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 10129 publications
Preview abstract
This paper presents NOMAD (Non-Matching Audio Distance), a differentiable perceptual similarity metric that measures the distance of a degraded signal against non-matching references. The proposed method is based on learning deep feature embeddings via a triplet loss guided by the Neurogram Similarity Index Measure (NSIM) to capture degradation intensity. During inference, the similarity score between any two audio samples is computed through Euclidean distance of their embeddings. NOMAD is fully unsupervised and can be used in general perceptual audio tasks for audio analysis e.g. quality assessment and generative tasks such as speech enhancement and speech synthesis. The proposed method is evaluated with 3 tasks. Ranking degradation intensity, predicting speech quality, and as a loss function for speech enhancement. Results indicate NOMAD outperforms other non-matching reference approaches in both ranking degradation intensity and quality assessment, exhibiting competitive performance with full-reference audio metrics. NOMAD demonstrates a promising technique that mimics human capabilities in assessing audio quality with non-matching references to learn perceptual embeddings without the need for human-generated labels.
View details
The Case for Globalizing Fairness: A Mixed Methods Study on the Perceptions of Colonialism, AI and Health in Africa
Iskandar Haykel
Aisha Walcott-Bryant
Sanmi Koyejo
Preview abstract
With growing machine learning (ML) and large language model applications in healthcare, there have been calls for fairness in ML to understand and mitigate ethical concerns these systems may pose. Fairness has implications for health in Africa, which already has inequitable power imbalances between the Global North and South. This paper seeks to explore fairness for global health, with Africa as a case study.
We conduct a scoping review to propose fairness attributes for consideration in the African context and delineate where they may come into play in different ML-enabled medical modalities. We then conduct qualitative research studies with 625 general population study participants in 5 countries in Africa and 28 experts in ML, Health, and/or policy focussed on Africa to obtain feedback on the proposed attributes. We delve specifically into understanding the interplay between AI, health and colonialism.
Our findings demonstrate that among experts there is a general mistrust that technologies that are solely developed by former colonizers can benefit Africans, and that associated resource constraints due to pre-existing economic and infrastructure inequities can be linked to colonialism. General population survey responses found about an average of 40% of people associate an undercurrent of colonialism to AI and this was most dominant amongst participants from South Africa. However the majority of the general population participants surveyed did not think there was a direct link between AI and colonialism.Colonial history, country of origin, National income level were specific axes of disparities that participants felt would cause an AI tool to be biased
This work serves as a basis for policy development around Artificial Intelligence for health in Africa and can be expanded to other regions.
View details
Preview abstract
We introduce SynCLR, a novel approach for learning visual representations exclusively from synthetic images and synthetic captions, without any real data. We synthesize a large dataset of image captions using LLMs, then use an off-the-shelf text-to-image model to generate multiple images corresponding to each synthetic caption. We perform visual representation learning on these synthetic images via contrastive learning, treating images sharing the same caption as positive pairs. The resulting representations transfer well to many downstream tasks, competing favorably with other general-purpose visual representation learners such as CLIP and DINO v2 in image classification tasks. Furthermore, in dense prediction tasks such as semantic segmentation, SynCLR outperforms previous self-supervised methods by a significant margin, e.g., improving over MAE and iBOT by 6.2 and 4.3 mIoU on ADE20k for ViT-B/16.
View details
Rethinking FID: Towards a Better Evaluation Metric for Image Generation
Sadeep Jayasumana
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2024)
Preview abstract
As with many machine learning problems, the progress of image generation methods hinges on good evaluation metrics. One of the most popular is the Frechet Inception Distance (FID). FID estimates the distance between a distribution of Inception-v3 features of real images, and those of images generated by the algorithm. We highlight important drawbacks of FID: Inception's poor representation of the rich and varied content generated by modern text-to-image models, incorrect normality assumptions, and poor sample complexity. We call for a reevaluation of FID's use as the primary quality metric for generated images. We empirically demonstrate that FID contradicts human raters, it does not reflect gradual improvement of iterative text-to-image models, it does not capture distortion levels, and that it produces inconsistent results when varying the sample size. We also propose an alternative new metric, CMMD, based on richer CLIP embeddings and the maximum mean discrepancy distance with the Gaussian RBF kernel. It is an unbiased estimator that does not make any assumptions on the probability distribution of the embeddings and is sample efficient. Through extensive experiments and analysis, we demonstrate that FID-based evaluations of text-to-image models may be unreliable, and that CMMD offers a more robust and reliable assessment of image quality.
View details
Preview abstract
Computing efficient traffic signal plans is often based on the amount of traffic in an intersection, its distribution over the various intersection movements and hours as well as on performance metrics such as traffic delay. In their simple and typical form plans are fixed in the same hour over weekdays. This allows low operation costs without the necessity for traffic detection and monitoring tools. A critical factor on the potential efficiency of such plans is the similarity of traffic patterns over the days along each of the intersection movements. We refer to such similarity as the traffic stability of the intersection and define simple metrics to measure it based on traffic volume and traffic delay. In this paper, we propose an automatic probe data based method, for city-wide estimation of traffic stability. We discuss how such measures can be used for signal planning such as in selecting plan resolution or as an indication as which intersections can benefit from dynamic but expensive traffic detection tools. We also identify events of major changes in traffic characteristics of an intersection. We demonstrate the framework by using real traffic statistics to study the traffic stability in the city of Haifa along its 162 intersections. We study the impact of the time of day on the stability, detect major changes in traffic and find intersections with high and low stability.
View details
Preview abstract
What is it to explain the outputs of an opaque machine learning model? Popular strategies in the literature are to develop
explainable machine learning techniques. These techniques approximate how the model works by providing local or global
information about the inner workings of a machine learning model. In this paper, we argue that, in some cases, explaining
machine learning outputs requires appealing to the third kind of explanation that we call socio-structural explanations.
The importance of socio-structural explanations is motivated by the observation that machine learning models are not
autonomous mathematico-computational entities. Instead, their very existence is intrinsically tied to the social context in
which they operate. Sometimes, the social structures are mirrored in the design and training of machine learning models
and hence appealing to the socio-structural explanations offers the relevant explanation for why the output is obtained.
By thoroughly examining a well-known case of racially biased algorithmic resource allocation in healthcare, we highlight
the significance of socio-structural explanations. One ramification of our proposal is that to understand how machine
learning models perpetuate unjust social harms, more is needed to interpret them by model interpretability methods.
Instead, providing socio-structural explanations adds explanatory adequacy as to how and why machine learning outputs
are obtained
View details
Prompt Cache: Modular Attention Reuse for Low Latency Inference
Anurag Khandelwal
Guojun Chen
In Gim
Lin Zhong
Seung-seob Lee
Νikhil Sarda
MLSys (2024)
Preview abstract
We present Prompt Cache, an approach for accelerating inference for large language models (LLM) by reusing
attention states across different LLM prompts. Many input prompts have overlapping text segments, such as
system messages, prompt templates, and documents provided for context. Our key insight is that by precomputing
and storing the attention states of these frequently occurring text segments on the inference server, we can
efficiently reuse them when these segments appear in user prompts. Prompt Cache employs a schema to explicitly
define such reusable text segments, called prompt modules. The schema ensures positional accuracy during
attention state reuse and provides users with an interface to access cached states in their prompt. Using a prototype
implementation, we evaluate Prompt Cache across several LLMs. We show that Prompt Cache significantly reduce
latency in time-to-first-token, especially for longer prompts such as document-based question answering and
recommendations. The improvements range from 8× for GPU-based inference to 60× for CPU-based inference,
all while maintaining output accuracy and without the need for model parameter modifications.
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
Its All Relative! -- A Synthetic Query Generation Approach for Improving Zero-Shot Relevance Prediction
Findings of the Association for Computational Linguistics: NAACL 2024
Preview abstract
Recent developments in large language models (LLMs) have shown promise in their ability to generate synthetic query-document pairs by prompting LLMs with as few as 8 demonstrations \cite{dai2022promptagator}.
This has enabled building better IR models especially for tasks which have no training data readily available.
Typically, such synthetic query generation (QGen) approaches condition on an input context (e.g. document) and generate a query that is relevant to that context or condition the QGen model additionally on the relevance label (e.g. relevant vs irrelevant) to generate queries across relevance buckets.
However, we find that such QGen approaches are sub-optimal as it requires the model to reason about the desired label and the input from only a handful of examples, which is not trivial, especially when the relevance buckets are nuanced.
In this work, we propose to reduce this burden of LLMs by generating queries simultaneously for different labels (e.g. relevance buckets).
We hypothesize that instead of asking the model to generate, say, an irrelevant query given an input context, asking the model to generate an irrelevant query with respect to a relevant query is a much simpler task setup for the model to reason about.
Extensive experimentation across seven IR datasets shows that synthetic queries generated in such a fashion translates to a better downstream performance, suggesting that the generated queries are indeed of higher quality.
View details
The Case for Validating Inputs in Software-Defined WANs
Rishabh Iyer
Isaac Keslassy
Sylvia Ratnasamy
The 23rd ACM Workshop on Hot Topics in Networks (HOTNETS ’24), ACM, Irvine, CA (2024) (to appear)
Preview abstract
We highlight a problem that the networking community has
largely overlooked: ensuring that the inputs to network controllers in software-
defined WANs are accurate. We we show that “incorrect” inputs are a common
cause of major outages in practice and propose new directions to address these.
View details
SkipWriter: LLM-Powered Abbreviated Writing on Tablets
Zheer Xu
Mukund Varma T
Proceedings of UIST 2024 (2024)
Preview abstract
Large Language Models (LLMs) may offer transformative opportunities for text input, especially for physically demanding modalities like handwriting. We studied a form of abbreviated handwriting by designing, developing and evaluating a prototype, named SkipWriter, that convert handwritten strokes of a variable-length, prefix- based abbreviation (e.g., “ho a y” as handwritten strokes) into the intended full phrase (e.g., “how are you” in the digital format) based
on preceding context. SkipWriter consists of an in-production hand-writing recognizer and a LLM fine-tuned on this skip-writing task. With flexible pen input, SkipWriter allows the user to add and revise prefix strokes when predictions don’t match the user’s intent. An user evaluation demonstrated a 60% reduction in motor movements with an average speed of 25.78 WPM. We also showed that this reduction is close to the ceiling of our model in an offline simulation.
View details
A Toolbox for Surfacing Health Equity Harms and Biases in Large Language Models
Heather Cole-Lewis
Nenad Tomašev
Liam McCoy
Leo Anthony Celi
Alanna Walton
Akeiylah DeWitt
Philip Mansfield
Sushant Prakash
Joelle Barral
Ivor Horn
Karan Singhal
Nature Medicine (2024)
Preview abstract
Large language models (LLMs) hold promise to serve complex health information needs but also have the potential to introduce harm and exacerbate health disparities. Reliably evaluating equity-related model failures is a critical step toward developing systems that promote health equity. We present resources and methodologies for surfacing biases with potential to precipitate equity-related harms in long-form, LLM-generated answers to medical questions and conduct a large-scale empirical case study with the Med-PaLM 2 LLM. Our contributions include a multifactorial framework for human assessment of LLM-generated answers for biases and EquityMedQA, a collection of seven datasets enriched for adversarial queries. Both our human assessment framework and our dataset design process are grounded in an iterative participatory approach and review of Med-PaLM 2 answers. Through our empirical study, we find that our approach surfaces biases that may be missed by narrower evaluation approaches. Our experience underscores the importance of using diverse assessment methodologies and involving raters of varying backgrounds and expertise. While our approach is not sufficient to holistically assess whether the deployment of an artificial intelligence (AI) system promotes equitable health outcomes, we hope that it can be leveraged and built upon toward a shared goal of LLMs that promote accessible and equitable healthcare.
View details
SEEDS: Emulation of Weather Forecast Ensembles with Diffusion Models
John Anderson
Science Advances, 10 (2024), eadk4489
Preview abstract
Probabilistic forecasting is crucial to decision-making under uncertainty about future weather. The dominant approach is to use an ensemble of forecasts to represent and quantify uncertainty in operational numerical weather prediction. However, generating ensembles is computationally costly. In this paper, we propose to generate ensemble forecasts at scale by leveraging recent advances in generative artificial intelligence. Our approach learns a data-driven probabilistic diffusion model from the 5-member ensemble GEFS reforecast dataset. The model can then be sampled efficiently to produce realistic weather forecasts, conditioned on a few members of the operational GEFS forecasting system. The generated ensembles have similar predictive skill as the full GEFS 31-member ensemble, evaluated against ERA5 reanalysis, and emulate well the statistics of large physics-based ensembles. We also apply the same methodology to developing a diffusion model for generative post-processing: the model directly learns to correct biases present in the emulated forecasting system by leveraging reanalysis data as labels during training. Ensembles from this generative post-processing model show greater reliability and accuracy, particularly in extreme event classification. In general, they are more reliable and forecast the probability of extreme weather more accurately than the GEFS operational ensemble. Our models achieve these results at less than 1/10th of the computational cost incurred by the operational GEFS system.
View details
PaLI-X: On Scaling up a Multilingual Vision and Language Model
Josip Djolonga
Piotr Padlewski
Basil Mustafa
Carlos Riquelme
Sebastian Goodman
Yi Tay
Siamak Shakeri
Daniel Salz
Michael Tschannen
Mandar Joshi
Filip Pavetić
Gang Li
Anurag Arnab
Yuanzhong Xu
Keran Rong
Neil Houlsby
Computer Vision and Pattern Recognition Conference (CVPR) (2024)
Preview abstract
We explore the boundaries of scaling up a multilingual vision and language model, both in terms of size of the components and the breadth of its training task mixture. Our model achieves new levels of performance on a wide-range of varied and complex tasks, including multiple image-based captioning and question-answering tasks, image-based document understanding and few-shot (in-context) learning, as well as object detection, video question answering, and video captioning. Our model advances the state-of-the-art on most vision-and-language benchmarks considered (20+ of them). Finally, we observe emerging capabilities, such as complex counting and multilingual object detection, tasks that are not explicitly in the training mix.
View details
PriorBoost: An Adaptive Algorithm for Learning from Aggregate Responses
Adel Javanmard
Proceedings of the 41st International Conference on Machine Learning (2024), pp. 21410-21429
Preview abstract
This work studies algorithms for learning from aggregate responses. We focus on the construction of aggregation sets (called \emph{bags} in the literature) for event-level loss functions. We prove for linear regression and generalized linear models (GLMs) that the optimal bagging problem reduces to one-dimensional size-constrained $k$-means clustering. Further, we theoretically quantify the advantage of using curated bags over random bags. We propose the \texttt{PriorBoost} algorithm, which iteratively forms increasingly homogenous bags with respect to (unseen) individual responses to improve model quality. We also explore label differential privacy for aggregate learning, and provide extensive experiments that demonstrate that \PriorBoost regularly achieves optimal quality, in contrast to non-adaptive algorithms for aggregate learning.
View details
