arXiv subjects

Shuai Zhao

Publications and source records attributed to Shuai Zhao.

At least 109 records · Page 6Linked to original sources

arXiv2024

Defending Against Weight-Poisoning Backdoor Attacks for Parameter-Efficient Fine-Tuning

Recently, various parameter-efficient fine-tuning (PEFT) strategies for application to language models have been proposed and successfully implemented. However, this raises the question of whether PEFT, which only updates a limited set of model parameters, constitutes security vulnerabilities when confronted with weight-poisoning backdoor attacks. In this study, we show that PEFT is more susceptible to weight-poisoning backdoor attacks compared to the full-parameter fine-tuning method, with pre-defined triggers remaining exploitable and pre-defined targets maintaining high confidence, even after fine-tuning. Motivated by this insight, we developed a Poisoned Sample Identification Module (PSIM) leveraging PEFT, which identifies poisoned samples through confidence, providing robust defense against weight-poisoning backdoor attacks. Specifically, we leverage PEFT to train the PSIM with randomly reset sample labels. During the inference process, extreme confidence serves as an indicator for poisoned samples, while others are clean. We conduct experiments on text classification tasks, five fine-tuning strategies, and three weight-poisoning backdoor attack methods. Experiments show near 100% success rates for weight-poisoning backdoor attacks when utilizing PEFT. Furthermore, our defensive approach exhibits overall competitive performance in mitigating weight-poisoning backdoor attacks.

cs.CR↗

arXiv2024

PE-GPT: A Physics-Informed Interactive Large Language Model for Power Converter Modulation Design

This paper proposes PE-GPT, a custom-tailored large language model uniquely adapted for power converter modulation design. By harnessing in-context learning and specialized tiered physics-informed neural networks, PE-GPT guides users through text-based dialogues, recommending actionable modulation parameters. The effectiveness of PE-GPT is validated through a practical design case involving dual active bridge converters, supported by hardware experimentation. This research underscores the transformative potential of large language models in power converter modulation design, offering enhanced accessibility, explainability, and efficiency, thereby setting a new paradigm in the field.

eess.SY↗

arXiv2024

Production of true para-muonium in linearly polarized photon fusions

True muonium (TM) -- the bound state of $μ^+μ^-$ -- has not been discovered yet. It was demonstrated that searching for TM via $γγ$ fusions in heavy ion collisions is feasible due to the enhancement of the atom number. We study the production of the true para-muonium (pTM) in the collisions of linearly polarized photons in the experiments of heavy-ion collisions, calculate the production rate as well as the transverse spectrum of pTM, and explore the discovery potential in nuclear experiments. Our results show that there is a significant correlation between the linearly polarized photon distribution and the transverse momentum distribution of pTM. The optimal kinematic region of the generated pTM is identified, which can provide a theoretical guide to the detection of pTM in experiments.

hep-ph↗

arXiv2024

Test-Time Adaptation with CLIP Reward for Zero-Shot Generalization in Vision-Language Models

One fascinating aspect of pre-trained vision-language models~(VLMs) learning under language supervision is their impressive zero-shot generalization capability. However, this ability is hindered by distribution shifts between the training and testing data. Previous test time adaptation~(TTA) methods for VLMs in zero-shot classification rely on minimizing the entropy of model outputs, tending to be stuck in incorrect model predictions. In this work, we propose TTA with feedback to rectify the model output and prevent the model from becoming blindly confident. Specifically, a CLIP model is adopted as the reward model during TTA and provides feedback for the VLM. Given a single test sample, the VLM is forced to maximize the CLIP reward between the input and sampled results from the VLM output distribution. The proposed \textit{reinforcement learning with CLIP feedback~(RLCF)} framework is highly flexible and universal. Beyond the classification task, with task-specific sampling strategies and a proper reward baseline choice, RLCF can be easily extended to not only discrimination tasks like retrieval but also generalization tasks like image captioning, improving the zero-shot generalization capacity of VLMs. According to the characteristics of these VL tasks, we build different fully TTA pipelines with RLCF to improve the zero-shot generalization ability of various VLMs. Extensive experiments along with promising empirical results demonstrate the effectiveness of RLCF. The code is available at https://github.com/mzhaoshuai/RLCF.

cs.CV↗

arXiv2024

Local unitary equivalence of arbitrary-dimensional multipartite quantum states

Local unitary equivalence is an important ingredient for quantifying and classifying entanglement. Verifying whether or not two quantum states are local unitary equivalent is a crucial problem, where only the case of multipartite pure states is solved. For mixed states, however, the verification of local unitary equivalence is still a challenging problem. In this paper, based on the coefficient matrices of generalized Bloch representations of quantum states, we find a variety of local unitary invariants for arbitrary-dimensional bipartite quantum states. These invariants are operational and can be used as necessary conditions for verifying the local unitary equivalence of two quantum states. Furthermore, we extend the construction to the arbitrary-dimensional multipartite case. We finally apply these invariants to estimate concurrence, a vital entanglement measure, showing the practicability of local unitary invariants in characterizing entanglement.

quant-ph↗

arXiv2024

Counting linearly polarized gluons with lattice QCD

We outline an approach to calculate the transverse-momentum-dependent distribution of linearly polarized gluons inside an unpolarized hadron on the lattice with the help of large momentum effective theory. To achieve this purpose, we propose calculating a Euclidean version of the degree of polarization for a fast-moving hadron on the lattice, which is ultraviolet finite, and no soft function subtraction is needed. It indicates a practical way to explore the distribution of the linearly polarized gluons in a proton and the linearly polarized gluon effects in hadron collisions on the lattice.

hep-ph↗

arXiv2024

Logarithmic moments of B-meson quasidistribution amplitude

It was demonstrated that the lattice simulation of $B$-meson light-cone distribution amplitude (LCDA) is feasible via the quasi-distribution amplitude (quasi-DA) in large momentum effective theory (LaMET). The structures of logarithmic moments (LMs) of $B$-meson quasi-DA is explored in this work. The one-loop results indicate mixing in the matching: the $n$-th LM would be not only factorized into the $n$-th LM of LCDA, but also other moments with different power, accompanied by short distance coefficients. These results supply the understanding of the matching in LaMET and may provide guidance to the lattice study of LMs or other parameters of $B$-meson LCDA.

hep-ph↗

arXiv2024

Loophole-free test of local realism via Hardy's violation

Bell's theorem states that quantum mechanical description on physical quantity cannot be fully explained by local realistic theories, and lays solid basis for various quantum information applications. Hardy's paradox is celebrated to be the simplest form of Bell's theorem concerning its "All versus Nothing" way to test local realism. However, due to experimental imperfections, existing tests of Hardy's paradox require additional assumptions of experimental systems, which constitute potential loopholes for faithfully testing local realistic theories. Here, we experimentally demonstrate Hardy's nonlocality through a photonic entanglement source. By achieving a detection efficiency of $82.2\%$, a quantum state fidelity of $99.10\%$ and applying high speed quantum random number generators for measurement setting switching, the experiment is implemented in a loophole-free manner. During $6$ hours of running, a strong violation of $P_{\text{Hardy}}=4.646\times 10^{-4}$ up to $5$ standard deviations is observed with $4.32\times 10^{9}$ trials. A null hypothesis test shows that the results can be explained by local realistic theories with an upper bound probability of $10^{-16348}$. These testing results present affirmative evidence against local realism, and provide an advancing benchmark for quantum information applications based on Hardy's paradox.

quant-ph↗

arXiv2023

Focus on Local Regions for Query-based Object Detection

Query-based methods have garnered significant attention in object detection since the advent of DETR, the pioneering query-based detector. However, these methods face challenges like slow convergence and suboptimal performance. Notably, self-attention in object detection often hampers convergence due to its global focus. To address these issues, we propose FoLR, a transformer-like architecture with only decoders. We improve the self-attention by isolating connections between irrelevant objects that makes it focus on local regions but not global regions. We also design the adaptive sampling method to extract effective features based on queries' local regions from feature maps. Additionally, we employ a look-back strategy for decoders to retain previous information, followed by the Feature Mixer module to fuse features and queries. Experimental results demonstrate FoLR's state-of-the-art performance in query-based detectors, excelling in convergence speed and computational efficiency. Index Terms: Local regions, Attention mechanism, Object detection

cs.CV↗

arXiv2023

Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models

The prompt-based learning paradigm, which bridges the gap between pre-training and fine-tuning, achieves state-of-the-art performance on several NLP tasks, particularly in few-shot settings. Despite being widely applied, prompt-based learning is vulnerable to backdoor attacks. Textual backdoor attacks are designed to introduce targeted vulnerabilities into models by poisoning a subset of training samples through trigger injection and label modification. However, they suffer from flaws such as abnormal natural language expressions resulting from the trigger and incorrect labeling of poisoned samples. In this study, we propose ProAttack, a novel and efficient method for performing clean-label backdoor attacks based on the prompt, which uses the prompt itself as a trigger. Our method does not require external triggers and ensures correct labeling of poisoned samples, improving the stealthy nature of the backdoor attack. With extensive experiments on rich-resource and few-shot text classification tasks, we empirically validate ProAttack's competitive performance in textual backdoor attacks. Notably, in the rich-resource setting, ProAttack achieves state-of-the-art attack success rates in the clean-label backdoor attack benchmark without external triggers.

cs.CL↗

arXiv2023

A Study of Unsupervised Evaluation Metrics for Practical and Automatic Domain Adaptation

Unsupervised domain adaptation (UDA) methods facilitate the transfer of models to target domains without labels. However, these methods necessitate a labeled target validation set for hyper-parameter tuning and model selection. In this paper, we aim to find an evaluation metric capable of assessing the quality of a transferred model without access to target validation labels. We begin with the metric based on mutual information of the model prediction. Through empirical analysis, we identify three prevalent issues with this metric: 1) It does not account for the source structure. 2) It can be easily attacked. 3) It fails to detect negative transfer caused by the over-alignment of source and target features. To address the first two issues, we incorporate source accuracy into the metric and employ a new MLP classifier that is held out during training, significantly improving the result. To tackle the final issue, we integrate this enhanced metric with data augmentation, resulting in a novel unsupervised UDA metric called the Augmentation Consistency Metric (ACM). Additionally, we empirically demonstrate the shortcomings of previous experiment settings and conduct large-scale experiments to validate the effectiveness of our proposed metric. Furthermore, we employ our metric to automatically search for the optimal hyper-parameter set, achieving superior performance compared to manually tuned sets across four common benchmarks. Codes will be available soon.

cs.CV↗

arXiv2023

Tilted Hardy paradoxes for device-independent randomness extraction

The device-independent paradigm has had spectacular successes in randomness generation, key distribution and self-testing, however most of these results have been obtained under the assumption that parties hold trusted and private random seeds. In efforts to relax the assumption of measurement independence, Hardy's non-locality tests have been proposed as ideal candidates. In this paper, we introduce a family of tilted Hardy paradoxes that allow to self-test general pure two-qubit entangled states, as well as certify up to $1$ bit of local randomness. We then use these tilted Hardy tests to obtain an improvement in the generation rate in the state-of-the-art randomness amplification protocols for Santha-Vazirani (SV) sources with arbitrarily limited measurement independence. Our result shows that device-independent randomness amplification is possible for arbitrarily biased SV sources and from almost separable states. Finally, we introduce a family of Hardy tests for maximally entangled states of local dimension $4, 8$ as the potential candidates for DI randomness extraction to certify up to the maximum possible $2 \log d$ bits of global randomness.

quant-ph↗

arXiv2023

Physics-Informed Neural Networks for Prognostics and Health Management of Lithium-Ion Batteries

For Prognostics and Health Management (PHM) of Lithium-ion (Li-ion) batteries, many models have been established to characterize their degradation process. The existing empirical or physical models can reveal important information regarding the degradation dynamics. However, there are no general and flexible methods to fuse the information represented by those models. Physics-Informed Neural Network (PINN) is an efficient tool to fuse empirical or physical dynamic models with data-driven models. To take full advantage of various information sources, we propose a model fusion scheme based on PINN. It is implemented by developing a semi-empirical semi-physical Partial Differential Equation (PDE) to model the degradation dynamics of Li-ion batteries. When there is little prior knowledge about the dynamics, we leverage the data-driven Deep Hidden Physics Model (DeepHPM) to discover the underlying governing dynamic models. The uncovered dynamics information is then fused with that mined by the surrogate neural network in the PINN framework. Moreover, an uncertainty-based adaptive weighting method is employed to balance the multiple learning tasks when training the PINN. The proposed methods are verified on a public dataset of Li-ion Phosphate (LFP)/graphite batteries.

eess.SP↗

arXiv2023

Accessing the subleading-twist $B$-meson light-cone distribution amplitude with Large-Momentum Effective Theory

Within the framework of large momentum effective theory (LaMET), we propose a hard-collinear factorization formula to extract the subleading-twist $B$-meson light-cone distribution amplitude (LCDA) from the quasidistribution amplitude calculated on the lattice. The one-loop matching coefficient and an integro-differential equation governing the evolution of the quasidistribution amplitude are derived. Our results could be useful either in evaluating the subleading-twist $B$-meson LCDA on the lattice or in the understanding the feasibility of LaMET on higher-twist distributions.

hep-ph↗

arXiv2023

A Transformer-Based Substitute Recommendation Model Incorporating Weakly Supervised Customer Behavior Data

The substitute-based recommendation is widely used in E-commerce to provide better alternatives to customers. However, existing research typically uses the customer behavior signals like co-view and view-but-purchase-another to capture the substitute relationship. Despite its intuitive soundness, we find that such an approach might ignore the functionality and characteristics of products. In this paper, we adapt substitute recommendation into language matching problem by taking product title description as model input to consider product functionality. We design a new transformation method to de-noise the signals derived from production data. In addition, we consider multilingual support from the engineering point of view. Our proposed end-to-end transformer-based model achieves both successes from offline and online experiments. The proposed model has been deployed in a large-scale E-commerce website for 11 marketplaces in 6 languages. Our proposed model is demonstrated to increase revenue by 19% based on an online A/B experiment.

cs.IR↗

arXiv2023

Evaluating Parameter-Efficient Transfer Learning Approaches on SURE Benchmark for Speech Understanding

Fine-tuning is widely used as the default algorithm for transfer learning from pre-trained models. Parameter inefficiency can however arise when, during transfer learning, all the parameters of a large pre-trained model need to be updated for individual downstream tasks. As the number of parameters grows, fine-tuning is prone to overfitting and catastrophic forgetting. In addition, full fine-tuning can become prohibitively expensive when the model is used for many tasks. To mitigate this issue, parameter-efficient transfer learning algorithms, such as adapters and prefix tuning, have been proposed as a way to introduce a few trainable parameters that can be plugged into large pre-trained language models such as BERT, and HuBERT. In this paper, we introduce the Speech UndeRstanding Evaluation (SURE) benchmark for parameter-efficient learning for various speech-processing tasks. Additionally, we introduce a new adapter, ConvAdapter, based on 1D convolution. We show that ConvAdapter outperforms the standard adapters while showing comparable performance against prefix tuning and LoRA with only 0.94% of trainable parameters on some of the task in SURE. We further explore the effectiveness of parameter efficient transfer learning for speech synthesis task such as Text-to-Speech (TTS).

cs.CL↗

arXiv2023

Gluon Transverse-Momentum-Dependent Distributions from Large-Momentum Effective Theory

We demonstrate that gluon transverse-momentum-dependent parton distribution functions (TMDPDFs) can be extracted from lattice calculations of appropriate Euclidean correlations in large-momentum effective theory (LaMET). Based on perturbative calculations of gluon unpolarized and helicity TMDPDFs, we present a matching formula connecting them and their LaMET counterparts, where the latter are renormalized in a scheme facilitating lattice calculations and converted to the $\overline{\rm MS}$ scheme. The hard matching kernel is given up to one-loop level. We also show that the perturbative result is independent of the prescription used for the pinch-pole singularity in the relevant correlations. Our results offer a guidance for the extraction of gluon TMDPDFs from lattice simulations, and have the potential to greatly facilitate perturbative calculations of the hard matching kernel.

hep-ph↗

arXiv2022

AutoPINN: When AutoML Meets Physics-Informed Neural Networks

Physics-Informed Neural Networks (PINNs) have recently been proposed to solve scientific and engineering problems, where physical laws are introduced into neural networks as prior knowledge. With the embedded physical laws, PINNs enable the estimation of critical parameters, which are unobservable via physical tools, through observable variables. For example, Power Electronic Converters (PECs) are essential building blocks for the green energy transition. PINNs have been applied to estimate the capacitance, which is unobservable during PEC operations, using current and voltage, which can be observed easily during operations. The estimated capacitance facilitates self-diagnostics of PECs. Existing PINNs are often manually designed, which is time-consuming and may lead to suboptimal performance due to a large number of design choices for neural network architectures and hyperparameters. In addition, PINNs are often deployed on different physical devices, e.g., PECs, with limited and varying resources. Therefore, it requires designing different PINN models under different resource constraints, making it an even more challenging task for manual design. To contend with the challenges, we propose Automated Physics-Informed Neural Networks (AutoPINN), a framework that enables the automated design of PINNs by combining AutoML and PINNs. Specifically, we first tailor a search space that allows finding high-accuracy PINNs for PEC internal parameter estimation. We then propose a resource-aware search strategy to explore the search space to find the best PINN model under different resource constraints. We experimentally demonstrate that AutoPINN is able to find more accurate PINN models than human-designed, state-of-the-art PINN models using fewer resources.

cs.LG↗