Publications
Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.
Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.
Sort By
1 - 15 of 10064 publications
Assessing Web Fingerprinting Risk
Robert Busa-Fekete
Antonio Sartori
Proceedings of the ACM Web Conference (WWW 2024)
Preview abstract
Modern Web APIs allow developers to provide extensively customized experiences for website visitors, but the richness of the device information they provide also make them vulnerable to being abused by malign actors to construct browser fingerprints, device-specific identifiers that enable covert tracking of users even when cookies are disabled.
Previous research has established entropy, a measure of information, as the key metric for quantifying fingerprinting risk. Earlier studies that estimated the entropy of Web APIs were based on data from a single website or were limited to an extremely small sample of clients. They also analyzed each Web API separately and then summed their entropies to quantify overall fingerprinting risk, an approach that can lead to gross overestimates.
We provide the first study of browser fingerprinting which addresses the limitations of prior work. Our study is based on actual visited pages and Web API function calls reported by tens of millions of real Chrome browsers in-the-wild. We accounted for the dependencies and correlations among Web APIs, which is crucial for obtaining more realistic entropy estimates. We also developed a novel experimental design that accurately estimates entropy while never observing too much information from any single user. Our results provide an understanding of the distribution of entropy for different website categories, confirm the utility of entropy as a fingerprinting proxy, and offer a method for evaluating browser enhancements which are intended to mitigate fingerprinting.
View details
Visual Program Tuning: Training Large Multimodal Models to Reason like Programs
Yushi Hu
Krishna Viswanathan
Kenji Hata
Enming Luo
Ranjay Krishna
Ariel Fuxman
Conference on Computer Vision and Pattern Recognition (2024)
Preview abstract
Solving complex visual tasks (e.g., “Who invented the musical instrument on the right?”) involves back-and-forth between visual processing and reasoning. Visual programming is a recent multimodal framework that has shown promise in conducting visual reasoning in an interpretable and compositional manner. However, this framework is error-prone—it can lead to a wrong answer whenever the program itself is wrong, or when any of the steps of the program are solved incorrectly, thus leading to worse overall performance than end-to-end systems trained with labeled data. Moreover, it is inefficient to involve multiple steps (i.e., generating and then running programs) during inference. Ideally, a single large multimodal model (LMM) should directly conduct similar reasoning and yield the correct answer.
In this work, we propose Visual Program Tuning (VPT), which leverages visual programs for teaching LLMs to reason via instruction tuning. VPT rewrites the execution traces of visual programs as chain-of-thought reasoning steps, and tunes an LMM to output not only the label but its reasoning as well. Extensive experiments on complex vision tasks show that models trained with VPT achieve state-of-the-art accuracy while being able to produce interpretable and faithful reasoning steps. PaLI-X + VPT outperforms all existing LMMs on a wide range of visual tasks, improving performance on counting, spatial relations, and compositional reasoning tasks. VPT is also helpful for quick adaptation on new tasks. Our experiments on content moderation show that fine-tuning LMMs with program-augmented examples is more sample efficient than traditional supervised training.
View details
Promises and Pitfalls of Generative Masked Language Modeling: Theoretical Framework and Practical Guidelines
Yuchen Li
Alexandre Kirchmeyer
Aashay Mehta
Yilong Qin
Andrej Risteski
International Conference on Machine Learning (2024) (to appear)
Preview abstract
Autoregressive language models are the currently dominant paradigm for text generation, however they have some fundamental limitations that cannot be remedied by scale ---for example inherently sequential and unidirectional generation. While alternate classes of models have been explored, we have limited mathematical understanding of their fundamental power and limitations. In this paper we focus on Generative Masked Language Models (GMLMs), a non-autoregressive paradigm in which we train a model to fit conditional probabilities of the data distribution via masking, which are subsequently used as inputs to a Markov Chain to draw samples from the model. These models empirically strike a promising speed-quality trade-off as each step can be typically parallelized by decoding the entire sequence in parallel. We develop a mathematical framework for analyzing and improving such models which sheds light on questions of sample complexity and inference speed and quality. Empirically, we adapt the T5 model for iteratively-refined parallel decoding, achieving 2-3x speedup in machine translation with minimal sacrifice in quality compared with autoregressive models. We run careful ablation experiments to give recommendations on key design choices, and make fine-grained observations on the common error modes in connection with our theory. Our mathematical analyses and empirical observations characterize both potentials and limitations of this approach, and can be applied to future works on improving understanding and performance of GMLMs.
View details
Experiencing InstructPipe: Building Multi-modal AI Pipelines via Prompting LLMs and Visual Programming
Zhongyi Zhou
Jing Jin
Xiuxiu Yuan
Jun Jiang
Jingtao Zhou
Yiyi Huang
Kristen Wright
Jason Mayes
Mark Sherwood
Ram Iyengar
Na Li
Extended Abstracts of the 2024 CHI Conference on Human Factors in Computing Systems, ACM, pp. 5
Preview abstract
Foundational multi-modal models have democratized AI access, yet the construction of complex, customizable machine learning pipelines by novice users remains a grand challenge. This paper demonstrates a visual programming system that allows novices to rapidly prototype multimodal AI pipelines. We first conducted a formative study with 58 contributors and collected 236 proposals of multimodal AI pipelines that served various practical needs. We then distilled our findings into a design matrix of primitive nodes for prototyping multimodal AI visual programming pipelines, and implemented a system with 65 nodes. To support users' rapid prototyping experience, we built InstructPipe, an AI assistant based on large language models (LLMs) that allows users to generate a pipeline by writing text-based instructions. We believe InstructPipe enhances novice users onboarding experience of visual programming and the controllability of LLMs by offering non-experts a platform to easily update the generation.
View details
PROMPT: A Fast and Extensible Memory Profiling Framework
Ziyang Xu
Yebin Chon
Yian Su
Zujun Tan
Simone Campanoni
David I. August
Proceedings of the ACM on Programming Languages, 8, Issue OOPSLA (2024)
Preview abstract
Memory profiling captures programs' dynamic memory behavior, assisting programmers in debugging, tuning, and enabling advanced compiler optimizations like speculation-based automatic parallelization. As each use case demands its unique program trace summary, various memory profiler types have been developed. Yet, designing practical memory profilers often requires extensive compiler expertise, adeptness in program optimization, and significant implementation effort. This often results in a void where aspirations for fast and robust profilers remain unfulfilled. To bridge this gap, this paper presents PROMPT, a framework for streamlined development of fast memory profilers. With PROMPT, developers need only specify profiling events and define the core profiling logic, bypassing the complexities of custom instrumentation and intricate memory profiling components and optimizations. Two state-of-the-art memory profilers were ported with PROMPT where all features preserved. By focusing on the core profiling logic, the code was reduced by more than 65% and the profiling overhead was improved by 5.3× and 7.1× respectively. To further underscore PROMPT's impact, a tailored memory profiling workflow was constructed for a sophisticated compiler optimization client. In 570 lines of code, this redesigned workflow satisfies the client’s memory profiling needs while achieving more than 90% reduction in profiling overhead and improved robustness compared to the original profilers.
View details
Bridging the Preference Gap between Retrievers and LLMs
Zixuan Ke
Qiaozhu Mei
Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (2024) (to appear)
Preview abstract
Large Language Models (LLMs) have demonstrated superior results across a wide range of tasks, and Retrieval-augmented Generation (RAG) is an effective way to enhance the performance by locating relevant information and placing it into the context window of the LLM. However, the relationship between retrievers and LLM in a RAG is still under-investigated. Most existing work treats the retriever and the LLM as independent components and leaves a gap between retrieving human-"friendly" information and assembling a LLM-"friendly" context. In this work, we examine a novel bridge mechanism. We validate the ranking and selection assumptions of retrievers in the context of RAG and propose a framework that chains together supervised and reinforcement learning to train a bridge model that optimizes the connection between the retriever and the LLM. Empirical results demonstrate the effectiveness of our method in both question-answering and personalized generation tasks.
View details
Preview abstract
Japanese text-to-pronunciation modelling is a notoriously data-intensive problem. Japanese data sources are often only partially annotated, and use different annotation standards for pronunciation and word segmentation. This talk introduces a set of techniques that enable ingesting data that may be partially annotated, use arbitrary word segmentations, and use a variety of pronunciation annotation standards.
View details
Preview abstract
The evolution of AI is a pivotal moment in history, but it’s not the first time we have experienced technological advances that have changed how humans work. By looking at the advances in automobiles, we are reminded of the importance of focusing on our developers' needs and goals.
View details
Preview abstract
One of the most basic problems for studying the "price of privacy over time" is the so called private counter problem, introduced by Dwork et al. (2010) and Chan et al. (2011). In this problem, we aim to track the number of events that occur over time, while hiding the existence of every single event. More specifically, in every time step $t\in[T]$ we learn (in an online fashion) that $\Delta_t\geq 0$ new events have occurred, and must respond with an estimate $n_t\approx\sum_{j=1}^t \Delta_j$. The privacy requirement is that all of the outputs together, across all time steps, satisfy event level differential privacy.
The main question here is how our error needs to depend on the total number of time steps $T$ and the total number of events $n$. Dwork et al. (2015) showed an upper bound of $O\left(\log(T)+\log^2(n)\right)$, and Henzinger et al. (2023) showed a lower bound of $\Omega\left(\min\{\log n, \log T\}\right)$. We show a new lower bound of $\Omega\left(\min\{n,\log T\}\right)$, which is tight w.r.t. the dependence on $T$, and is tight in the sparse case where $\log^2 n=O(\log T)$. Our lower bound has the following implications:
* We show that our lower bound extends to the online thresholds problem, where the goal is to privately answer many "quantile queries" when these queries are presented one-by-one. This resolves an open question of Bun et al. (2017).
* Our lower bound implies, for the first time, a separation between the number of mistakes obtainable by a private online learner and a non-private online learner. This partially resolves a COLT'22 open question published by Sanyal and Ramponi.
* Our lower bound also yields the first separation between the standard model of private online learning and a recently proposed relaxed variant of it, called private online prediction.
View details
The effect of uncertainty in humidity and model parameters on the prediction of contrail energy forcing
Marc Shapiro
Zebediah Engberg
Tharun Sankar
Marc E.J. Stettler
Roger Teoh
Ulrich Schumann
Susanne Rohs
Erica Brand
Environmental Research Communications, 6 (2024), pp. 095015
Preview abstract
Previous work has shown that while the net effect of aircraft condensation trails (contrails) on the climate is warming, the exact magnitude of the energy forcing per meter of contrail remains uncertain. In this paper, we explore the skill of a Lagrangian contrail model (CoCiP) in identifying flight segments with high contrail energy forcing. We find that skill is greater than climatological predictions alone, even accounting for uncertainty in weather fields and model parameters. We estimate the uncertainty due to humidity by using the ensemble ERA5 weather reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF) as Monte Carlo inputs to CoCiP. We unbias and correct under-dispersion on the ERA5 humidity data by forcing a match to the distribution of in situ humidity measurements taken at cruising altitude. We take CoCiP energy forcing estimates calculated using one of the ensemble members as a proxy for ground truth, and report the skill of CoCiP in identifying segments with large positive proxy energy forcing. We further estimate the uncertainty due to model parameters in CoCiP by performing Monte Carlo simulations with CoCiP model parameters drawn from uncertainty distributions consistent with the literature. When CoCiP outputs are averaged over seasons to form climatological predictions, the skill in predicting the proxy is 44%, while the skill of per-flight CoCiP outputs is 84%. If these results carry over to the true (unknown) contrail EF, they indicate that per-flight energy forcing predictions can reduce the number of potential contrail avoidance route adjustments by 2x, hence reducing both the cost and fuel impact of contrail avoidance.
View details
Unsupervised representation learning on high-dimensional clinical data improves genomic discovery and prediction
Babak Behsaz
Zachary Ryan Mccaw
Davin Hill
Robert Luben
Dongbing Lai
John Bates
Howard Yang
Tae-Hwi Schwantes-An
Yuchen Zhou
Anthony Khawaja
Andrew Carroll
Brian Hobbs
Michael Cho
Nature Genetics (2024)
Preview abstract
Although high-dimensional clinical data (HDCD) are increasingly available in biobank-scale datasets, their use for genetic discovery remains challenging. Here we introduce an unsupervised deep learning model, Representation Learning for Genetic Discovery on Low-Dimensional Embeddings (REGLE), for discovering associations between genetic variants and HDCD. REGLE leverages variational autoencoders to compute nonlinear disentangled embeddings of HDCD, which become the inputs to genome-wide association studies (GWAS). REGLE can uncover features not captured by existing expert-defined features and enables the creation of accurate disease-specific polygenic risk scores (PRSs) in datasets with very few labeled data. We apply REGLE to perform GWAS on respiratory and circulatory HDCD—spirograms measuring lung function and photoplethysmograms measuring blood volume changes. REGLE replicates known loci while identifying others not previously detected. REGLE are predictive of overall survival, and PRSs constructed from REGLE loci improve disease prediction across multiple biobanks. Overall, REGLE contain clinically relevant information beyond that captured by existing expert-defined features, leading to improved genetic discovery and disease prediction.
View details
Limoncello: Prefetchers for Scale
Carlos Villavieja
Baris Kasikci
Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Association for Computing Machinery, New York, NY, United States (2024)
Preview abstract
This paper presents Limoncello, a novel software system that dynamically configures data prefetching for high utilization systems. We demonstrate that in resource-constrained environments, such as large data centers, traditional methods of hardware prefetching can increase memory latency and decrease available memory bandwidth. To address this, Limoncello dynamically configures data prefetching, disabling hardware prefetchers when memory bandwidth utilization is high and leveraging targeted software prefetching to reduce cache misses when hardware prefetchers are disabled. Limoncello is software-centric and does not require any modifications to hardware. Our evaluation of the deployment on a real-world hyperscale system reveals that Limoncello unlocks significant performance gains for high-utilization systems: it improves application throughput by 10%, due to a 15% reduction in memory latency, while maintaining minimal change in cache miss rate for targeted library functions.
View details
Preview abstract
Proving mathematical theorems at the olympiad level represents a notable milestone in human-level automated reasoning, owing to their reputed difficulty among the world’s best talents in pre-university mathematics. Current machine-learning approaches, however, are not applicable to most mathematical domains owing to the high cost of translating human proofs into machine-verifiable format. The problem is even worse for geometry because of its unique translation challenges, resulting in severe scarcity of training data. We propose AlphaGeometry, a theorem prover for Euclidean plane geometry that sidesteps the need for human demonstrations by synthesizing millions of theorems and proofs across different levels of complexity. AlphaGeometry is a neuro-symbolic system that uses a neural language model, trained from scratch on our large-scale synthetic data, to guide a symbolic deduction engine through infinite branching points in challenging problems. On a test set of 30 latest olympiad-level problems, AlphaGeometry solves 25, outperforming the previous best method that only solves ten problems and approaching the performance of an average International Mathematical Olympiad (IMO) gold medallist. Notably, AlphaGeometry produces human-readable proofs, solves all geometry problems in the IMO 2000 and 2015 under human expert evaluation and discovers a generalized version of a translated IMO theorem in 2004.
View details
SAC126 - DNSSEC Delegation Signer (DS) Record Automation
Internet Corporation for Assigned Names and Numbers (ICANN), ICANN Security and Stability Advisory Committee (SSAC) Reports and Advisories (2024), pp. 39
Preview abstract
The deployment of Domain Name System (DNS) Security Extensions (DNSSEC) has been
hindered by a number of obstacles. This report focuses on one: the management of Delegation
Signer (DS) records, which connect a child zone’s DNSSEC public key and signatures to the
chain of trust provided by its parent zone (e.g., a zone corresponding to a top-level domain).
DNSSEC is not simply enabled by signing a delegated domain’s DNS zone with DNSSEC
signatures. It is also necessary to configure (and later maintain) appropriate DS records, which
involves coordinated actions by the DNS operator, registrant, registrar, and registry.
In the case where the domain’s DNS service is operated by the registrar, this process can be
reduced to a simple internal operation by the registrar. If the functions are separated, this is not
possible. This report is therefore focused on when the domain’s DNS service is not operated by
the registrar, but by a third-party DNS operator.
In such a scenario, current practice holds the registrant responsible for coordinating DS
maintenance. The registrant (or someone appointed by them) needs to first obtain DNSSEC
public key parameters from the DNS operator, and convey these parameters to the registrar
(potentially via a reseller). The registrar will then need to relay these DNSSEC public key
parameters to the registry, who will use them to create and publish the DS record in the parent
zone. This process often involves idiosyncratic interfaces for each combination of DNS operator
and registrar, requiring a level of engagement and time investment, awareness, and
understanding that often do not match with what the registrant knows or expects. The complexity of the process further introduces opportunity for error.
This can be alleviated by employing automation for the data exchanges required for DS
maintenance so that, when the domain’s DNS service is operated by a third party, registries or
registrars can, without human involvement, obtain all information needed for keeping DS records up to date. Various approaches to achieve this are possible, such as a scheme where the registry or registrar actively contacts the Child DNS operator, or vice versa. The different approaches come with different challenges with respect to authentication, timing, and efficiency.
The IETF has standardized specifications around the first approach, where the parent pulls
information from the Child DNS operator, and operational experience has been gained over
recent years. However, some standardization gaps remain (such as to improve efficiency and
error handling). In addition, the industry could benefit from further development of best practices in deploying the technology.
The SSAC believes that automated DS maintenance should be a goal for the domain name
industry. To make this a reality, the SSAC makes several recommendations with the goal to spur
industry players and ICANN towards an industry best practice for DNSSEC DS automation.
View details
Measuring Developer Experience with a Longitudinal Survey
Jessica Lin
Jill Dicker
IEEE Software (2024)
Preview abstract
At Google, we’ve been running a quarterly large-scale survey with developers since 2018. In this article, we will discuss how we run EngSat, some of our key learnings over the past 6 years, and how we’ve evolved our approach to meet new needs and challenges.
View details
