arXiv subjects

Thomas Humphries

Publications and source records attributed to Thomas Humphries.

12 recordsLinked to original sources

arXiv2026

PE-means: Improved Differentially Private $k$-means Clustering through Private Evolution

We study the problem of differentially private (DP) $k$-means clustering in Euclidean space. Previous solutions rely on summing the private data directly, which induces a sensitivity proportional to the domain. We introduce PE-means, an extension of the private evolution (PE) algorithm (an increasingly popular method for synthetic data generation), to the problem of $k$-means clustering. The key advantage of PE is that it only computes a private histogram with constant sensitivity to guide the evolution. Our adaptation of PE includes new evolutionary operators for clustering, as well as other algorithmic improvements of independent interest. Overall, PE-means achieves an average improvement of 26% in clustering loss over state-of-the-art baselines such as Google's LSH-based algorithm and DP-Lloyd variants.

cs.LG↗

arXiv2026

Interpreting the Error of Differentially Private Median Queries through Randomization Intervals

It can be difficult for practitioners to interpret the quality of differentially private (DP) statistics due to the added noise. One method to help analysts understand the amount of error introduced by DP is to return a Randomization Interval (RI), along with the statistic. A RI is a type of confidence interval that bounds the error introduced by DP. For queries where the noise distribution depends on the input, such as the median, prior work degrades the quality of the median itself to obtain a high-quality RI. In this work, we propose PostRI, a solution to compute a RI after the median has been estimated. PostRI enables a median estimation with 14%-850% higher utility than related work, while maintaining a narrow RI.

cs.CR↗

arXiv2024

Plug-and-play superiorization

The superiorization methodology (SM) is an optimization heuristic in which an iterative algorithm, which aims to solve a particular problem, is ``superiorized'' to promote solutions that are improved with respect to some secondary criterion. This superiorization is achieved by perturbing iterates of the algorithm in nonascending directions of a prescribed function that penalizes undesirable characteristics in the solution; the solution produced by the superiorized algorithm should therefore be improved with respect to the value of this function. In this paper, we broaden the SM to allow for the perturbations to be introduced by an arbitrary procedure instead, using a plug-and-play approach. This allows for operations such as image denoisers or deep neural networks, which have applications to a broad class of problems, to be incorporated within the superiorization methodology. As proof of concept, we perform numerical simulations involving low-dose and sparse-view computed tomography image reconstruction, comparing the plug-and-play approach to two conventionally superiorized algorithms, as well as a post-processing approach. The plug-and-play approach provides comparable or better image quality in most cases, while also providing advantages in terms of computing time, and data fidelity of the solutions.

math.OC↗

arXiv2024

FastLloyd: Federated, Accurate, Secure, and Tunable $k$-Means Clustering with Differential Privacy

We study the problem of privacy-preserving $k$-means clustering in the horizontally federated setting. Existing federated approaches using secure computation suffer from substantial overheads and do not offer output privacy. At the same time, differentially private (DP) $k$-means algorithms either assume a trusted central curator or significantly degrade utility by adding noise in the local DP model. Naively combining the secure and central DP solutions results in a protocol with impractical overhead. Instead, our work provides enhancements to both the DP and secure computation components, resulting in a design that is faster, more private, and more accurate than previous work. By utilizing the computational DP model, we design a lightweight, secure aggregation-based approach that achieves five orders of magnitude speed-up over state-of-the-art related work. Furthermore, we not only maintain the utility of the state-of-the-art in the central model of DP, but we improve the utility further by designing a new DP clustering mechanism.

cs.CR↗

arXiv2024

Fast and Private Inference of Deep Neural Networks by Co-designing Activation Functions

Machine Learning as a Service (MLaaS) is an increasingly popular design where a company with abundant computing resources trains a deep neural network and offers query access for tasks like image classification. The challenge with this design is that MLaaS requires the client to reveal their potentially sensitive queries to the company hosting the model. Multi-party computation (MPC) protects the client's data by allowing encrypted inferences. However, current approaches suffer from prohibitively large inference times. The inference time bottleneck in MPC is the evaluation of non-linear layers such as ReLU activation functions. Motivated by the success of previous work co-designing machine learning and MPC, we develop an activation function co-design. We replace all ReLUs with a polynomial approximation and evaluate them with single-round MPC protocols, which give state-of-the-art inference times in wide-area networks. Furthermore, to address the accuracy issues previously encountered with polynomial activations, we propose a novel training algorithm that gives accuracy competitive with plaintext models. Our evaluation shows between $3$ and $110\times$ speedups in inference time on large models with up to $23$ million parameters while maintaining competitive inference accuracy.

cs.CR↗

arXiv2023

PEPSI: Practically Efficient Private Set Intersection in the Unbalanced Setting

Two parties with private data sets can find shared elements using a Private Set Intersection (PSI) protocol without revealing any information beyond the intersection. Circuit PSI protocols privately compute an arbitrary function of the intersection - such as its cardinality, and are often employed in an unbalanced setting where one party has more data than the other. Existing protocols are either computationally inefficient or require extensive server-client communication on the order of the larger set. We introduce Practically Efficient PSI or PEPSI, a non-interactive solution where only the client sends its encrypted data. PEPSI can process an intersection of 1024 client items with a million server items in under a second, using less than 5 MB of communication. Our work is over 4 orders of magnitude faster than an existing non-interactive circuit PSI protocol and requires only 10% of the communication. It is also up to 20 times faster than the work of Ion et al., which computes a limited set of functions and has communication costs proportional to the larger set. Our work is the first to demonstrate that non-interactive circuit PSI can be practically applied in an unbalanced setting.

cs.CR↗

arXiv2023

A New Spectral Conjugate Subgradient Method with Application in Computed Tomography Image Reconstruction

A new spectral conjugate subgradient method is presented to solve nonsmooth unconstrained optimization problems. The method combines the spectral conjugate gradient method for smooth problems with the spectral subgradient method for nonsmooth problems. We study the effect of two different choices of line search, as well as three formulas for determining the conjugate directions. In addition to numerical experiments with standard nonsmooth test problems, we also apply the method to several image reconstruction problems in computed tomography, using total variation regularization. Performance profiles are used to compare the performance of the algorithm using different line search strategies and conjugate directions to that of the original spectral subgradient method. Our results show that the spectral conjugate subgradient algorithm outperforms the original spectral subgradient method, and that the use of the Polak-Ribiere formula for conjugate directions provides the best and most robust performance.

math.OC↗

arXiv2023

Investigating Membership Inference Attacks under Data Dependencies

Training machine learning models on privacy-sensitive data has become a popular practice, driving innovation in ever-expanding fields. This has opened the door to new attacks that can have serious privacy implications. One such attack, the Membership Inference Attack (MIA), exposes whether or not a particular data point was used to train a model. A growing body of literature uses Differentially Private (DP) training algorithms as a defence against such attacks. However, these works evaluate the defence under the restrictive assumption that all members of the training set, as well as non-members, are independent and identically distributed. This assumption does not hold for many real-world use cases in the literature. Motivated by this, we evaluate membership inference with statistical dependencies among samples and explain why DP does not provide meaningful protection (the privacy parameter $ε$ scales with the training set size $n$) in this more general case. We conduct a series of empirical evaluations with off-the-shelf MIAs using training sets built from real-world data showing different types of dependencies among samples. Our results reveal that training set dependencies can severely increase the performance of MIAs, and therefore assuming that data samples are statistically independent can significantly underestimate the performance of MIAs.

cs.CR↗

arXiv2022

Cache Me If You Can: Accuracy-Aware Inference Engine for Differentially Private Data Exploration

Differential privacy (DP) allows data analysts to query databases that contain users' sensitive information while providing a quantifiable privacy guarantee to users. Recent interactive DP systems such as APEx provide accuracy guarantees over the query responses, but fail to support a large number of queries with a limited total privacy budget, as they process incoming queries independently from past queries. We present an interactive, accuracy-aware DP query engine, CacheDP, which utilizes a differentially private cache of past responses, to answer the current workload at a lower privacy budget, while meeting strict accuracy guarantees. We integrate complex DP mechanisms with our structured cache, through novel cache-aware DP cost optimization. Our thorough evaluation illustrates that CacheDP can accurately answer various workload sequences, while lowering the privacy loss as compared to related work.

cs.CR↗

arXiv2022

Selective MPC: Distributed Computation of Differentially Private Key-Value Statistics

Key-value data is a naturally occurring data type that has not been thoroughly investigated in the local trust model. Existing local differentially private (LDP) solutions for computing statistics over key-value data suffer from the inherent accuracy limitations of each user adding their own noise. Multi-party computation (MPC) maintains better accuracy than LDP and similarly does not require a trusted central party. However, naively applying MPC to key-value data results in prohibitively expensive computation costs. In this work, we present selective multi-party computation, a novel approach to distributed computation that leverages DP leakage to efficiently and accurately compute statistics over key-value data. By providing each party with a view of a random subset of the data, we can capture subtractive noise. We prove that our protocol satisfies pure DP and is provably secure in the combined DP/MPC model. Our empirical evaluation demonstrates that we can compute statistics over 10,000 keys in 20 seconds and can scale up to 30 servers while obtaining results for a single key in under a second.

cs.CR↗

arXiv2021

Self-Attention Generative Adversarial Network for Iterative Reconstruction of CT Images

Computed tomography (CT) uses X-ray measurements taken from sensors around the body to generate tomographic images of the human body. Conventional reconstruction algorithms can be used if the X-ray data are adequately sampled and of high quality; however, concerns such as reducing dose to the patient, or geometric limitations on data acquisition, may result in low quality or incomplete data. Images reconstructed from these data using conventional methods are of poor quality, due to noise and other artifacts. The aim of this study is to train a single neural network to reconstruct high-quality CT images from noisy or incomplete CT scan data, including low-dose, sparse-view, and limited-angle scenarios. To accomplish this task, we train a generative adversarial network (GAN) as a signal prior, to be used in conjunction with the iterative simultaneous algebraic reconstruction technique (SART) for CT data. The network includes a self-attention block to model long-range dependencies in the data. We compare our Self-Attention GAN for CT image reconstruction with several state-of-the-art approaches, including denoising cycle GAN, CIRCLE GAN, and a total variation superiorized algorithm. Our approach is shown to have comparable overall performance to CIRCLE GAN, while outperforming the other two approaches.

eess.IV↗

arXiv2015

Convergence analysis of a polyenergetic SART algorithm

Purpose: We analyze a recently proposed polyenergetic version of the simultaneous algebraic reconstruction technique (SART). This algorithm, denoted pSART, replaces the monoenergetic forward projection operation used by SART with a post-log, polyenergetic forward projection, while leaving the rest of the algorithm unchanged. While the proposed algorithm provides good results empirically, convergence of the algorithm was not established mathematically in the original paper. Methods: We analyze pSART as a nonlinear fixed point iteration by explicitly computing the Jacobian of the iteration. A necessary condition for convergence is that the spectral radius of the Jacobian, evaluated at the fixed point, is less than one. A short proof of convergence for SART is also provided as a basis for comparison. Results: We show that the pSART algorithm is not guaranteed to converge, in general. The Jacobian of the iteration depends on several factors, including the system matrix and how one models the energy dependence of the linear attenuation coefficient. We provide a simple numerical example that shows that the spectral radius of the Jacobian matrix is not guaranteed to be less than one. A second set of numerical experiments using realistic CT system matrices, however, indicates that conditions for convergence are likely to be satisfied in practice. Conclusion: Although pSART is not mathematically guaranteed to converge, our numerical experiments indicate that it will tend to converge at roughly the same rate as SART for system matrices of the type encountered in CT imaging. Thus we conclude that the algorithm is still a useful method for reconstruction of polyenergetic CT data.

math.NA↗