Search for: All records

In this paper, we propose a new secure machine learning inference platform assisted by a small dedicated security processor, which will be easier to protect and deploy compared to today's TEEs integrated into high-performance processors. Our platform provides three main advantages over the state-of-the-art: (i) We achieve significant performance improvements compared to state-of-the-art distributed Privacy-Preserving Machine Learning (PPML) protocols, with only a small security processor that is comparable to a discrete security chip such as the Trusted Platform Module (TPM) or on-chip security subsystems in SoCs similar to the Apple enclave processor. In the semi-honest setting with WAN/GPU, our scheme is 4X-63X faster than Falcon (PoPETs'21) and AriaNN (PoPETs'22) and 3.8X-12X more communication efficient. We achieve even higher performance improvements in the malicious setting. (ii) Our platform guarantees security with abort against malicious adversaries under honest majority assumption. (iii) Our technique is not limited by the size of secure memory in a TEE and can support high-capacity modern neural networks like ResNet18 and Transformer. While previous work investigated the use of high-performance TEEs in PPML, this work represents the first to show that even tiny secure hardware with very limited performance can be leveraged to significantly speed-up distributed PPML protocols if the protocol can be carefully designed for lightweight trusted hardware. 

Abstract Language Models (LM) have been extensively utilized for learning DNA sequence patterns and generating synthetic sequences. In this paper, we present a novel approach for the generation of synthetic DNA data using pangenomes in combination with LM. We introduce three innovative pangenome-based tokenization schemes, including two that can decouple from private data, while enhance long DNA sequence generation. Our experimental results demonstrate the superiority of pangenome-based tokenization over classical methods in generating high-utility synthetic DNA sequences, highlighting a promising direction for the public sharing of genomic datasets. 

We created GNQA, a generative pre-trained transformer (GPT) knowledge base driven by a performant retrieval augmented generation (RAG) with a focus on aging, dementia, Alzheimer’s and diabetes. We uploaded a corpus of three thousand peer reviewed publications on these topics into the RAG. To address concerns about inaccurate responses and GPT ‘hallucinations’, we implemented a context provenance tracking mechanism that enables researchers to validate responses against the original material and to get references to the original papers. To assess the effectiveness of contextual information we collected evaluations and feedback from both domain expert users and ‘citizen scientists’ on the relevance of GPT responses. A key innovation of our study is automated evaluation by way of a RAG assessment system (RAGAS). RAGAS combines human expert assessment with AI-driven evaluation to measure the effectiveness of RAG systems. When evaluating the responses to their questions, human respondents give a “thumbs-up” 76% of the time. Meanwhile, RAGAS scores 90% on answer relevance on questions posed by experts. And when GPT-generates questions, RAGAS scores 74% on answer relevance. With RAGAS we created a benchmark that can be used to continuously assess the performance of our knowledge base. Full GNQA functionality is embedded in the freeGeneNetwork.orgweb service, an open-source system containing over 25 years of experimental data on model organisms and human. The code developed for this study is published under a free and open-source software license athttps://git.genenetwork.org/gn-ai/tree/README.md. 

Storage-as-a-service (STaaS) permits the client to outsource her data to the cloud, thereby reducing data management and maintenance costs. However, STaaS also brings significant data integrity and soundness concerns since the storage provider might not keep the client data intact and retrievable all the time (e.g., cost saving via deletions). Proof of Retrievability (PoR) can validate the integrity and retrievability of remote data effectively. This technique can be useful for regular audits to monitor data compromises, as well as to comply with standard data regulations. In particular, cold storage applications (e.g., MS Azure, Amazon Glacier) require regular and frequent audits with less frequent data modification. Yet, despite their merits, existing PoR techniques generally focus on other metrics (e.g., low storage, fast update, metadata privacy) but not audit efficiency (e.g., low audit time, small proof size). Hence, there is a need to develop new PoR techniques that achieve efficient data audit while preserving update and retrieval performance. In this paper, we propose Porla, a new PoR framework that permits efficient data audit, update, and retrieval functionalities simultaneously. Porla permits data audit in both private and public settings, each of which features asymptotically (and concretely) smaller audit-proof size and lower audit time than all the prior works while retaining the same asymptotic data update overhead. Porla achieves all these properties by composing erasure codes with verifiable computation techniques which, to our knowledge, is a new approach to PoR design. We address several challenges that arise in such a composition by creating a new homomorphic authenticated commitment scheme, which can be of independent interest. We fully implemented Porla and evaluated its performance on commodity cloud (i.e., Amazon EC2) under various settings. Experimental results demonstrated that Porla achieves two to four orders of magnitude smaller audit proof size with 4x–18000x lower audit time than all prior schemes in both private and public audit settings at the cost of only 2x–3x slower update. 

Channel charting (CC) has been proposed recently to enable logical positioning of user equipments (UEs) in the neighborhood of a multi-antenna base-station solely from channel-state information (CSI). CC relies on dimensionality reduction of high-dimensional CSI features in order to construct a channel chart that captures spatial and radio geometries so that UEs close in space are close in the channel chart. In this paper, we demonstrate that autoencoder (AE)-based CC can be augmented with side information that is obtained during the CSI acquisition process. More specifically, we propose to include pairwise representation constraints into AEs with the goal of improving the quality of the learned channel charts. We show that such representation-constrained AEs recover the global geometry of the learned channel charts, which enables CC to perform approximate positioning without global navigation satellite systems or supervised learning methods that rely on extensive and expensive measurement campaigns.