurons that exhibit significant similarity differences by transplanting these prepared patches, thereby minimally altering the fine-tuned model. Extensive experiments demonstrate significant safety enhancements in fine-tuned models across multiple downstream tasks, while greatly maintaining task-level accuracy. Our findings suggest regions of some safety-critical neurons show noticeable differences after fine-tuning, which can be effectively corrected by transplanting neurons from the reference model without requiring additional training. The code will be available at \url{https://github.com/xinykou/NLSR}

cs.CL↗

arXiv2024

"Ghost of the past": identifying and resolving privacy leakage from LLM's memory through proactive user interaction

Memories, encompassing past inputs in context window and retrieval-augmented generation (RAG), frequently surface during human-LLM interactions, yet users are often unaware of their presence and the associated privacy risks. To address this, we propose MemoAnalyzer, a system for identifying, visualizing, and managing private information within memories. A semi-structured interview (N=40) revealed that low privacy awareness was the primary challenge, while proactive privacy control emerged as the most common user need. MemoAnalyzer uses a prompt-based method to infer and identify sensitive information from aggregated past inputs, allowing users to easily modify sensitive content. Background color temperature and transparency are mapped to inference confidence and sensitivity, streamlining privacy adjustments. A 5-day evaluation (N=36) comparing MemoAnalyzer with the default GPT setting and a manual modification baseline showed MemoAnalyzer significantly improved privacy awareness and protection without compromising interaction speed. Our study contributes to privacy-conscious LLM design, offering insights into privacy protection for Human-AI interactions.

cs.HC↗

arXiv2024

Natural Language but Omitted? On the Ineffectiveness of Large Language Models' privacy policy from End-users' Perspective

LLMs driven products were increasingly prevalent in our daily lives, With a natural language based interaction style, people may potentially leak their personal private information. Thus, privacy policy and user agreement played an important role in regulating and alerting people. However, there lacked the work examining the reading of LLM's privacy policy. Thus, we conducted the first user study to let participants read the privacy policy and user agreement with two different styles (a cursory and detailed style). We found users lack important information upon cursory reading and even detailed reading. Besides, their privacy concerns was not solved even upon detailed reading. We provided four design implications based on the findings.

cs.HC↗

arXiv2024

A safety realignment framework via subspace-oriented model fusion for large language models

The current safeguard mechanisms for large language models (LLMs) are indeed susceptible to jailbreak attacks, making them inherently fragile. Even the process of fine-tuning on apparently benign data for downstream tasks can jeopardize safety. One potential solution is to conduct safety fine-tuning subsequent to downstream fine-tuning. However, there's a risk of catastrophic forgetting during safety fine-tuning, where LLMs may regain safety measures but lose the task-specific knowledge acquired during downstream fine-tuning. In this paper, we introduce a safety realignment framework through subspace-oriented model fusion (SOMF), aiming to combine the safeguard capabilities of initially aligned model and the current fine-tuned model into a realigned model. Our approach begins by disentangling all task vectors from the weights of each fine-tuned model. We then identify safety-related regions within these vectors by subspace masking techniques. Finally, we explore the fusion of the initial safely aligned LLM with all task vectors based on the identified safety subspace. We validate that our safety realignment framework satisfies the safety requirements of a single fine-tuned model as well as multiple models during their fusion. Our findings confirm that SOMF preserves safety without notably compromising performance on downstream tasks, including instruction following in Chinese, English, and Hindi, as well as problem-solving capabilities in Code and Math.

cs.CL↗

arXiv2024

Modulating trap properties by Cr3+-doping in Zn2SiO4: Mn2+ nano phosphor for optical information storage

Photo stimulated luminescent materials are one of the most attractive alternatives for next generation optical information storage technologies. However, there are still some challenges in regulating appropriate energy levels in luminescent materials for optical information storage. Herein, a green emission nanophosphor Zn2SiO4: Cr3+, Mn2+ with the trap depth of 1.05 eV, fulfilling the requirements for optical information storage, was fabricated for the first time through the solution combustion method and subsequent heat treatment at 1000 degree centigrade for 2h. The crystal structure, micromorphology, photoluminescence (PL), photoluminescence excitation (PLE), and afterglow properties of Zn2SiO4: xCr3+, yMn2+ were studied systematically. By applying the strategy of trap depth engineering, high trap density with proper trap depth was observed when Cr3+ ions were introduced into Zn2SiO4: Mn2+. Thermoluminescence (TL) glow curve analysis through the initial rise (IR) method was conducted to gain some insight into the information of traps. As proof of application, information storage was experimentally achieved by choosing 275 nm illumination for information writing and 980 nm NIR excitation for information reading. The results indicate that Zn2SiO4: Cr3+, Mn2+ phosphor holds promise for potential applications in the field of optical information storage.

cond-mat.mtrl-sci↗

arXiv2024

CogniDot: Vasoactivity-based Cognitive Load Monitoring with a Miniature On-skin Sensor

Vascular activities offer valuable signatures for psychological monitoring applications. We present CogniDot, an affordable, miniature skin sensor placed on the temporal area on the head that senses cognitive loads with a single-pixel color sensor. With its energy-efficient design, bio-compatible adhesive, and compact size (22mm diameter, 8.5mm thickness), it is ideal for long-term monitoring of mind status. We showed in detail the hardware design of our sensor. The user study results with 12 participants show that CogniDot can accurately differentiate between three levels of cognitive loads with a within-user accuracy of 97%. We also discuss its potential for broader applications.

cs.HC↗

arXiv2024

Fine-Grained Detoxification via Instance-Level Prefixes for Large Language Models

Impressive results have been achieved in natural language processing (NLP) tasks through the training of large language models (LLMs). However, these models occasionally produce toxic content such as insults, threats, and profanity in response to certain prompts, thereby constraining their practical utility. To tackle this issue, various finetuning-based and decoding-based approaches have been utilized to mitigate toxicity. However, these methods typically necessitate additional costs such as high-quality training data or auxiliary models. In this paper, we propose fine-grained detoxification via instance-level prefixes (FGDILP) to mitigate toxic text without additional cost. Specifically, FGDILP contrasts the contextualized representation in attention space using a positive prefix-prepended prompt against multiple negative prefix-prepended prompts at the instance level. This allows for constructing fine-grained subtoxicity vectors, which enables collaborative detoxification by fusing them to correct the normal generation process when provided with a raw prompt. We validate that FGDILP enables controlled text generation with regard to toxicity at both the utterance and context levels. Our method surpasses prompt-based baselines in detoxification, although at a slight cost to generation fluency and diversity.

cs.CL↗

arXiv2023

Exploring the interfacial coupling between graphene and the antiferromagnetic insulator MnPSe$_3$

Interfacial coupling between graphene and other 2D materials can give rise to intriguing physical phenomena. In particular, several theoretical studies predict that the interplay between graphene and an antiferromagnetic insulator could lead to the emergence of quantum anomalous Hall phases. However, such phases have not been observed experimentally yet, and further experimental studies are needed to reveal the interaction between graphene and antiferromagnetic insulators. Here, we report the study in heterostructures composed of graphene and the antiferromagnetic insulator MnPSe$_3$. It is found that the MnPSe$_3$ has little impact on the quantum Hall phases apart from doping graphene via interfacial charge transfer. However, the magnetic order can contribute indirectly via process like Kondo effect, as evidenced by the observed minimum in the temperature-resistance curve between 20-40 K, far below the Néel temperature (70 K).

cond-mat.mtrl-sci↗

arXiv2023

Towards Benchmarking and Assessing Visual Naturalness of Physical World Adversarial Attacks

Physical world adversarial attack is a highly practical and threatening attack, which fools real world deep learning systems by generating conspicuous and maliciously crafted real world artifacts. In physical world attacks, evaluating naturalness is highly emphasized since human can easily detect and remove unnatural attacks. However, current studies evaluate naturalness in a case-by-case fashion, which suffers from errors, bias and inconsistencies. In this paper, we take the first step to benchmark and assess visual naturalness of physical world attacks, taking autonomous driving scenario as the first attempt. First, to benchmark attack naturalness, we contribute the first Physical Attack Naturalness (PAN) dataset with human rating and gaze. PAN verifies several insights for the first time: naturalness is (disparately) affected by contextual features (i.e., environmental and semantic variations) and correlates with behavioral feature (i.e., gaze signal). Second, to automatically assess attack naturalness that aligns with human ratings, we further introduce Dual Prior Alignment (DPA) network, which aims to embed human knowledge into model reasoning process. Specifically, DPA imitates human reasoning in naturalness assessment by rating prior alignment and mimics human gaze behavior by attentive prior alignment. We hope our work fosters researches to improve and automatically assess naturalness of physical world attacks. Our code and dataset can be found at https://github.com/zhangsn-19/PAN.

cs.CV↗

arXiv2023

Modeling the Trade-off of Privacy Preservation and Activity Recognition on Low-Resolution Images

A computer vision system using low-resolution image sensors can provide intelligent services (e.g., activity recognition) but preserve unnecessary visual privacy information from the hardware level. However, preserving visual privacy and enabling accurate machine recognition have adversarial needs on image resolution. Modeling the trade-off of privacy preservation and machine recognition performance can guide future privacy-preserving computer vision systems using low-resolution image sensors. In this paper, using the at-home activity of daily livings (ADLs) as the scenario, we first obtained the most important visual privacy features through a user survey. Then we quantified and analyzed the effects of image resolution on human and machine recognition performance in activity recognition and privacy awareness tasks. We also investigated how modern image super-resolution techniques influence these effects. Based on the results, we proposed a method for modeling the trade-off of privacy preservation and activity recognition on low-resolution images.

cs.HC↗

arXiv2022

Joint spectrum shrinking maps on projections

Let $\mathcal H$ be a finite dimensional complex Hilbert space with dimension $n \ge 3$ and $\mathcal P(\mathcal H)$ the set of projections on $\mathcal H$. Let $φ: \mathcal P(\mathcal H) \to \mathcal P(\mathcal H)$ be a surjective map. We show that $φ$ shrinks the joint spectrum of any two projections if and only if it is joint spectrum preserving for any two projections and thus is induced by a ring automorphism on $\mathbb C$ in a particular way. In addition, for an arbitrary $k \ge 3$, $φ$ shrinks the joint spectrum of any $k$ projections if and only if it is induced by a unitary or an anti-unitary. Assume that $ϕ$ is a surjective map on the Grassmann space of rank one projections. We show that $ϕ$ is joint spectrum preserving for any $n$ rank one projections if and only if it can be extended to a surjective map on $\mathcal P(\mathcal{H})$ which is spectrum preserving for any two projections. Moreover, for any $k >n$, $ϕ$ is joint spectrum shrinking for any $k$ rank one projections if and only if it is induced by a unitary or an anti-unitary.

math.FA↗

arXiv2022

Vortex-driven periodic and aperiodic magnetoresistance oscillations in cuprates

The study of the interaction between superconductivity and charge ordering is helpful to resolve the pairing mechanism in high-temperature superconductors. Recently, several resistance oscillations studies trigger the speculation that a long-range charge ordering, with an enormous mesh size of several tens of nanometer, can possibly emerge in underdoped high Tc superconductor. However, spectroscopy studies have not traced this kind of long-range charge ordering. Here, we clarify the disagreement between the transport and spectroscopy studies on the mysterious long-range charge ordering by investigating the magneto-oscillations in underdoped Bi2Sr2CaCu2O8+δ flakes. Inspired by the observation that the oscillations evolve from a periodic to an aperiodic one with decreasing doping level, we conclude that the magneto-oscillations can be generated by the interaction between vortices and superconducting loops that enclose randomly distributed underdoped puddles while an assumption of long-range charge ordering is not necessary.

cond-mat.supr-con↗

arXiv2022

Josephson Effect in NbS$_{2}$ van der Waals Junctions

Van der Waals (vdW) Josephson junctions can possibly accelerate the development of advanced superconducting device that utilizes the unique properties of two-dimensional (2D) transition metal dichalcogenide (TMD) superconductors such as spin-orbit coupling, spin-valley locking. Here, we fabricate vertically stacked NbS$_{2}$/NbS$_{2}$ Josephson junctions using a modified all-dry transfer technique and characterize the device performance via systematic low-temperature transport measurements. The experimental results show that the superconducting transition temperature of the NbS$_{2}$/NbS$_{2}$ Josephson junction is 5.84 K, and the critical current density reaches 3975 A/cm$^{2}$ at 2K. Moreover, we extract a superconducting energy gap $Δ=0.58$ meV, which is considerably smaller than that expected from the single band s-wave Bardeen-Cooper-Schrieffer (BCS) model ($Δ=0.89$ meV).

cond-mat.supr-con↗

arXiv2022

Thermally enhanced photoluminescence and temperature sensing properties of Sc$_2$W$_3$O$_{12}$:Eu$^{3+}$ phosphors

Currently,lanthanide ions doped luminescence materials applying as optical thermometers have arose much concern. Basing on the different responses of two emissions to temperature, the fluorescence intensity ratio (FIR) technique can be executed and further estimate the sensitivities to assess the optical thermometry performances. In this study, we introduce different doping concentrations of Eu$^{3+}$ ions into negative expansion material Sc$_2$W$_3$O$_{12}$:Eu$^{3+}$, accessing to the thermal enhanced luminescence from 373 to 548 K, and investigate the temperature sensing properties in detail. All samples exhibit good thermally enhanced luminescence behavior. The emission intensity of Sc$_2$W$_3$O$_{12}$: 6 mol% Eu$^{3+}$ phosphors reaches at 147.81% of initial intensity at 473 K. As the Eu doping concentration increases, the resistance of the samples to thermal quenching decreases. The FIR technique based on the transitions 5D0-7F1 (592 nm) and 5D0-7F2 (613 nm) of Eu$^{3+}$ ions demonstrate a maximum relative temperature sensitivity of 3.063% K-1 at 298 K for Sc$_2$W$_3$O$_{12}$:Eu$^{3+}$: 6 mol% Eu$^{3+}$ phosphors. The sensitivity of sample decreases with the increase of Eu$^{3+}$ concentration. Benefiting from the thermal enhanced luminescence performance and good temperature sensing properties, the Sc$_2$W$_3$O$_{12}$:Eu$^{3+}$: Eu$^{3+}$ phosphors can be applies as optical thermometers.

cond-mat.mtrl-sci↗

arXiv2021

LGA-RCNN: Loss-Guided Attention for Object Detection

Object detection is widely studied in computer vision filed. In recent years, certain representative deep learning based detection methods along with solid benchmarks are proposed, which boosts the development of related researchs. However, existing detection methods still suffer from undesirable performance under challenges such as camouflage, blur, inter-class similarity, intra-class variance and complex environment. To address this issue, we propose LGA-RCNN which utilizes a loss-guided attention (LGA) module to highlight representative region of objects. Then, those highlighted local information are fused with global information for precise classification and localization.

cs.CV↗

arXiv2021

AFD-Net: Adaptive Fully-Dual Network for Few-Shot Object Detection

Few-shot object detection (FSOD) aims at learning a detector that can fast adapt to previously unseen objects with scarce annotated examples, which is challenging and demanding. Existing methods solve this problem by performing subtasks of classification and localization utilizing a shared component (e.g., RoI head) in the detector, yet few of them take the distinct preferences of two subtasks towards feature embedding into consideration. In this paper, we carefully analyze the characteristics of FSOD, and present that a general few-shot detector should consider the explicit decomposition of two subtasks, as well as leveraging information from both of them to enhance feature representations. To the end, we propose a simple yet effective Adaptive Fully-Dual Network (AFD-Net). Specifically, we extend Faster R-CNN by introducing Dual Query Encoder and Dual Attention Generator for separate feature extraction, and Dual Aggregator for separate model reweighting. Spontaneously, separate state estimation is achieved by the R-CNN detector. Besides, for the acquisition of enhanced feature representations, we further introduce Adaptive Fusion Mechanism to adaptively perform feature fusion in different subtasks. Extensive experiments on PASCAL VOC and MS COCO in various settings show that, our method achieves new state-of-the-art performance by a large margin, demonstrating its effectiveness and generalization ability.

cs.CV↗

arXiv2021

Two-Step Image Dehazing with Intra-domain and Inter-domain Adaptation

Caused by the difference of data distributions, intra-domain gap and inter-domain gap are widely present in image processing tasks. In the field of image dehazing, certain previous works have paid attention to the inter-domain gap between the synthetic domain and the real domain. However, those methods only establish the connection from the source domain to the target domain without taking into account the large distribution shift within the target domain (intra-domain gap). In this work, we propose a Two-Step Dehazing Network (TSDN) with an intra-domain adaptation and a constrained inter-domain adaptation. First, we subdivide the distributions within the synthetic domain into subsets and mine the optimal subset (easy samples) by loss-based supervision. To alleviate the intra-domain gap of the synthetic domain, we propose an intra-domain adaptation to align distributions of other subsets to the optimal subset by adversarial learning. Finally, we conduct the constrained inter-domain adaptation from the real domain to the optimal subset of the synthetic domain, alleviating the domain shift between domains as well as the distribution shift within the real domain. Extensive experimental results demonstrate that our framework performs favorably against the state-of-the-art algorithms both on the synthetic datasets and the real datasets.

cs.CV↗