arXiv subjects

Zhili Chen

Publications and source records attributed to Zhili Chen.

At least 37 records · Page 2Linked to original sources

arXiv2024

On the complexity of isomorphism problems for tensors, groups, and polynomials III: actions by classical groups

We study the complexity of isomorphism problems for d-way arrays, or tensors, under natural actions by classical groups such as orthogonal, unitary, and symplectic groups. Such problems arise naturally in statistical data analysis and quantum information. We study two types of complexity-theoretic questions. First, for a fixed action type (isomorphism, conjugacy, etc.), we relate the complexity of the isomorphism problem over a classical group to that over the general linear group. Second, for a fixed group type (orthogonal, unitary, or symplectic), we compare the complexity of the decision problems for different actions. Our main results are as follows. First, for orthogonal and symplectic groups acting on 3-way arrays, the isomorphism problems reduce to the corresponding problem over the general linear group. Second, for orthogonal and unitary groups, the isomorphism problems of five natural actions on 3-way arrays are polynomial-time equivalent, and the d-tensor isomorphism problem reduces to the 3-tensor isomorphism problem for any fixed d>3. For unitary groups, the preceding result implies that LOCC classification of tripartite quantum states is at least as difficult as LOCC classification of d-partite quantum states for any d. Lastly, we also show that the graph isomorphism problem reduces to the tensor isomorphism problem over orthogonal and unitary groups.

cs.CC↗

arXiv2024

PPAD: Iterative Interactions of Prediction and Planning for End-to-end Autonomous Driving

We present a new interaction mechanism of prediction and planning for end-to-end autonomous driving, called PPAD (Iterative Interaction of Prediction and Planning Autonomous Driving), which considers the timestep-wise interaction to better integrate prediction and planning. An ego vehicle performs motion planning at each timestep based on the trajectory prediction of surrounding agents (e.g., vehicles and pedestrians) and its local road conditions. Unlike existing end-to-end autonomous driving frameworks, PPAD models the interactions among ego, agents, and the dynamic environment in an autoregressive manner by interleaving the Prediction and Planning processes at every timestep, instead of a single sequential process of prediction followed by planning. Specifically, we design ego-to-agent, ego-to-map, and ego-to-BEV interaction mechanisms with hierarchical dynamic key objects attention to better model the interactions. The experiments on the nuScenes benchmark show that our approach outperforms state-of-the-art methods.

cs.CV↗

arXiv2024

Learning High-resolution Vector Representation from Multi-Camera Images for 3D Object Detection

The Bird's-Eye-View (BEV) representation is a critical factor that directly impacts the 3D object detection performance, but the traditional BEV grid representation induces quadratic computational cost as the spatial resolution grows. To address this limitation, we present a new camera-based 3D object detector with high-resolution vector representation: VectorFormer. The presented high-resolution vector representation is combined with the lower-resolution BEV representation to efficiently exploit 3D geometry from multi-camera images at a high resolution through our two novel modules: vector scattering and gathering. To this end, the learned vector representation with richer scene contexts can serve as the decoding query for final predictions. We conduct extensive experiments on the nuScenes dataset and demonstrate state-of-the-art performance in NDS and inference time. Furthermore, we investigate query-BEV-based methods incorporated with our proposed vector representation and observe a consistent performance improvement.

cs.CV↗

arXiv2024

ALI-DPFL: Differentially Private Federated Learning with Adaptive Local Iterations

Federated Learning (FL) is a distributed machine learning technique that allows model training among multiple devices or organizations by sharing training parameters instead of raw data. However, adversaries can still infer individual information through inference attacks (e.g. differential attacks) on these training parameters. As a result, Differential Privacy (DP) has been widely used in FL to prevent such attacks. We consider differentially private federated learning in a resource-constrained scenario, where both privacy budget and communication rounds are constrained. By theoretically analyzing the convergence, we can find the optimal number of local DPSGD iterations for clients between any two sequential global updates. Based on this, we design an algorithm of Differentially Private Federated Learning with Adaptive Local Iterations (ALI-DPFL). We experiment our algorithm on the MNIST, FashionMNIST and Cifar10 datasets, and demonstrate significantly better performances than previous work in the resource-constraint scenario. Code is available at https://github.com/cheng-t/ALI-DPFL.

cs.LG↗

arXiv2024

DP-DCAN: Differentially Private Deep Contrastive Autoencoder Network for Single-cell Clustering

Single-cell RNA sequencing (scRNA-seq) is important to transcriptomic analysis of gene expression. Recently, deep learning has facilitated the analysis of high-dimensional single-cell data. Unfortunately, deep learning models may leak sensitive information about users. As a result, Differential Privacy (DP) is increasingly used to protect privacy. However, existing DP methods usually perturb whole neural networks to achieve differential privacy, and hence result in great performance overheads. To address this challenge, in this paper, we take advantage of the uniqueness of the autoencoder that it outputs only the dimension-reduced vector in the middle of the network, and design a Differentially Private Deep Contrastive Autoencoder Network (DP-DCAN) by partial network perturbation for single-cell clustering. Since only partial network is added with noise, the performance improvement is obvious and twofold: one part of network is trained with less noise due to a bigger privacy budget, and the other part is trained without any noise. Experimental results of six datasets have verified that DP-DCAN is superior to the traditional DP scheme with whole network perturbation. Moreover, DP-DCAN demonstrates strong robustness to adversarial attacks.

cs.LG↗

arXiv2024

Cross-Cluster Shifting for Efficient and Effective 3D Object Detection in Autonomous Driving

We present a new 3D point-based detector model, named Shift-SSD, for precise 3D object detection in autonomous driving. Traditional point-based 3D object detectors often employ architectures that rely on a progressive downsampling of points. While this method effectively reduces computational demands and increases receptive fields, it will compromise the preservation of crucial non-local information for accurate 3D object detection, especially in the complex driving scenarios. To address this, we introduce an intriguing Cross-Cluster Shifting operation to unleash the representation capacity of the point-based detector by efficiently modeling longer-range inter-dependency while including only a negligible overhead. Concretely, the Cross-Cluster Shifting operation enhances the conventional design by shifting partial channels from neighboring clusters, which enables richer interaction with non-local regions and thus enlarges the receptive field of clusters. We conduct extensive experiments on the KITTI, Waymo, and nuScenes datasets, and the results demonstrate the state-of-the-art performance of Shift-SSD in both detection accuracy and runtime efficiency.

cs.CV↗

arXiv2024

Multi-Party Private Set Intersection: A Circuit-Based Protocol with Jaccard Similarity for Secure and Efficient Anomaly Detection in Network Traffic

We present a new circuit-based protocol for multi-party private set intersection (PSI) that allows m parties to compute the intersection of their datasets without revealing any additional information about the items outside the intersection. Building upon the two-party Sort-Compare-Shuffle (SCS) protocol, we seamlessly extend it to a multi-party setting. Demonstrating its practicality through implementation, our protocol exhibits acceptable performance. Specifically, with 7 parties, each possessing a set size of 2^{12}, our protocol completes in just 19 seconds. Moreover, circuit-based protocols like ours have an advantage over using custom protocols to perform more complex computation. We substantiate this advantage by incorporating a module for calculating the Jaccard similarity metric of the private sets which can be used in the application domain of network traffic analysis for anomaly detection. This extension showcases the versatility of our protocol beyond set intersection computations, demonstrating its efficacy in preserving privacy while efficiently identifying abnormal patterns in network flow.

cs.CR↗

arXiv2023

DPSUR: Accelerating Differentially Private Stochastic Gradient Descent Using Selective Update and Release

Machine learning models are known to memorize private data to reduce their training loss, which can be inadvertently exploited by privacy attacks such as model inversion and membership inference. To protect against these attacks, differential privacy (DP) has become the de facto standard for privacy-preserving machine learning, particularly those popular training algorithms using stochastic gradient descent, such as DPSGD. Nonetheless, DPSGD still suffers from severe utility loss due to its slow convergence. This is partially caused by the random sampling, which brings bias and variance to the gradient, and partially by the Gaussian noise, which leads to fluctuation of gradient updates. Our key idea to address these issues is to apply selective updates to the model training, while discarding those useless or even harmful updates. Motivated by this, this paper proposes DPSUR, a Differentially Private training framework based on Selective Updates and Release, where the gradient from each iteration is evaluated based on a validation test, and only those updates leading to convergence are applied to the model. As such, DPSUR ensures the training in the right direction and thus can achieve faster convergence than DPSGD. The main challenges lie in two aspects -- privacy concerns arising from gradient evaluation, and gradient selection strategy for model update. To address the challenges, DPSUR introduces a clipping strategy for update randomization and a threshold mechanism for gradient selection. Experiments conducted on MNIST, FMNIST, CIFAR-10, and IMDB datasets show that DPSUR significantly outperforms previous works in terms of convergence speed and model utility.

cs.LG↗

arXiv2023

Truncated Laplace and Gaussian mechanisms of RDP

The Laplace mechanism and the Gaussian mechanism are primary mechanisms in differential privacy, widely applicable to many scenarios involving numerical data. However, due to the infinite-range random variables they generate, the Laplace and Gaussian mechanisms may return values that are semantically impossible, such as negative numbers. To address this issue, we have designed the truncated Laplace mechanism and Gaussian mechanism. For a given truncation interval [a, b], the truncated Gaussian mechanism ensures the same Renyi Differential Privacy (RDP) as the untruncated mechanism, regardless of the values chosen for the truncation interval [a, b]. Similarly, the truncated Laplace mechanism, for specified interval [a, b], maintains the same RDP as the untruncated mechanism. We provide the RDP expressions for each of them. We believe that our study can further enhance the utility of differential privacy in specific applications.

cs.CR↗

arXiv2023

Secure and Scalable Circuit-based Protocol for Multi-Party Private Set Intersection

We propose a novel protocol for computing a circuit which implements the multi-party private set intersection functionality (PSI). Circuit-based approach has advantages over using custom protocols to achieve this task, since many applications of PSI do not require the computation of the intersection itself, but rather specific functional computations over the items in the intersection. Our protocol represents the pioneering circuit-based multi-party PSI protocol, which builds upon and optimizes the two-party SCS \cite{huang2012private} protocol. By using secure computation between two parties, our protocol sidesteps the complexities associated with multi-party interactions and demonstrates good scalability. In order to mitigate the high overhead associated with circuit-based constructions, we have further enhanced our protocol by utilizing simple hashing scheme and permutation-based hash functions. These tricks have enabled us to minimize circuit size by employing bucketing techniques while simultaneously attaining noteworthy reductions in both computation and communication expenses.

cs.CR↗

arXiv2022

SA-DPSGD: Differentially Private Stochastic Gradient Descent based on Simulated Annealing

Differential privacy (DP) provides a formal privacy guarantee that prevents adversaries with access to machine learning models from extracting information about individual training points. Differentially private stochastic gradient descent (DPSGD) is the most popular training method with differential privacy in image recognition. However, existing DPSGD schemes lead to significant performance degradation, which prevents the application of differential privacy. In this paper, we propose a simulated annealing-based differentially private stochastic gradient descent scheme (SA-DPSGD) which accepts a candidate update with a probability that depends both on the update quality and on the number of iterations. Through this random update screening, we make the differentially private gradient descent proceed in the right direction in each iteration, and result in a more accurate model finally. In our experiments, under the same hyperparameters, our scheme achieves test accuracies 98.35%, 87.41% and 60.92% on datasets MNIST, FashionMNIST and CIFAR10, respectively, compared to the state-of-the-art result of 98.12%, 86.33% and 59.34%. Under the freely adjusted hyperparameters, our scheme achieves even higher accuracies, 98.89%, 88.50% and 64.17%. We believe that our method has a great contribution for closing the accuracy gap between private and non-private image classification.

cs.CR↗

arXiv2022

Adap DP-FL: Differentially Private Federated Learning with Adaptive Noise

Federated learning seeks to address the issue of isolated data islands by making clients disclose only their local training models. However, it was demonstrated that private information could still be inferred by analyzing local model parameters, such as deep neural network model weights. Recently, differential privacy has been applied to federated learning to protect data privacy, but the noise added may degrade the learning performance much. Typically, in previous work, training parameters were clipped equally and noises were added uniformly. The heterogeneity and convergence of training parameters were simply not considered. In this paper, we propose a differentially private scheme for federated learning with adaptive noise (Adap DP-FL). Specifically, due to the gradient heterogeneity, we conduct adaptive gradient clipping for different clients and different rounds; due to the gradient convergence, we add decreasing noises accordingly. Extensive experiments on real-world datasets demonstrate that our Adap DP-FL outperforms previous methods significantly.

cs.LG↗

arXiv2022

Geo-MOEA: A Multi-Objective Evolutionary Algorithm with Geo-obfuscation for Mobile Crowdsourcing Workers

The rapid development of mobile Internet and sharing economy brings the prosperity of Spatial Crowdsourcing (SC). SC applications assign various tasks according to reported location information of task's requesters and outsourced workers (such as DiDi, MeiTuan and Uber). However, SC-servers are often untrustworthy and the exposure of users' locations raises privacy concerns. In this paper, we design a framework called Geo-MOEA (Multi-Objective Evolutionary Algorithm with Geo-obfuscation) to protect location privacy of workers involved on SC platform in mobile networks environment. We propose an adaptive regionalized obfuscation approach with inference error bounds based on geo-indistinguishability (a strong notion of differential privacy), which is suitable for the context of large-scale location data and task allocations. This enables each worker to report a pseudo-location that is adaptively generated with a personalized inference error threshold. Moreover, as a popular computational intelligence method, MOEA is introduced to optimize the trade-off between SC service availability and privacy protection while ensuring theoretically the most general condition on protection location sets for larger search space. Finally, the experimental results on two public datasets show that our Geo-MOEA approach achieves up to 20% reduction in service quality loss while guaranteeing differential and geo-distortion location privacy.

cs.CR↗

arXiv2022

DPIVE: A Regionalized Location Obfuscation Scheme with Personalized Privacy Levels

The popularity of cyber-physical systems is fueling the rapid growth of location-based services. This poses the risk of location privacy disclosure. Effective privacy preservation is foremost for various mobile applications. Recently, geo-indistinguishability and expected inference error are proposed for limiting location leakages. In this paper, we argue that personalization means regionalization for geo-indistinguishability, and we propose a regionalized location obfuscation mechanism called DPIVE with personalized utility sensitivities. This substantially corrects the differential and distortion privacy problem of PIVE framework proposed by Yu et al. on NDSS 2017. We develop DPIVE with two phases. In Phase I, we determine disjoint sets by partitioning all possible positions such that different locations in the same set share the Protection Location Set (PLS). In Phase II, we construct a probability distribution matrix in which the rows corresponding to the same PLS have their own sensitivity of utility (PLS diameter). Moreover, by designing QK-means algorithm for more search space in 2-D space, we improve DPIVE with refined location partition and present fine-grained personalization, enabling each location to have its own privacy level endowed with a customized privacy budget. Experiments with two public datasets demonstrate that our mechanisms have the superior performance, typically on skewed locations.

cs.CR↗

arXiv2022

LDP-FPMiner: FP-Tree Based Frequent Itemset Mining with Local Differential Privacy

Data aggregation in the setting of local differential privacy (LDP) guarantees strong privacy by providing plausible deniability of sensitive data. Existing works on this issue mostly focused on discovering heavy hitters, leaving the task of frequent itemset mining (FIM) as an open problem. To the best of our knowledge, the-state-of-the-art LDP solution to FIM is the SVSM protocol proposed recently. The SVSM protocol is mainly based on the padding and sampling based frequency oracle (PSFO) protocol, and regarded an itemset as an independent item without considering the frequency consistency among itemsets. In this paper, we propose a novel LDP approach to FIM called LDP-FPMiner based on frequent pattern tree (FP-tree). Our proposal exploits frequency consistency among itemsets by constructing and optimizing a noisy FP-tree with LDP. Specifically, it works as follows. First, the most frequent items are identified, and the item domain is cut down accordingly. Second, the maximum level of the FP-tree is estimated. Third, a noisy FP-tree is constructed and optimized by using itemset frequency consistency, and then mined to obtain the k most frequent itemsets. Experimental results show that the LDP-FPMiner significantly improves over the state-of-the-art approach, SVSM, especially in the case of a high privacy level.

cs.DB↗

arXiv2022

DCDChain: A Credible Architecture of Digital Copyright Detection Based on Blockchain

Copyright detection is an effective method to prevent piracy. However, untrustworthy detection parties may lead to falsified detection results. Due to its credibility and tamper resistance, blockchain has been applied to copyright protection. Previous works mainly utilized blockchain for reliable copyright information storage or copyrighted digital media trading. As far as we know, the problem of credible copyright detection has not been addressed. In this paper, we propose a credible copyright detection architecture based on the blockchain, called DCDChain. In this architecture, the detection agency first detects copyrights off the chain, then uploads the detection records to the blockchain. Since data on the blockchain are publicly accessible, media providers can verify the correctness of the copyright detection, and appeal to a smart contract if there is any dissent. The smart contract then arbitrates the disputes by verifying the correctness of detection on the chain. The detect-verify-and-arbitrate mechanism guarantees the credibility of copyright detection. Security analysis and experimental simulations show that the digital copyright detection architecture is credible, secure and efficient. The proposed credible copyright detection scheme is highly important for copyright protection. The future work is to improve the scheme by designing more effective locality sensitive hash algorithms for various digital media.

cs.CR↗

arXiv2022

Efficient reversible data hiding via two layers of double-peak embedding

Reversible data hiding continues to attract significant attention in recent years. In particular, an increasing number of authors focus on the higher significant bit (HSB) plane of an image which can yield more redundant space. On the other hand, the lower significant bit planes are often ignored for embedding in existing schemes due to their harm to the embedding rate. This paper proposes an efficient reversible data hiding scheme via a double-peak two-layer embedding (DTLE) strategy with prediction error expansion. The higher six-bit planes of the image are assigned as the HSB plane, and double prediction error peaks are applied in either embedding layer. This makes fuller use of the redundancy space of images compared with the one error peak strategy. Moreover, we carry out the median-edge detector pre-processing for complex images to reduce the size of the auxiliary information. A series of experimental results show that our DTLE approach achieves up to 83% higher embedding rate on real-world datasets while guaranteeing better image quality.

cs.CR↗

arXiv2022

Point Cloud Compression with Sibling Context and Surface Priors

We present a novel octree-based multi-level framework for large-scale point cloud compression, which can organize sparse and unstructured point clouds in a memory-efficient way. In this framework, we propose a new entropy model that explores the hierarchical dependency in an octree using the context of siblings' children, ancestors, and neighbors to encode the occupancy information of each non-leaf octree node into a bitstream. Moreover, we locally fit quadratic surfaces with a voxel-based geometry-aware module to provide geometric priors in entropy encoding. These strong priors empower our entropy framework to encode the octree into a more compact bitstream. In the decoding stage, we apply a two-step heuristic strategy to restore point clouds with better reconstruction quality. The quantitative evaluation shows that our method outperforms state-of-the-art baselines with a bitrate improvement of 11-16% and 12-14% on the KITTI Odometry and nuScenes datasets, respectively.

cs.CV↗