arXiv subjects

Akash Bharadwaj

Publications and source records attributed to Akash Bharadwaj.

9 recordsLinked to original sources

arXiv2026

Steal the Patch Size: Adversarially Manipulate Vision-Language Models

We present a black-box model-stealing attack that recovers private vision-tokenizer configurations of deployed vision-language models (VLMs), including the visual patch size and input preprocessing pipeline. The key idea is a task-level side channel induced by ViT-style patchification: when a synthetic grid image is aligned with the hidden patch grid, boundary cues are erased at tokenization, causing periodic accuracy drop. By sweeping the grid cell size and measuring these collapses, we infer the patch size; by introducing padding and a consistency-check test, we further identify whether preprocessing is dynamic- or fixed-resolution and recover the target resize resolution. Across open-source Qwen-VL variants and proprietary models including GPT and Claude, we reliably recover tokenizer-related parameters. Finally, we show that such leakage enables preprocessing-aware transfer attacks and model-targeted adversarial manipulation.

cs.CV↗

arXiv2026

Configurable Reward Model for Balanced Safety Alignment

Aligning large language models (LLMs) to heterogeneous and rapidly evolving safety requirements remains a critical challenge. Existing instruction-tuned LLMs and standalone safety classifiers often fail to generalize to new safety configurations, motivating the need for Reward Models (RMs) that are explicitly configurable to changing specifications. We introduce the Configurable Safety Reward Model (CSRM), which is jointly optimized for calibrated safety compliance and reward modeling. Our approach is supported by configuration-targeted data augmentation that enforces instruction adherence while preserving relative severity structure. The resulting RM is sensitive to fine-grained safety configurations and conversational nuances, substantially improving generalization to previously unseen safety configurations. CSRM achieves state-of-the-art performance on recent configurable safety benchmarks, including CoSApien (94.6% F1) and DynaBench (75.8% F1), without requiring additional human annotation. When used for downstream safety alignment, CSRM yields LLMs with a significantly improved helpfulness-safety tradeoff compared to existing baselines.

cs.CL↗

arXiv2025

Jailbreak-Zero: A Path to Pareto Optimal Red Teaming for Large Language Models

This paper introduces Jailbreak-Zero, a novel red teaming methodology that shifts the paradigm of Large Language Model (LLM) safety evaluation from a constrained example-based approach to a more expansive and effective policy-based framework. By leveraging an attack LLM to generate a high volume of diverse adversarial prompts and then fine-tuning this attack model with a preference dataset, Jailbreak-Zero achieves Pareto optimality across the crucial objectives of policy coverage, attack strategy diversity, and prompt fidelity to real user inputs. The empirical evidence demonstrates the superiority of this method, showcasing significantly higher attack success rates against both open-source and proprietary models like GPT-40 and Claude 3.5 when compared to existing state-of-the-art techniques. Crucially, Jailbreak-Zero accomplishes this while producing human-readable and effective adversarial prompts with minimal need for human intervention, thereby presenting a more scalable and comprehensive solution for identifying and mitigating the safety vulnerabilities of LLMs.

cs.CL↗

arXiv2023

Federated Experiment Design under Distributed Differential Privacy

Experiment design has a rich history dating back over a century and has found many critical applications across various fields since then. The use and collection of users' data in experiments often involve sensitive personal information, so additional measures to protect individual privacy are required during data collection, storage, and usage. In this work, we focus on the rigorous protection of users' privacy (under the notion of differential privacy (DP)) while minimizing the trust toward service providers. Specifically, we consider the estimation of the average treatment effect (ATE) under DP, while only allowing the analyst to collect population-level statistics via secure aggregation, a distributed protocol enabling a service provider to aggregate information without accessing individual data. Although a vital component in modern A/B testing workflows, private distributed experimentation has not previously been studied. To achieve DP, we design local privatization mechanisms that are compatible with secure aggregation and analyze the utility, in terms of the width of confidence intervals, both asymptotically and non-asymptotically. We show how these mechanisms can be scaled up to handle the very large number of participants commonly found in practice. In addition, when introducing DP noise, it is imperative to cleverly split privacy budgets to estimate both the mean and variance of the outcomes and carefully calibrate the confidence intervals according to the DP noise. Last, we present comprehensive experimental evaluations of our proposed schemes and show the privacy-utility trade-offs in experiment design.

cs.CR↗

arXiv2023

When does the student surpass the teacher? Federated Semi-supervised Learning with Teacher-Student EMA

Semi-Supervised Learning (SSL) has received extensive attention in the domain of computer vision, leading to development of promising approaches such as FixMatch. In scenarios where training data is decentralized and resides on client devices, SSL must be integrated with privacy-aware training techniques such as Federated Learning. We consider the problem of federated image classification and study the performance and privacy challenges with existing federated SSL (FSSL) approaches. Firstly, we note that even state-of-the-art FSSL algorithms can trivially compromise client privacy and other real-world constraints such as client statelessness and communication cost. Secondly, we observe that it is challenging to integrate EMA (Exponential Moving Average) updates into the federated setting, which comes at a trade-off between performance and communication cost. We propose a novel approach FedSwitch, that improves privacy as well as generalization performance through Exponential Moving Average (EMA) updates. FedSwitch utilizes a federated semi-supervised teacher-student EMA framework with two features - local teacher adaptation and adaptive switching between teacher and student for pseudo-label generation. Our proposed approach outperforms the state-of-the-art on federated image classification, can be adapted to real-world constraints, and achieves good generalization performance with minimal communication cost overhead.

cs.LG↗

arXiv2022

The communication cost of security and privacy in federated frequency estimation

We consider the federated frequency estimation problem, where each user holds a private item $X_i$ from a size-$d$ domain and a server aims to estimate the empirical frequency (i.e., histogram) of $n$ items with $n \ll d$. Without any security and privacy considerations, each user can communicate its item to the server by using $\log d$ bits. A naive application of secure aggregation protocols would, however, require $d\log n$ bits per user. Can we reduce the communication needed for secure aggregation, and does security come with a fundamental cost in communication? In this paper, we develop an information-theoretic model for secure aggregation that allows us to characterize the fundamental cost of security and privacy in terms of communication. We show that with security (and without privacy) $\Omega\left( n \log d \right)$ bits per user are necessary and sufficient to allow the server to compute the frequency distribution. This is significantly smaller than the $d\log n$ bits per user needed by the naive scheme, but significantly higher than the $\log d$ bits per user needed without security. To achieve differential privacy, we construct a linear scheme based on a noisy sketch which locally perturbs the data and does not require a trusted server (a.k.a. distributed differential privacy). We analyze this scheme under $\ell_2$ and $\ell_\infty$ loss. By using our information-theoretic framework, we show that the scheme achieves the optimal accuracy-privacy trade-off with optimal communication cost, while matching the performance in the centralized case where data is stored in the central server.

cs.IT↗

arXiv2021

Sample and Threshold Differential Privacy: Histograms and applications

Federated analytics seeks to compute accurate statistics from data distributed across users' devices while providing a suitable privacy guarantee and being practically feasible to implement and scale. In this paper, we show how a strong $(\epsilon, \delta)$-Differential Privacy (DP) guarantee can be achieved for the fundamental problem of histogram generation in a federated setting, via a highly practical sampling-based procedure that does not add noise to disclosed data. Given the ubiquity of sampling in practice, we thus obtain a DP guarantee almost for free, avoid over-estimating histogram counts, and allow easy reasoning about how privacy guarantees may obscure minorities and outliers. Using such histograms, related problems such as heavy hitters and quantiles can be answered with provable error and privacy guarantees. Experimental results show that our sample-and-threshold approach is accurate and scalable.

cs.CR↗

arXiv2021

Opacus: User-Friendly Differential Privacy Library in PyTorch

We introduce Opacus, a free, open-source PyTorch library for training deep learning models with differential privacy (hosted at opacus.ai). Opacus is designed for simplicity, flexibility, and speed. It provides a simple and user-friendly API, and enables machine learning practitioners to make a training pipeline private by adding as little as two lines to their code. It supports a wide variety of layers, including multi-head attention, convolution, LSTM, GRU (and generic RNN), and embedding, right out of the box and provides the means for supporting other user-defined layers. Opacus computes batched per-sample gradients, providing higher efficiency compared to the traditional "micro batch" approach. In this paper we present Opacus, detail the principles that drove its implementation and unique features, and benchmark it against other frameworks for training models with differential privacy as well as standard PyTorch.

cs.LG↗

arXiv2020

To Test Machine Comprehension, Start by Defining Comprehension

Many tasks aim to measure machine reading comprehension (MRC), often focusing on question types presumed to be difficult. Rarely, however, do task designers start by considering what systems should in fact comprehend. In this paper we make two key contributions. First, we argue that existing approaches do not adequately define comprehension; they are too unsystematic about what content is tested. Second, we present a detailed definition of comprehension -- a "Template of Understanding" -- for a widely useful class of texts, namely short narratives. We then conduct an experiment that strongly suggests existing systems are not up to the task of narrative understanding as we define it.

cs.CL↗