Search

Large language models (LLMs) provide powerful means to leverage prior knowledge for predictive modeling when data is limited. In this work, we demonstrate how LLMs can use their compressed world knowledge to generate intrinsically interpretable machine learning models, i.e., decision trees, without any training data. We find that these zero-shot decision trees can surpass data-driven trees on some small-sized tabular datasets and that embeddings derived from these trees perform on par with data-driven tree-based embeddings on average. Our knowledge-driven decision tree induction and embedding approaches therefore serve as strong new baselines for data-driven machine learning methods in the low-data regime.

Non-equilibrium clustering and percolation are investigated in an archetypal model of two-dimensional active matter using dynamic simulations of self-propelled Brownian repulsive particles. We concentrate on the single-phase region up to moderate levels of activity, before motility-induced phase separation (MIPS) sets in. Weak activity promotes cluster formation and lowers the percolation threshold. However, driving the system further out of equilibrium partly reverses this effect, resulting in a minimum in the critical density for the formation of system-spanning clusters and introducing re-entrant percolation as a function of activity in the pre-MIPS regime. This non-monotonic behaviour arises from competition between activity-induced effective attraction (which eventually leads to MIPS) and activity-driven cluster breakup. Using an adapted iterative Boltzmann inversion method, we derive effective potentials to map weakly active cases onto a passive (equilibrium) model with conservative attraction, which can be characterised by Monte Carlo simulations. While the active and passive systems have practically identical radial distribution functions, we find decisive differences in higher-order structural correlations, to which the percolation threshold is highly sensitive. For sufficiently strong activity, no passive pairwise potential can reproduce the radial distribution function of the active system.

Large language models (LLMs) are now being considered and even deployed for applications that support high-stakes decision-making, such as recruitment and clinical decisions. While several methods have been proposed for measuring bias, there remains a gap between predictions, which are what the proposed methods consider, and how they are used to make decisions. In this work, we introduce Rank-Allocational-Based Bias Index (RABBI), a model-agnostic bias measure that assesses potential allocational harms arising from biases in LLM predictions. We compare RABBI and current bias metrics on two allocation decision tasks. We evaluate their predictive validity across ten LLMs and utility for model selection. Our results reveal that commonly-used bias metrics based on average performance gap and distribution distance fail to reliably capture group disparities in allocation outcomes, whereas RABBI exhibits a strong correlation with allocation disparities. Our work highlights the need to account for how models are used in contexts with limited resource constraints.

Detecting anomalous events in satellite telemetry is a critical task in space operations. This task, however, is extremely time-consuming, error-prone and human dependent, thus automated data-driven anomaly detection algorithms have been emerging at a steady pace. However, there are no publicly available datasets of real satellite telemetry accompanied with the ground-truth annotations that could be used to train and verify anomaly detection supervised models. In this article, we address this research gap and introduce the AI-ready benchmark dataset (OPSSAT-AD) containing the telemetry data acquired on board OPS-SAT -- a CubeSat mission which has been operated by the European Space Agency which has come to an end during the night of 22--23 May 2024 (CEST). The dataset is accompanied with the baseline results obtained using 30 supervised and unsupervised classic and deep machine learning algorithms for anomaly detection. They were trained and validated using the training-test dataset split introduced in this work, and we present a suggested set of quality metrics which should be always calculated to confront the new algorithms for anomaly detection while exploiting OPSSAT-AD. We believe that this work may become an important step toward building a fair, reproducible and objective validation procedure that can be used to quantify the capabilities of the emerging anomaly detection techniques in an unbiased and fully transparent way., Comment: 13 pages, 8 figures, 3 tables

Membership inference attacks are used as a key tool for disclosure auditing. They aim to infer whether an individual record was used to train a model. While such evaluations are useful to demonstrate risk, they are computationally expensive and often make strong assumptions about potential adversaries' access to models and training environments, and thus do not provide tight bounds on leakage from potential attacks. We show how prior claims around black-box access being sufficient for optimal membership inference do not hold for stochastic gradient descent, and that optimal membership inference indeed requires white-box access. Our theoretical results lead to a new white-box inference attack, IHA (Inverse Hessian Attack), that explicitly uses model parameters by taking advantage of computing inverse-Hessian vector products. Our results show that both auditors and adversaries may be able to benefit from access to model parameters, and we advocate for further research into white-box methods for membership inference., Comment: Accepted to Transactions on Machine Learning Research (TMLR)

Serious privacy concerns arise with the use of patient data in rule-based clinical decision support systems (CDSS). The goal of a privacy-preserving CDSS is to learn a population ruleset from individual clients' local rulesets, while protecting the potentially sensitive information contained in the rulesets. We present the first work focused on this problem and develop a framework for learning population rulesets with local differential privacy (LDP), suitable for use within a distributed CDSS and other distributed settings. Our rule discovery protocol uses a Monte-Carlo Tree Search (MCTS) method integrated with LDP to search a rule grammar in a structured way and find rule structures clients are likely to have. Randomized response queries are sent to clients to determine promising paths to search within the rule grammar. In addition, we introduce an adaptive budget allocation method which dynamically determines how much privacy loss budget to use at each query, resulting in better privacy-utility trade-offs. We evaluate our approach using three clinical datasets and find that we are able to learn population rulesets with high coverage (breadth of rules) and clinical utility even at low privacy loss budgets.

While third-party cookies have been a key component of the digital marketing ecosystem for years, they allow users to be tracked across web sites in ways that raise serious privacy concerns. Google has proposed the Privacy Sandbox initiative to enable ad targeting without third-party cookies. While there have been several studies focused on other aspects of this initiative, there has been little analysis to date as to how well the system achieves the intended goal of preventing request linking. This work focuses on analyzing linkage privacy risks for the reporting mechanisms proposed in the Protected Audience (PrAu) proposal (previously known as FLEDGE), which is intended to enable online remarketing without using third-party cookies. We summarize the overall workflow of PrAu and highlight potential privacy risks associated with its proposed design, focusing on scenarios in which adversaries attempt to link requests to different sites to the same user. We show how a realistic adversary would be still able to use the privacy-protected reporting mechanisms to link user requests and conduct mass surveillance, even with correct implementations of all the currently proposed privacy mechanisms.

We consider the problem of building non-invertible quantum symmetries (as characterized by actions of unitary fusion categories) on noncommutative tori. We introduce a general method to construct actions of fusion categories on inductive limit C*-algberas using finite dimenionsal data, and then apply it to obtain AT-actions of arbitrary Haagerup-Izumi categories on noncommutative 2-tori, of the even part of the $E_{8}$ subfactor on a noncommutative 3-torus, and of $\text{PSU}(2)_{15}$ on a noncommutative 4-torus.

Statistical fairness stipulates equivalent outcomes for every protected group, whereas causal fairness prescribes that a model makes the same prediction for an individual regardless of their protected characteristics. Counterfactual data augmentation (CDA) is effective for reducing bias in NLP models, yet models trained with CDA are often evaluated only on metrics that are closely tied to the causal fairness notion; similarly, sampling-based methods designed to promote statistical fairness are rarely evaluated for causal fairness. In this work, we evaluate both statistical and causal debiasing methods for gender bias in NLP models, and find that while such methods are effective at reducing bias as measured by the targeted metric, they do not necessarily improve results on other bias metrics. We demonstrate that combinations of statistical and causal debiasing techniques are able to reduce bias measured through both types of metrics., Comment: NAACL 2024 (Findings)

Membership inference attacks (MIAs) attempt to predict whether a particular datapoint is a member of a target model's training data. Despite extensive research on traditional machine learning models, there has been limited work studying MIA on the pre-training data of large language models (LLMs). We perform a large-scale evaluation of MIAs over a suite of language models (LMs) trained on the Pile, ranging from 160M to 12B parameters. We find that MIAs barely outperform random guessing for most settings across varying LLM sizes and domains. Our further analyses reveal that this poor performance can be attributed to (1) the combination of a large dataset and few training iterations, and (2) an inherently fuzzy boundary between members and non-members. We identify specific settings where LLMs have been shown to be vulnerable to membership inference and show that the apparent success in such settings can be attributed to a distribution shift, such as when members and non-members are drawn from the seemingly identical domain but with different temporal ranges. We release our code and data as a unified benchmark package that includes all existing MIAs, supporting future work., Comment: Accepted at Conference on Language Modeling (COLM), 2024

Machine learning is susceptible to poisoning attacks, in which an attacker controls a small fraction of the training data and chooses that data with the goal of inducing some behavior unintended by the model developer in the trained model. We consider a realistic setting in which the adversary with the ability to insert a limited number of data points attempts to control the model's behavior on a specific subpopulation. Inspired by previous observations on disparate effectiveness of random label-flipping attacks on different subpopulations, we investigate the properties that can impact the effectiveness of state-of-the-art poisoning attacks against different subpopulations. For a family of 2-dimensional synthetic datasets, we empirically find that dataset separability plays a dominant role in subpopulation vulnerability for less separable datasets. However, well-separated datasets exhibit more dependence on individual subpopulation properties. We further discover that a crucial subpopulation property is captured by the difference in loss on the clean dataset between the clean model and a target model that misclassifies the subpopulation, and a subpopulation is much easier to attack if the loss difference is small. This property also generalizes to high-dimensional benchmark datasets. For the Adult benchmark dataset, we show that we can find semantically-meaningful subpopulation properties that are related to the susceptibilities of a selected group of subpopulations. The results in this paper are accompanied by a fully interactive web-based visualization of subpopulation poisoning attacks found at https://uvasrg.github.io/visualizing-poisoning, Comment: 18 pages, 11 figures

Numerous works study black-box attacks on image classifiers. However, these works make different assumptions on the adversary's knowledge and current literature lacks a cohesive organization centered around the threat model. To systematize knowledge in this area, we propose a taxonomy over the threat space spanning the axes of feedback granularity, the access of interactive queries, and the quality and quantity of the auxiliary data available to the attacker. Our new taxonomy provides three key insights. 1) Despite extensive literature, numerous under-explored threat spaces exist, which cannot be trivially solved by adapting techniques from well-explored settings. We demonstrate this by establishing a new state-of-the-art in the less-studied setting of access to top-k confidence scores by adapting techniques from well-explored settings of accessing the complete confidence vector, but show how it still falls short of the more restrictive setting that only obtains the prediction label, highlighting the need for more research. 2) Identification the threat model of different attacks uncovers stronger baselines that challenge prior state-of-the-art claims. We demonstrate this by enhancing an initially weaker baseline (under interactive query access) via surrogate models, effectively overturning claims in the respective paper. 3) Our taxonomy reveals interactions between attacker knowledge that connect well to related areas, such as model inversion and extraction attacks. We discuss how advances in other areas can enable potentially stronger black-box attacks. Finally, we emphasize the need for a more realistic assessment of attack success by factoring in local attack runtime. This approach reveals the potential for certain attacks to achieve notably higher success rates and the need to evaluate attacks in diverse and harder settings, highlighting the need for better selection criteria., Comment: Accepted at SaTML 2024

Large Language Models (LLMs) are advancing at a remarkable pace, with myriad applications under development. Unlike most earlier machine learning models, they are no longer built for one specific application but are designed to excel in a wide range of tasks. A major part of this success is due to their huge training datasets and the unprecedented number of model parameters, which allow them to memorize large amounts of information contained in the training data. This memorization goes beyond mere language, and encompasses information only present in a few documents. This is often desirable since it is necessary for performing tasks such as question answering, and therefore an important part of learning, but also brings a whole array of issues, from privacy and security to copyright and beyond. LLMs can memorize short secrets in the training data, but can also memorize concepts like facts or writing styles that can be expressed in text in many different ways. We propose a taxonomy for memorization in LLMs that covers verbatim text, facts, ideas and algorithms, writing styles, distributional properties, and alignment goals. We describe the implications of each type of memorization - both positive and negative - for model performance, privacy, security and confidentiality, copyright, and auditing, and ways to detect and prevent memorization. We further highlight the challenges that arise from the predominant way of defining memorization with respect to model behavior instead of model weights, due to LLM-specific phenomena such as reasoning capabilities or differences between decoding algorithms. Throughout the paper, we describe potential risks and opportunities arising from memorization in LLMs that we hope will motivate new research directions.

This document presents a summary of the 2023 Terrestrial Very-Long-Baseline Atom Interferometry Workshop hosted by CERN. The workshop brought together experts from around the world to discuss the exciting developments in large-scale atom interferometer (AI) prototypes and their potential for detecting ultralight dark matter and gravitational waves. The primary objective of the workshop was to lay the groundwork for an international TVLBAI proto-collaboration. This collaboration aims to unite researchers from different institutions to strategize and secure funding for terrestrial large-scale AI projects. The ultimate goal is to create a roadmap detailing the design and technology choices for one or more km-scale detectors, which will be operational in the mid-2030s. The key sections of this report present the physics case and technical challenges, together with a comprehensive overview of the discussions at the workshop together with the main conclusions., Comment: Summary of the Terrestrial Very-Long-Baseline Atom Interferometry Workshop held at CERN: https://indico.cern.ch/event/1208783/

Mendelian randomization (MR) is an epidemiological method that can be used to strengthen causal inference regarding the relationship between a modifiable environmental exposure and a medically relevant trait and to estimate the magnitude of this relationship1. Recently, there has been considerable interest in using MR to examine potential causal relationships between parental phenotypes and outcomes amongst their offspring. In a recent issue of BMC Research Notes, Woolf et al (2023) present a new method, GWAS by subtraction, to derive genome-wide summary statistics for paternal smoking and other paternal phenotypes with the goal that these estimates can then be used in downstream (including two sample) MR studies. Whilst a potentially useful goal, Woolf et al. (2023) focus on the wrong parameter of interest for useful genome-wide association studies (GWAS) and downstream cross-generational MR studies, and the estimator that they derive is neither efficient nor appropriate for such use., Comment: 8 pages, 0 figures

We study indiscriminate poisoning for linear learners where an adversary injects a few crafted examples into the training data with the goal of forcing the induced model to incur higher test error. Inspired by the observation that linear learners on some datasets are able to resist the best known attacks even without any defenses, we further investigate whether datasets can be inherently robust to indiscriminate poisoning attacks for linear learners. For theoretical Gaussian distributions, we rigorously characterize the behavior of an optimal poisoning attack, defined as the poisoning strategy that attains the maximum risk of the induced model at a given poisoning budget. Our results prove that linear learners can indeed be robust to indiscriminate poisoning if the class-wise data distributions are well-separated with low variance and the size of the constraint set containing all permissible poisoning points is also small. These findings largely explain the drastic variation in empirical attack performance of the state-of-the-art poisoning attacks on linear learners across benchmark datasets, making an important initial step towards understanding the underlying reasons some learning tasks are vulnerable to data poisoning attacks., Comment: NeurIPS 2023 camera-ready version, 39 pages

Motivated by the Freed-Hopkins-Teleman theorem we study graded equivariant higher twists of $K$-theory for the groups $G = SU(n)$ induced by exponential functors. We compute the rationalisation of these groups for all $n$ and all non-trivial functors $F$ using the Mayer-Vietoris spectral sequence. Similar to the classical case only the $K$-theory in degree $\dim(G)$ is non-trivial and the non-vanishing group is a quotient of a localisation of the representation ring $R(G) \otimes \mathbb{Q}$ by a higher fusion ideal $J_{F,\mathbb{Q}}$. We give generators for this ideal and prove that these can be obtained as derivatives of a potential., Comment: 33 pages, one figure, minor mistake fixed in Section 3.1.1, setting has been adjusted to exponential functors that are symmetric monoidal and take values in graded vector spaces

In this article, we present a detailed review of current practices and state-of-the-art methodologies in the field of differential privacy (DP), with a focus of advancing DP's deployment in real-world applications. Key points and high-level contents of the article were originated from the discussions from "Differential Privacy (DP): Challenges Towards the Next Frontier," a workshop held in July 2022 with experts from industry, academia, and the public sector seeking answers to broad questions pertaining to privacy and its implications in the design of industry-grade systems. This article aims to provide a reference point for the algorithmic and design decisions within the realm of privacy, highlighting important challenges and potential research directions. Covering a wide spectrum of topics, this article delves into the infrastructure needs for designing private systems, methods for achieving better privacy/utility trade-offs, performing privacy attacks and auditing, as well as communicating privacy with broader audiences and stakeholders.

Transfer learning is a popular method for tuning pretrained (upstream) models for different downstream tasks using limited data and computational resources. We study how an adversary with control over an upstream model used in transfer learning can conduct property inference attacks on a victim's tuned downstream model. For example, to infer the presence of images of a specific individual in the downstream training set. We demonstrate attacks in which an adversary can manipulate the upstream model to conduct highly effective and specific property inference attacks (AUC score $> 0.9$), without incurring significant performance loss on the main task. The main idea of the manipulation is to make the upstream model generate activations (intermediate features) with different distributions for samples with and without a target property, thus enabling the adversary to distinguish easily between downstream models trained with and without training examples that have the target property. Our code is available at https://github.com/yulongt23/Transfer-Inference., Comment: Accepted to CVPR 2023

This paper surveys the long-standing connections and impact between Vaughan Jones's theory of subfactors and various topics in mathematical physics, namely statistical mechanics,quantum field theory,quantum information and two-dimensional conformal field theory., Comment: 22 pages, 1 figure. To appear in an issue of the Bulletin of the AMS, dedicated to the mathematical legacy of Vaughan Jones

We focus on the problem of generating high-quality, private synthetic glucose traces, a task generalizable to many other time series sources. Existing methods for time series data synthesis, such as those using Generative Adversarial Networks (GANs), are not able to capture the innate characteristics of glucose data and cannot provide any formal privacy guarantees without severely degrading the utility of the synthetic data. In this paper we present GlucoSynth, a novel privacy-preserving GAN framework to generate synthetic glucose traces. The core intuition behind our approach is to conserve relationships amongst motifs (glucose events) within the traces, in addition to temporal dynamics. Our framework incorporates differential privacy mechanisms to provide strong formal privacy guarantees. We provide a comprehensive evaluation on the real-world utility of the data using 1.2 million glucose traces; GlucoSynth outperforms all previous methods in its ability to generate high-quality synthetic glucose traces with strong privacy guarantees.

With tools like GitHub Copilot, automatic code suggestion is no longer a dream in software engineering. These tools, based on large language models, are typically trained on massive corpora of code mined from unvetted public sources. As a result, these models are susceptible to data poisoning attacks where an adversary manipulates the model's training by injecting malicious data. Poisoning attacks could be designed to influence the model's suggestions at run time for chosen contexts, such as inducing the model into suggesting insecure code payloads. To achieve this, prior attacks explicitly inject the insecure code payload into the training data, making the poison data detectable by static analysis tools that can remove such malicious data from the training set. In this work, we demonstrate two novel attacks, COVERT and TROJANPUZZLE, that can bypass static analysis by planting malicious poison data in out-of-context regions such as docstrings. Our most novel attack, TROJANPUZZLE, goes one step further in generating less suspicious poison data by never explicitly including certain (suspicious) parts of the payload in the poison data, while still inducing a model that suggests the entire payload when completing code (i.e., outside docstrings). This makes TROJANPUZZLE robust against signature-based dataset-cleansing methods that can filter out suspicious sequences from the training data. Our evaluation against models of two sizes demonstrates that both COVERT and TROJANPUZZLE have significant implications for practitioners when selecting code used to train or tune code-suggestion models.

Deploying machine learning models in production may allow adversaries to infer sensitive information about training data. There is a vast literature analyzing different types of inference risks, ranging from membership inference to reconstruction attacks. Inspired by the success of games (i.e., probabilistic experiments) to study security properties in cryptography, some authors describe privacy inference risks in machine learning using a similar game-based style. However, adversary capabilities and goals are often stated in subtly different ways from one presentation to the other, which makes it hard to relate and compose results. In this paper, we present a game-based framework to systematize the body of knowledge on privacy inference risks in machine learning. We use this framework to (1) provide a unifying structure for definitions of inference risks, (2) formally establish known relations among definitions, and (3) to uncover hitherto unknown relations that would have been difficult to spot otherwise., Comment: 20 pages, to appear in 2023 IEEE Symposium on Security and Privacy

A distribution inference attack aims to infer statistical properties of data used to train machine learning models. These attacks are sometimes surprisingly potent, but the factors that impact distribution inference risk are not well understood and demonstrated attacks often rely on strong and unrealistic assumptions such as full knowledge of training environments even in supposedly black-box threat scenarios. To improve understanding of distribution inference risks, we develop a new black-box attack that even outperforms the best known white-box attack in most settings. Using this new attack, we evaluate distribution inference risk while relaxing a variety of assumptions about the adversary's knowledge under black-box access, like known model architectures and label-only access. Finally, we evaluate the effectiveness of previously proposed defenses and introduce new defenses. We find that although noise-based defenses appear to be ineffective, a simple re-sampling defense can be highly effective. Code is available at https://github.com/iamgroot42/dissecting_distribution_inference, Comment: Accepted at SaTML 2023 (updated Yifu's email address)

Traditional (fickle) adversarial examples involve finding a small perturbation that does not change an input's true label but confuses the classifier into outputting a different prediction. Conversely, obstinate adversarial examples occur when an adversary finds a small perturbation that preserves the classifier's prediction but changes the true label of an input. Adversarial training and certified robust training have shown some effectiveness in improving the robustness of machine learnt models to fickle adversarial examples. We show that standard adversarial training methods focused on reducing vulnerability to fickle adversarial examples may make a model more vulnerable to obstinate adversarial examples, with experiments for both natural language inference and paraphrase identification tasks. To counter this phenomenon, we introduce Balanced Adversarial Training, which incorporates contrastive learning to increase robustness against both fickle and obstinate adversarial examples., Comment: EMNLP 2022

Models can expose sensitive information about their training data. In an attribute inference attack, an adversary has partial knowledge of some training records and access to a model trained on those records, and infers the unknown values of a sensitive feature of those records. We study a fine-grained variant of attribute inference we call \emph{sensitive value inference}, where the adversary's goal is to identify with high confidence some records from a candidate set where the unknown attribute has a particular sensitive value. We explicitly compare attribute inference with data imputation that captures the training distribution statistics, under various assumptions about the training data available to the adversary. Our main conclusions are: (1) previous attribute inference methods do not reveal more about the training data from the model than can be inferred by an adversary without access to the trained model, but with the same knowledge of the underlying distribution as needed to train the attribute inference attack; (2) black-box attribute inference attacks rarely learn anything that cannot be learned without the model; but (3) white-box attacks, which we introduce and evaluate in the paper, can reliably identify some records with the sensitive value attribute that would not be predicted without having access to the model. Furthermore, we show that proposed defenses such as differentially private training and removing vulnerable records from training do not mitigate this privacy risk. The code for our experiments is available at \url{https://github.com/bargavj/EvaluatingDPML}., Comment: 13 (main body) + 4 (references and appendix) pages. To appear in CCS'22