arXiv subjects

Ravi Kumar

Publications and source records attributed to Ravi Kumar.

At least 73 records · Page 4Linked to original sources

arXiv2024

Differentially Private Optimization with Sparse Gradients

Motivated by applications of large embedding models, we study differentially private (DP) optimization problems under sparsity of individual gradients. We start with new near-optimal bounds for the classic mean estimation problem but with sparse data, improving upon existing algorithms particularly for the high-dimensional regime. Building on this, we obtain pure- and approximate-DP algorithms with almost optimal rates for stochastic convex optimization with sparse gradients; the former represents the first nearly dimension-independent rates for this problem. Finally, we study the approximation of stationary points for the empirical loss in approximate-DP optimization and obtain rates that depend on sparsity instead of dimension, modulo polylogarithmic factors.

cs.LG↗

arXiv2024

Differential Privacy on Trust Graphs

We study differential privacy (DP) in a multi-party setting where each party only trusts a (known) subset of the other parties with its data. Specifically, given a trust graph where vertices correspond to parties and neighbors are mutually trusting, we give a DP algorithm for aggregation with a much better privacy-utility trade-off than in the well-studied local model of DP (where each party trusts no other party). We further study a robust variant where each party trusts all but an unknown subset of at most $t$ of its neighbors (where $t$ is a given parameter), and give an algorithm for this setting. We complement our algorithms with lower bounds, and discuss implications of our work to other tasks in private learning and analytics.

cs.CR↗

arXiv2024

Unlearn and Burn: Adversarial Machine Unlearning Requests Destroy Model Accuracy

Machine unlearning algorithms, designed for selective removal of training data from models, have emerged as a promising approach to growing privacy concerns. In this work, we expose a critical yet underexplored vulnerability in the deployment of unlearning systems: the assumption that the data requested for removal is always part of the original training set. We present a threat model where an attacker can degrade model accuracy by submitting adversarial unlearning requests for data not present in the training set. We propose white-box and black-box attack algorithms and evaluate them through a case study on image classification tasks using the CIFAR-10 and ImageNet datasets, targeting a family of widely used unlearning methods. Our results show extremely poor test accuracy following the attack: 3.6% on CIFAR-10 and 0.4% on ImageNet for white-box attacks, and 8.5% on CIFAR-10 and 1.3% on ImageNet for black-box attacks. Additionally, we evaluate various verification mechanisms to detect the legitimacy of unlearning requests and reveal the challenges in verification, as most of the mechanisms fail to detect stealthy attacks without severely impairing their ability to process valid requests. These findings underscore the urgent need for research on more robust request verification methods and unlearning protocols, should the deployment of machine unlearning systems become more prevalent in the future.

cs.CR↗

arXiv2024

Probing Elastic Isotropy in Entropy Stabilized Transition Metal Oxides: Experimental Estimation of Single Crystal Elastic Constants from Polycrystalline Materials

Single Crystal Elastic Constants (SECs) are pivotal for understanding material deformation, validating interatomic potentials, and enabling crucial material simulations. The entropy stabilized oxide showcases intriguing properties, underscoring the necessity for the determination of precise SECs to establish reliable interatomic potential and unlock its full potential using simulations. This study presents an innovative methodology for estimating SECs from polycrystalline materials, requiring only two diffraction elastic constants and isotropic elastic constants for crystals with cubic symmetry. Validation using phase-pure nickel demonstrated good agreement with existing literature values, with a maximum 11.5% deviation for $C_{12}$ values. Extending the methodology to [(MgNiCoCuZn)O], SECs were calculated: 219 GPa for $C_{11}$, 116 GPa for $C_{12}$, and 51 GPa for $C_{44}$. Comparison with literature-reported values from DFT calculations revealed a significant divergence, ranging from 25% to 59% in the bulk and shear modulus calculated using the Voigt-Reuss-Hill average. To comprehend this disparity, we conducted DFT calculations and thoroughly examined the factors influencing these values. This study not only introduces a straightforward and dependable SEC estimation methodology but also provides precise experimental SEC values for [(MgNiCoCuZn)O] entropy stabilized oxides at ambient conditions, crucial for developing accurate interatomic potentials in future research.

cond-mat.mtrl-sci↗

arXiv2024

Absence of heat flow in ν = 0 quantum Hall ferromagnet in bilayer graphene

The charge neutrality point of bilayer graphene, denoted as ν = 0 state, manifests competing phases marked by spontaneously broken isospin (spin/valley/layer) symmetries under external magnetic and electric fields. However, due to their electrically insulating nature, identifying these phases through electrical conductance measurements remains challenging. A recent theoretical proposal introduces a novel approach, employing thermal transport measurements to detect these competing phases. Here, we experimentally explore the bulk thermal transport of the ν = 0 state in bilayer graphene to investigate its ground states and collective excitations associated with isospin. While the theory anticipates a finite thermal conductance in the ν = 0 state, our findings unveil an absence of detectable thermal conductance. Through variations in the external electric field and temperature-dependent measurements, our results suggest towards gapped collective excitations at ν = 0 state. Our findings underscore the necessity for further investigations into the nature of ν = 0.

cond-mat.mes-hall↗

arXiv2024

Mind the Privacy Unit! User-Level Differential Privacy for Language Model Fine-Tuning

Large language models (LLMs) have emerged as powerful tools for tackling complex tasks across diverse domains, but they also raise privacy concerns when fine-tuned on sensitive data due to potential memorization. While differential privacy (DP) offers a promising solution by ensuring models are 'almost indistinguishable' with or without any particular privacy unit, current evaluations on LLMs mostly treat each example (text record) as the privacy unit. This leads to uneven user privacy guarantees when contributions per user vary. We therefore study user-level DP motivated by applications where it necessary to ensure uniform privacy protection across users. We present a systematic evaluation of user-level DP for LLM fine-tuning on natural language generation tasks. Focusing on two mechanisms for achieving user-level DP guarantees, Group Privacy and User-wise DP-SGD, we investigate design choices like data selection strategies and parameter tuning for the best privacy-utility tradeoff.

cs.CL↗

arXiv2024

Rapid Likelihood Free Inference of Compact Binary Coalescences using Accelerated Hardware

We report a gravitational-wave parameter estimation algorithm, AMPLFI, based on likelihood-free inference using normalizing flows. The focus of AMPLFI is to perform real-time parameter estimation for candidates detected by machine-learning based compact binary coalescence search, Aframe. We present details of our algorithm and optimizations done related to data-loading and pre-processing on accelerated hardware. We train our model using binary black-hole (BBH) simulations on real LIGO-Virgo detector noise. Our model has $\sim 6$ million trainable parameters with training times $\lesssim 24$ hours. Based on online deployment on a mock data stream of LIGO-Virgo data, Aframe + AMPLFI is able to pick up BBH candidates and infer parameters for real-time alerts from data acquisition with a net latency of $\sim 6$s.

gr-qc↗

arXiv2024

Improving Online Algorithms via ML Predictions

In this work we study the problem of using machine-learned predictions to improve the performance of online algorithms. We consider two classical problems, ski rental and non-clairvoyant job scheduling, and obtain new online algorithms that use predictions to make their decisions. These algorithms are oblivious to the performance of the predictor, improve with better predictions, but do not degrade much if the predictions are poor.

cs.DS↗

arXiv2024

On Convex Optimization with Semi-Sensitive Features

We study the differentially private (DP) empirical risk minimization (ERM) problem under the semi-sensitive DP setting where only some features are sensitive. This generalizes the Label DP setting where only the label is sensitive. We give improved upper and lower bounds on the excess risk for DP-ERM. In particular, we show that the error only scales polylogarithmically in terms of the sensitive domain size, improving upon previous results that scale polynomially in the sensitive domain size (Ghazi et al., 2021).

cs.LG↗

arXiv2024

On Computing Pairwise Statistics with Local Differential Privacy

We study the problem of computing pairwise statistics, i.e., ones of the form $\binom{n}{2}^{-1} \sum_{i \ne j} f(x_i, x_j)$, where $x_i$ denotes the input to the $i$th user, with differential privacy (DP) in the local model. This formulation captures important metrics such as Kendall's $τ$ coefficient, Area Under Curve, Gini's mean difference, Gini's entropy, etc. We give several novel and generic algorithms for the problem, leveraging techniques from DP algorithms for linear queries.

cs.DS↗

arXiv2024

How Private are DP-SGD Implementations?

We demonstrate a substantial gap between the privacy guarantees of the Adaptive Batch Linear Queries (ABLQ) mechanism under different types of batch sampling: (i) Shuffling, and (ii) Poisson subsampling; the typical analysis of Differentially Private Stochastic Gradient Descent (DP-SGD) follows by interpreting it as a post-processing of ABLQ. While shuffling-based DP-SGD is more commonly used in practical implementations, it has not been amenable to easy privacy analysis, either analytically or even numerically. On the other hand, Poisson subsampling-based DP-SGD is challenging to scalably implement, but has a well-understood privacy analysis, with multiple open-source numerically tight privacy accountants available. This has led to a common practice of using shuffling-based DP-SGD in practice, but using the privacy analysis for the corresponding Poisson subsampling version. Our result shows that there can be a substantial gap between the privacy analysis when using the two types of batch sampling, and thus advises caution in reporting privacy parameters for DP-SGD.

cs.LG↗

arXiv2024

Individualized Privacy Accounting via Subsampling with Applications in Combinatorial Optimization

In this work, we give a new technique for analyzing individualized privacy accounting via the following simple observation: if an algorithm is one-sided add-DP, then its subsampled variant satisfies two-sided DP. From this, we obtain several improved algorithms for private combinatorial optimization problems, including decomposable submodular maximization and set cover. Our error guarantees are asymptotically tight and our algorithm satisfies pure-DP while previously known algorithms (Gupta et al., 2010; Chaturvedi et al., 2021) are approximate-DP. We also show an application of our technique beyond combinatorial optimization by giving a pure-DP algorithm for the shifting heavy hitter problem in a stream; previously, only an approximateDP algorithm was known (Kaplan et al., 2021; Cohen & Lyu, 2023).

cs.DS↗

arXiv2024

Searching for gravitational wave optical counterparts with the Zwicky Transient Facility: summary of O4a

During the first half of the fourth observing run (O4a) of the International Gravitational Wave Network (IGWN), the Zwicky Transient Facility (ZTF) conducted a systematic search for kilonova (KN) counterparts to binary neutron star (BNS) and neutron star-black hole (NSBH) merger candidates. Here, we present a comprehensive study of the five high-significance (FAR < 1 per year) BNS and NSBH candidates in O4a. Our follow-up campaigns relied on both target-of-opportunity observations (ToO) and re-weighting of the nominal survey schedule to maximize coverage. We describe the toolkit we have been developing, Fritz, an instance of SkyPortal, instrumental in coordinating and managing our telescope scheduling, candidate vetting, and follow-up observations through a user-friendly interface. ZTF covered a total of 2841 deg$^2$ within the skymaps of the high-significance GW events, reaching a median depth of g~20.2 mag. We circulated 15 candidates, but found no viable KN counterpart to any of the GW events. Based on the ZTF non-detections of the high-significance events in O4a, we used a Bayesian approach, nimbus, to quantify the posterior probability of KN model parameters that are consistent with our non-detections. Our analysis favors KNe with initial absolute magnitude fainter than -16 mag. The joint posterior probability of a GW170817-like KN associated with all our O4a follow-ups was 64%. Additionally, we use a survey simulation software, simsurvey, to determine that our combined filtered efficiency to detect a GW170817-like KN is 36%, when considering the 5 confirmed astrophysical events in O3 (1 BNS and 4 NSBH), along with our O4a follow-ups. Following Kasliwal et al. (2020), we derived joint constraints on the underlying KN luminosity function based on our O3 and O4a follow-ups, determining that no more than 76% of KNe fading at 1 mag/day can peak at a magnitude brighter than -17.5 mag.

astro-ph.HE↗

arXiv2024

Online Load and Graph Balancing for Random Order Inputs

Online load balancing for heterogeneous machines aims to minimize the makespan (maximum machine workload) by scheduling arriving jobs with varying sizes on different machines. In the adversarial setting, where an adversary chooses not only the collection of job sizes but also their arrival order, the problem is well-understood and the optimal competitive ratio is known to be $Θ(\log m)$ where $m$ is the number of machines. In the more realistic random arrival order model, the understanding is limited. Previously, the best lower bound on the competitive ratio was only $Ω(\log \log m)$. We significantly improve this bound by showing an $Ω( \sqrt {\log m})$ lower bound, even for the restricted case where each job has a unit size on two machines and infinite size on the others. On the positive side, we propose an $O(\log m/\log \log m)$-competitive algorithm, demonstrating that better performance is possible in the random arrival model.

cs.DS↗

arXiv2024

Harnessing the Duality of Magnetism and Conductivity: A Review of Oxide based Dilute Magnetic Semiconductors

Over the last two decades, the new branch of spintronics, i.e., semiconductor spintronics, has gained more attention because it integrates the characteristics of conventional semiconductors, such as optical bandgap and charge carriers, helpful for processing and computing pieces of information combined with magnets for data storage applications in a single device. Likewise, substituting transition metal (TM) ions to induce magnetic qualities into semiconductors or oxides creates dilute magnetic semiconductors (DMSs) or oxides (DMOs) with high electronic, photonic, and magnetic functionality. This review article discusses the historical outline of magnetic semiconductors with their origin and mechanism. It also includes a concise overview of various DMO systems based on their conductivity (p-type and n-type) to elucidate the synthesis, origin, and control mechanisms and further evoke the prepared spintronics devices. The occurrence of RTFM with transparency and conductivity can be helpful in spintronics device fabrications, which was assumed to be governed by the formation of intrinsic defects, charge carriers, morphology, and the induced exchange interactions between ions. The DMOs-based spintronics devices, such as magneto-optical devices, transparent ferromagnets, and spin-based solar cells, exploit both semiconducting and magnetic properties, which have also been discussed in this review article with outlook and perspectives.

cond-mat.mtrl-sci↗

arXiv2024

Fock-space relativistic coupled-cluster calculations of clock transition properties in Pb$^{2+}$

We have implemented an all-particle multireference Fock-space relativistic coupled-cluster theory to probe $6s^2{\;^1}S_{0} - 6s6p{\;^3P^o_{0}}$ clock transition in an even isotope of Pb$^{2+}$. We have computed, excitation energy for several low lying states, E1 and M1 transition amplitudes, and the lifetime of the clock state. Moreover, we have also calculated the ground state dipole polarizability using perturbed relativistic coupled-cluster theory. To improve the accuracy of results, we incorporated the corrections from the relativistic and QED effects in all our calculations. The contributions from triple excitations are accounted perturbatively. Our computed excitation energies are in excellent agreement with the experimental values for all the states. Our result for lifetime, $9.76\times10^{6}$ s, of clock state is $\approx$ 8.5\% larger than the previous value using CI+MBPT [Phys. Rev. Lett. {\bf 127}, 013201 (2021)]. Based on our analysis, we find that the contributions from the {\em valence-valence} correlations arising from higher energy configurations and the corrections from the perturbative triples and QED effects are essential to get accurate clock transition properties in Pb$^{2+}$. Our computed value of dipole polarizability is in good agreement with the available theoretical and experimental data.

physics.atom-ph↗

arXiv2024

Differentially Private Ad Conversion Measurement

In this work, we study ad conversion measurement, a central functionality in digital advertising, where an advertiser seeks to estimate advertiser website (or mobile app) conversions attributed to ad impressions that users have interacted with on various publisher websites (or mobile apps). Using differential privacy (DP), a notion that has gained in popularity due to its strong mathematical guarantees, we develop a formal framework for private ad conversion measurement. In particular, we define the notion of an operationally valid configuration of the attribution rule, DP adjacency relation, contribution bounding scope and enforcement point. We then provide, for the set of configurations that most commonly arises in practice, a complete characterization, which uncovers a delicate interplay between attribution and privacy.

cs.CR↗

arXiv2024

Aharonov-Bohm interference and the evolution of phase jumps in fractional quantum Hall Fabry-Perot interferometers based on bi-layer graphene

Quasi-particles in fractional quantum Hall states are collective excitations that carry fractional charge and anyonic statistics. While the fractional charge affects semi-classical characteristics such as shot noise and charging energies, the anyonic statistics is most notable in quantum interference phenomena. In this study, we utilize a versatile bilayer graphene-based Fabry-Pérot Interferometer (FPI) that facilitates the study of a broad spectrum of operating regimes, from Coulomb-dominated to Aharonov-Bohm, for both integer and fractional quantum Hall states. Focusing on the $ν$=$1 \over 3$ fractional quantum Hall state, we study the Aharonov-Bohm interference of quasi-particles when the magnetic flux through an interference loop and the charge density within the loop are independently varied. When their combined variation is such that the Landau filling remains $1 \over 3$ we observe pristine Aharonov-Bohm oscillations with a period of three flux quanta, as is expected from the interference of quasi-particles of one-third of the electron charge. When the combined variation is such that it leads to quasi-particles addition or removal from the loop, phase jumps emerge, and alter the phase evolution. Notably, across all cases, the average phase consistently increases by 2$π$ with each addition of one electron to the loop.

cond-mat.mes-hall↗