Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
22works
0followers
17topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

22 published item(s)

preprint2026arXiv

TeamTR: Trust-Region Fine-Tuning for Multi-Agent LLM Coordination

Multi-agent LLM systems have shown promise for complex reasoning, yet recent evaluations reveal they often underperform single-model baselines. We identify a structural failure mode in sequential fine-tuning of shared-context teams: updating one agent shifts the team's context distribution, and when subsequent updates are evaluated on cached rollouts, this mismatch compounds. We formalize this as the compounding occupancy shift and prove that stale-occupancy evaluation incurs a penalty that scales quadratically with the number of agents. In contrast, intermediate-occupancy evaluation reduces this to linear scaling. We propose TeamTR, a trust-region framework that resamples trajectories after each component update and enforces per-agent divergence control, yielding rigorous per-update and per-stage improvement lower bounds. Experiments show that TeamTR outperforms single-agent and sequential baselines with 7.1% on average, mitigates coordination regressions, and supports plug-and-play component replacement. Code is available at https://github.com/Yydc/TeamTR.

preprint2024arXiv

SAME: Sample Reconstruction against Model Extraction Attacks

While deep learning models have shown significant performance across various domains, their deployment needs extensive resources and advanced computing infrastructure. As a solution, Machine Learning as a Service (MLaaS) has emerged, lowering the barriers for users to release or productize their deep learning models. However, previous studies have highlighted potential privacy and security concerns associated with MLaaS, and one primary threat is model extraction attacks. To address this, there are many defense solutions but they suffer from unrealistic assumptions and generalization issues, making them less practical for reliable protection. Driven by these limitations, we introduce a novel defense mechanism, SAME, based on the concept of sample reconstruction. This strategy imposes minimal prerequisites on the defender's capabilities, eliminating the need for auxiliary Out-of-Distribution (OOD) datasets, user query history, white-box model access, and additional intervention during model training. It is compatible with existing active defense methods. Our extensive experiments corroborate the superior efficacy of SAME over state-of-the-art solutions. Our code is available at https://github.com/xythink/SAME.

preprint2023arXiv

GiT: Graph Interactive Transformer for Vehicle Re-identification

Transformers are more and more popular in computer vision, which treat an image as a sequence of patches and learn robust global features from the sequence. However, pure transformers are not entirely suitable for vehicle re-identification because vehicle re-identification requires both robust global features and discriminative local features. For that, a graph interactive transformer (GiT) is proposed in this paper. In the macro view, a list of GiT blocks are stacked to build a vehicle re-identification model, in where graphs are to extract discriminative local features within patches and transformers are to extract robust global features among patches. In the micro view, graphs and transformers are in an interactive status, bringing effective cooperation between local and global features. Specifically, one current graph is embedded after the former level's graph and transformer, while the current transform is embedded after the current graph and the former level's transformer. In addition to the interaction between graphs and transforms, the graph is a newly-designed local correction graph, which learns discriminative local features within a patch by exploring nodes' relationships. Extensive experiments on three large-scale vehicle re-identification datasets demonstrate that our GiT method is superior to state-of-the-art vehicle re-identification approaches.

preprint2022arXiv

CHIP: CHannel Independence-based Pruning for Compact Neural Networks

Filter pruning has been widely used for neural network compression because of its enabled practical acceleration. To date, most of the existing filter pruning works explore the importance of filters via using intra-channel information. In this paper, starting from an inter-channel perspective, we propose to perform efficient filter pruning using Channel Independence, a metric that measures the correlations among different feature maps. The less independent feature map is interpreted as containing less useful information$/$knowledge, and hence its corresponding filter can be pruned without affecting model capacity. We systematically investigate the quantification metric, measuring scheme and sensitiveness$/$reliability of channel independence in the context of filter pruning. Our evaluation results for different models on various datasets show the superior performance of our approach. Notably, on CIFAR-10 dataset our solution can bring $0.90\%$ and $0.94\%$ accuracy increase over baseline ResNet-56 and ResNet-110 models, respectively, and meanwhile the model size and FLOPs are reduced by $42.8\%$ and $47.4\%$ (for ResNet-56) and $48.3\%$ and $52.1\%$ (for ResNet-110), respectively. On ImageNet dataset, our approach can achieve $40.8\%$ and $44.8\%$ storage and computation reductions, respectively, with $0.15\%$ accuracy increase over the baseline ResNet-50 model. The code is available at https://github.com/Eclipsess/CHIP_NeurIPS2021.

preprint2022arXiv

Compute Cost Amortized Transformer for Streaming ASR

We present a streaming, Transformer-based end-to-end automatic speech recognition (ASR) architecture which achieves efficient neural inference through compute cost amortization. Our architecture creates sparse computation pathways dynamically at inference time, resulting in selective use of compute resources throughout decoding, enabling significant reductions in compute with minimal impact on accuracy. The fully differentiable architecture is trained end-to-end with an accompanying lightweight arbitrator mechanism operating at the frame-level to make dynamic decisions on each input while a tunable loss function is used to regularize the overall level of compute against predictive performance. We report empirical results from experiments using the compute amortized Transformer-Transducer (T-T) model conducted on LibriSpeech data. Our best model can achieve a 60% compute cost reduction with only a 3% relative word error rate (WER) increase.

preprint2022arXiv

Instanton homology and knot detection on thickened surfaces

Suppose $Σ$ is a compact oriented surface (possibly with boundary) that has genus zero, and L is a link in the interior of $(-1,1)\timesΣ$. We prove that the Asaeda-Przytycki-Sikora (APS) homology of L has rank 2 if and only if L is isotopic to an embedded knot in $\{0\}\timesΣ$. As a consequence, the APS homology detects the unknot in $(-1,1)\timesΣ$. This is the first detection result for generalized Khovanov homology that is valid on an infinite family of manifolds, and it partially solves a conjecture in arxiv:2005.12863. Our proof is different from the previous detection results obtained by instanton homology because in this case, the second page of Kronheimer-Mrowka's spectral sequence is not isomorphic to the APS homology. We also characterize all links in product manifolds that have minimal sutured instanton homology, which may be of independent interest.

preprint2022arXiv

Interfacial Charge-transfer Excitonic Insulator in a Two-dimensional Organic-inorganic Superlattice

Excitonic insulators are long-sought-after quantum materials predicted to spontaneously open a gap by the Bose condensation of bound electron-hole pairs, namely, excitons, in their ground state. Since the theoretical conjecture, extensive efforts have been devoted to pursuing excitonic insulator platforms for exploring macroscopic quantum phenomena in real materials. Reliable evidences of excitonic character have been obtained in layered chalcogenides as promising candidates. However, owing to the interference of intrinsic lattice instabilities, it is still debatable whether those features, such as charge density wave and gap opening, are primarily driven by the excitonic effect or by the lattice transition. Herein, we develop a novel charge-transfer excitonic insulator in organic-inorganic superlattice interfaces, which serves as an ideal platform to decouple the excitonic effect from the lattice effect. In this system, we observe the narrow gap opening and the formation of a charge density wave without periodic lattice distortion, providing visualized evidence of exciton condensation occurring in thermal equilibrium. Our findings identify spontaneous interfacial charge transfer as a new strategy for developing novel excitonic insulators and investigating their correlated many-body physics.

preprint2022arXiv

Investigation of the Effect of Quantum Measurement on Parity-Time Symmetry

Symmetry, including the parity-time ($\mathcal{PT}$)-symmetry, is a striking topic, widely discussed and employed in many fields. It is well-known that quantum measurement can destroy or disturb quantum systems. However, can and how does quantum measurement destroy the symmetry of the measured system? To answer the pertinent question, we establish the correlation between the quantum measurement and Floquet $\mathcal{PT}$-symmetry and investigate for the first time how the measurement frequency and measurement strength affect the $\mathcal{PT}$-symmetry of the measured system using the $^{40}\mathrm{Ca}^{+}$ ion. It is already shown that the measurement at high frequencies would break the $\mathcal{PT}$ symmetry. Notably, even for an inadequately fast measurement frequency, if the measurement strength is sufficiently strong, the $\mathcal{PT}$ symmetry breaking can occur. The current work can enhance our knowledge of quantum measurement and symmetry and may inspire further research on the effect of quantum measurement on symmetry.

preprint2022arXiv

Revisiting the Evidence for an Intermediate-mass Black Hole in the Center of NGC 6624 with Simulations

The acceleration of LMXB 4U 1820-30 that derived from its orbital-period derivative $\dot P_{\rm b}$ was supposed to be the evidence for an Intermediate-mass Black Hole (IMBH) in the Galactic globular cluster (GC) NGC 6624. However, we find that the anomalous $\dot P_{\rm b}$ is mainly due to the gravitational wave emission, rather than the acceleration in cluster potential. Using the standard structure models of GCs, we simulate acceleration distributions for pulsars in the central region of the cluster. By fitting the acceleration of J1823-3021A with the simulated distribution profiles (maximum values), it is suggested that an IMBH with mass $M\gtrsim 950^{+550}_{-350}~M_{\odot}$ may reside in the cluster center. We further show that the second period derivative $\ddot P$ of J1823-3021A is probably due to the gravitational perturbation of a nearby star.

preprint2022arXiv

RIBAC: Towards Robust and Imperceptible Backdoor Attack against Compact DNN

Recently backdoor attack has become an emerging threat to the security of deep neural network (DNN) models. To date, most of the existing studies focus on backdoor attack against the uncompressed model; while the vulnerability of compressed DNNs, which are widely used in the practical applications, is little exploited yet. In this paper, we propose to study and develop Robust and Imperceptible Backdoor Attack against Compact DNN models (RIBAC). By performing systematic analysis and exploration on the important design knobs, we propose a framework that can learn the proper trigger patterns, model parameters and pruning masks in an efficient way. Thereby achieving high trigger stealthiness, high attack success rate and high model efficiency simultaneously. Extensive evaluations across different datasets, including the test against the state-of-the-art defense mechanisms, demonstrate the high robustness, stealthiness and model efficiency of RIBAC. Code is available at https://github.com/huyvnphan/ECCV2022-RIBAC

preprint2022arXiv

Self-Testing of a Single Quantum System: Theory and Experiment

Certifying individual quantum devices with minimal assumptions is crucial for the development of quantum technologies. Here, we investigate how to leverage single-system contextuality to realize self-testing. We develop a robust self-testing protocol based on the simplest contextuality witness for the simplest contextual quantum system, the Klyachko-Can-Binicioğlu-Shumovsky (KCBS) inequality for the qutrit. We establish a lower bound on the fidelity of the state and the measurements (to an ideal configuration) as a function of the value of the witness under a pragmatic assumption on the measurements we call the KCBS orthogonality condition. We apply the method in an experiment with randomly chosen measurements on a single trapped $^{40}{\rm Ca}^+$ and near-perfect detection efficiency. The observed statistics allow us to self-test the system and provide the first experimental demonstration of quantum self-testing of a single system. Further, we quantify and report that deviations from our assumptions are minimal, an aspect previously overlooked by contextuality experiments.

preprint2021arXiv

Enabling Fast and Universal Audio Adversarial Attack Using Generative Model

Recently, the vulnerability of DNN-based audio systems to adversarial attacks has obtained the increasing attention. However, the existing audio adversarial attacks allow the adversary to possess the entire user's audio input as well as granting sufficient time budget to generate the adversarial perturbations. These idealized assumptions, however, makes the existing audio adversarial attacks mostly impossible to be launched in a timely fashion in practice (e.g., playing unnoticeable adversarial perturbations along with user's streaming input). To overcome these limitations, in this paper we propose fast audio adversarial perturbation generator (FAPG), which uses generative model to generate adversarial perturbations for the audio input in a single forward pass, thereby drastically improving the perturbation generation speed. Built on the top of FAPG, we further propose universal audio adversarial perturbation generator (UAPG), a scheme crafting universal adversarial perturbation that can be imposed on arbitrary benign audio input to cause misclassification. Extensive experiments show that our proposed FAPG can achieve up to 167X speedup over the state-of-the-art audio adversarial attack methods. Also our proposed UAPG can generate universal adversarial perturbation that achieves much better attack performance than the state-of-the-art solutions.

preprint2021arXiv

Probing Dark Contents in Globular Clusters With Timing Effects of Pulsar Acceleration

We investigate pulsar timing residuals due to the coupling effect of the pulsar transverse acceleration and the R$\rm{\ddot{o}}$mer delay. The effect is relatively small and usually negligible. Only for pulsars in globular clusters, it is possibly important. The maximum residual amplitude, which is from the pulsar near the surface of the core of the cluster, is about tens nanoseconds, and may hardly be identified for most of the globular clusters currently. However, an intermediate-mass black hole in the centre of a cluster can increase apparently the timing residual magnitudes. Particularly for the pulsars in the innermost core region, their residual magnitudes may be significant. The high-magnitude residuals, which above critical lines of each cluster, are strong evidences for the presence of a black hole or dark remnants of comparable total mass in the centre of the cluster. We also explored the timing effects of line-of-sight accelerations for the pulsars. The distribution of measured line-of-sight accelerations are simulated with a Monte Carlo method. A two-dimensional Kolmogorov-Smirnov tests are performed to reexamine the consistency of distributions of the simulated and reported data for various values of parameters of the clusters. It is shown that the structure parameters of Terzan 5 can be well constrained by comparing the distribution of measured line-of-sight accelerations with the distributions from Monte Carlo simulations. We provide that the cluster has an upper limit on the central black hole/dark remnant mass of $ \sim 6000 M_{\odot} $.

preprint2020arXiv

Estimation of the Laser Frequency Nosie Spectrum by Continuous Dynamical Decoupling

Decoherence induced by the laser frequency noise is one of the most important obstacles in the quantum information processing. In order to suppress this decoherence, the noise power spectral density needs to be accurately characterized. In particular, the noise spectrum measurement based on the coherence characteristics of qubits would be a meaningful and still challenging method. Here, we theoretically analyze and experimentally obtain the spectrum of laser frequency noise based on the continuous dynamical decoupling technique. We first estimate the mixture-noise (including laser and magnetic noises) spectrum up to $(2π)$530 kHz by monitoring the transverse relaxation from an initial state $+X$, followed by a gradient descent data process protocol. Then the contribution from the laser noise is extracted by enconding the qubits on different Zeeman sublevels. We also investigate two sufficiently strong noise components by making an analogy between these noises and driving lasers whose linewidth assumed to be negligible. This method is verified experimentally and finally helps to characterize the noise.

preprint2020arXiv

ISTD-GCN: Iterative Spatial-Temporal Diffusion Graph Convolutional Network for Traffic Speed Forecasting

Most of the existing algorithms for traffic speed forecasting split spatial features and temporal features to independent modules, and then associate information from both dimensions. However, features from spatial and temporal dimensions influence mutually, separated extractions isolate such dependencies, and might lead to inaccurate results. In this paper, we incorporate the perspective of information diffusion to model spatial features and temporal features synchronously. Intuitively, vertices not only diffuse information to the neighborhood but also to the subsequent state along with the temporal dimension. Therefore, we can model such heterogeneous spatial-temporal structures as a homogeneous process of diffusion. On this basis, we propose an Iterative Spatial-Temporal Diffusion Graph Convolutional Network (ISTD-GCN) to extract spatial and temporal features synchronously, thus dependencies between both dimensions can be better modeled. Experiments on two traffic datasets illustrate that our ISTD-GCN outperforms 10 baselines in traffic speed forecasting tasks. The source code is available at https://github.com/Anonymous.

preprint2020arXiv

PERMDNN: Efficient Compressed DNN Architecture with Permuted Diagonal Matrices

Deep neural network (DNN) has emerged as the most important and popular artificial intelligent (AI) technique. The growth of model size poses a key energy efficiency challenge for the underlying computing platform. Thus, model compression becomes a crucial problem. However, the current approaches are limited by various drawbacks. Specifically, network sparsification approach suffers from irregularity, heuristic nature and large indexing overhead. On the other hand, the recent structured matrix-based approach (i.e., CirCNN) is limited by the relatively complex arithmetic computation (i.e., FFT), less flexible compression ratio, and its inability to fully utilize input sparsity. To address these drawbacks, this paper proposes PermDNN, a novel approach to generate and execute hardware-friendly structured sparse DNN models using permuted diagonal matrices. Compared with unstructured sparsification approach, PermDNN eliminates the drawbacks of indexing overhead, non-heuristic compression effects and time-consuming retraining. Compared with circulant structure-imposing approach, PermDNN enjoys the benefits of higher reduction in computational complexity, flexible compression ratio, simple arithmetic computation and full utilization of input sparsity. We propose PermDNN architecture, a multi-processing element (PE) fully-connected (FC) layer-targeted computing engine. The entire architecture is highly scalable and flexible, and hence it can support the needs of different applications with different model configurations. We implement a 32-PE design using CMOS 28nm technology. Compared with EIE, PermDNN achieves 3.3x~4.8x higher throughout, 5.9x~8.5x better area efficiency and 2.8x~4.0x better energy efficiency on different workloads. Compared with CirCNN, PermDNN achieves 11.51x higher throughput and 3.89x better energy efficiency.

preprint2020arXiv

Real-time, Universal, and Robust Adversarial Attacks Against Speaker Recognition Systems

As the popularity of voice user interface (VUI) exploded in recent years, speaker recognition system has emerged as an important medium of identifying a speaker in many security-required applications and services. In this paper, we propose the first real-time, universal, and robust adversarial attack against the state-of-the-art deep neural network (DNN) based speaker recognition system. Through adding an audio-agnostic universal perturbation on arbitrary enrolled speaker's voice input, the DNN-based speaker recognition system would identify the speaker as any target (i.e., adversary-desired) speaker label. In addition, we improve the robustness of our attack by modeling the sound distortions caused by the physical over-the-air propagation through estimating room impulse response (RIR). Experiment using a public dataset of 109 English speakers demonstrates the effectiveness and robustness of our proposed attack with a high attack success rate of over 90%. The attack launching time also achieves a 100X speedup over contemporary non-universal attacks.

preprint2020arXiv

The Timing Residual Patterns Due to Pulsar Acceleration

The form of timing residuals due to errors in pulsar spin period $P$ and its derivative $\dot{P}$, in positions, as well as in proper motions, have been well presented for decades in the literature. However, the residual patterns due to errors in the pulsar acceleration have not been reported previously, while a pulsar in the galaxy or a globular cluster (GC) will be unavoidably accelerated. The coupling effect of the pulsar transverse acceleration and the R$\rm{\ddot{o}}$mer delay on timing residuals are simulated in this work. The results show that the residual due to the effect can be identified by the oscillation envelopes of the residuals. It is also shown that the amplitude of the residual due to the effect is usually relatively small, however, it may probably be observable for pulsars distributing in the vicinity of the core of a nearby GC.

preprint2018arXiv

Sutured Manifolds and Polynomial Invariants from Higher Rank Bundles

For each integer $N\geq 2$, Mariño and Moore defined generalized Donaldson invariants by the methods of quantum field theory, and made predictions about the values of these invariants. Subsequently, Kronheimer gave a rigorous definition of generalized Donaldson invariants using the moduli spaces of anti-self-dual connections on hermitian vector bundles of rank $N$. In this paper, Mariño and Moore's predictions are confirmed for simply connected elliptic surfaces without multiple fibers and certain surfaces of general type in the case that $N=3$. The primary motivation is to study 3-manifold instanton Floer homologies which are defined by higher rank bundles. In particular, the computation of the generalized Donaldson invariants are exploited to define a Floer homology theory for sutured 3-manifolds.