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 10128 publications
Rambler: Supporting Writing With Speech via LLM-Assisted Gist Manipulation
Susan Lin
Jeremy Warner
J.D. Zamfirescu-Pereira
Matthew G Lee
Sauhard Jain
Michael Xuelin Huang
Bjoern Hartmann
Can Liu
Proceedings of the 2024 CHI Conference on Human Factors in Computing Systems, Association for Computing Machinery, New York, NY, USA
Preview abstract
Dictation enables efficient text input on mobile devices. However, writing with speech can produce disfluent, wordy, and incoherent text and thus requires heavy post-processing. This paper presents Rambler, an LLM-powered graphical user interface that supports gist-level manipulation of dictated text with two main sets of functions: gist extraction and macro revision. Gist extraction generates keywords and summaries as anchors to support the review and interaction with spoken text. LLM-assisted macro revisions allow users to respeak, split, merge, and transform dictated text without specifying precise editing locations. Together they pave the way for interactive dictation and revision that help close gaps between spontaneously spoken words and well-structured writing. In a comparative study with 12 participants performing verbal composition tasks, Rambler outperformed the baseline of a speech-to-text editor + ChatGPT, as it better facilitates iterative revisions with enhanced user control over the content while supporting surprisingly diverse user strategies.
View details
Using an LLM to Help With Code Understanding
Daye Nam
Vincent Hellendoorn
Bogdan Vasilescu
Brad A. Myers
ICSE '24: Proceedings of the IEEE/ACM 46th International Conference on Software Engineering (2024)
Preview abstract
Understanding code is challenging, especially when working in new and complex development environments. Code comments and documentation can help, but are typically scarce or hard to navigate. Large language models (LLMs) are revolutionizing the process of writing code. Can they do the same for helping understand it? In this study, we provide a first investigation of an LLM-based conversational UI built directly in the IDE that is geared towards code understanding. Our IDE plugin queries OpenAI's GPT-3.5-turbo model with four high-level requests without the user having to write explicit prompts: to explain a highlighted section of code, provide details of API calls used in the code, explain key domain-specific terms, and provide usage examples for an API. The plugin also allows for open-ended prompts, which are automatically contextualized to the LLM with the program being edited. We evaluate this system in a user study with 32 participants, which confirms that using our plugin can aid task completion more than web search. We additionally provide a thorough analysis of the ways developers use, and perceive the usefulness of, our system, among others finding that the usage and benefits differ between students and professionals. We conclude that in-IDE prompt-less interaction with LLMs is a promising future direction for tool builders.
View details
Using Early Readouts to Mediate Featural Bias in Distillation
Rishabh Tiwari
Durga Sivasubramanian
Anmol Mekala
Ganesh Ramakrishnan
WACV 2024 (2024)
Preview abstract
Deep networks tend to learn spurious feature-label correlations in real-world supervised learning tasks. This vulnerability is aggravated in distillation, where a (student) model may have less representational capacity than the corresponding teacher model. Often, knowledge of specific problem features is used to reweight instances & rebalance the learning process. We propose a novel early readout mechanism whereby we attempt to predict the label using representations from earlier network layers. We show that these early readouts automatically identify problem instances or groups in the form of confident, incorrect predictions. We improve group fairness measures across benchmark datasets by leveraging these signals to mediate between teacher logits and supervised label. We extend our results to the closely related but distinct problem of domain generalization, which also critically depends on the quality of learned features. We provide secondary analyses that bring insight into the role of feature learning in supervision and distillation.
View details
Preview abstract
While large, generative, multilingual models are rapidly being developed and deployed, their safety and fairness evaluations primarily hinge on resources collected in the English language and some limited translations. This has been demonstrated to be insufficient, and severely lacking in nuances of unsafe language and stereotypes prevalent in different languages and the geographical pockets they are prevalent in. Gathering these resources, at scale, in varied languages and regions also poses a challenge as it requires expansive sociolinguistic knowledge and can also be prohibitively expensive. We utilize an established methodology of coupling LLM generations with distributed annotations to overcome these gaps and create the resource SeeGULL Multilingual, spanning 20 languages across 23 regions.
View details
How we use GenAI in SRE
CommitConf, Madrid (2024)
Preview abstract
Google services are powered by the largest network of computers in the world. Site Reliabity Engineers (SRE) make sure that the whole stack is cool: datacenters are safe, well provisionedl; we have fallback mechanims, and data integrity; to making sure we design our stack properly, using the right storage, replication and software trade-offs.
Generative AI is a great tool to make us super-effective: having access to tools to generate our most toily configurations, to classify risks and events, to manage large swaths of machines with agents or to automate complex workflows cheaply.
This talk will cover the journey that SRE started years ago to become a truly AI-First discipline and the latest advancements in tooling, practices and workflows.
View details
Preview abstract
Image-Text pretraining on a web-scale image caption dataset has become the default recipe for open vocabulary classification and retrieval models thanks to the success of CLIP and its variants. Several works have also used CLIP features for dense prediction tasks and have shown the emergence of open-set abilities. However, the contrastive objective only focuses on image and text alignment and does not incentivise image feature learning for dense prediction tasks. In this work, we propose the simple addition of local-to-global correspondence learning by self-distillation as an additional objective for contrastive pre-training to propose SILC. We show that distilling local image features from an EMA teacher model significantly improves model performance on tasks including classification, retrieval, and especially segmentation. We further show that SILC scales better with the same training duration compared to the baselines. Our improved SILC sets a new state-of-the-art for zero-shot classification, few shot classification, image retrieval, zero-shot segmentation, and open vocabulary segmentation.
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
AI-powered patching: the future of automated vulnerability fixes
Jan Keller
Jan Nowakowski
Google Security Engineering Technical Report (2024) (to appear)
Preview abstract
As AI continues to advance at rapid speed, so has its ability to unearth hidden security vulnerabilities in all types of software. Every bug uncovered is an opportunity to patch and strengthen code—but as detection continues to improve, we need to be prepared with new automated solutions that bolster our ability to fix those bugs. That’s why our Secure AI Framework (SAIF) includes a fundamental pillar addressing the need to “automate defenses to keep pace with new and existing threats.”
This paper shares lessons from our experience leveraging AI to scale our ability to fix bugs, specifically those found by sanitizers in C/C++, Java, and Go code. By automating a pipeline to prompt Large Language Models (LLMs) to generate code fixes for human review, we have harnessed our Gemini model to successfully fix 15% of sanitizer bugs discovered during unit tests, resulting in hundreds of bugs patched. Given the large number of sanitizer bugs found each year, this seemingly modest success rate will with time save significant engineering effort. We expect this success rate to continually improve and anticipate that LLMs can be used to fix bugs in various languages across the software development lifecycle.
View details
General Geospatial Inference with a Population Dynamics Foundation Model
Chaitanya Kamath
Shravya Shetty
David Schottlander
Yael Mayer
Joydeep Paul
Jamie McPike
Sheila de Guia
Niv Efron
(2024) (to appear)
Preview abstract
Understanding complex relationships between human behavior and local contexts is crucial for various applications in public health, social science, and environmental studies. Traditional approaches often make use of small sets of manually curated, domain-specific variables to represent human behavior, and struggle to capture these intricate connections, particularly when dealing with diverse data types. To address this challenge, this work introduces a novel approach that leverages the power of graph neural networks (GNNs). We first construct a large dataset encompassing human-centered variables aggregated at postal code and county levels across the United States. This dataset captures rich information on human behavior (internet search behavior and mobility patterns) along with environmental factors (local facility availability, temperature, and air quality). Next, we propose a GNN-based framework designed to encode the connections between these diverse features alongside the inherent spatial relationships between postal codes and their containing counties. We then demonstrate the effectiveness of our approach by benchmarking the model on 27 target variables spanning three distinct domains: health, socioeconomic factors, and environmental measurements. Through spatial interpolation, extrapolation, and super-resolution tasks, we show that the proposed method can effectively utilize the rich feature set to achieve accurate predictions across diverse geospatial domains.
View details
Preview abstract
Specialized Large multi-modal models (LMMs) have exhibited remarkable performance across numerous tasks, however, generalist LMMs suffer from performance degradation when training with a large collection of tasks. Recent research suggests Mixture of Experts (MoE) Models help instruction tuning, however, for LMMs of parameter size around O(50-100B), the prohibitive cost of replicating and storing the expert models severely limits the number of experts we can use.
We propose Omni-SMoLA that softly mixes many multimodal low rank experts to large models without introducing significant new parameter count compared to conventional MoE models. The core idea is that the large model provides a foundational backbone and different lightweight experts learn specialized knowledge residually. Extensive experiments demonstrate that the SMoLA approach helps improve the generalist performance across a broad range of visual question answering and captioning tasks, achieving a new state-of-the-art generalist performance that matches or outperforms single specialized LMM baselines.
View details
SceneFun3D: Fine-Grained Functionality and Affordance Understanding in 3D Scenes
Delitzas Alexandros
Ayça Takmaz
Marc Pollefeys
Francis Engelmann
CVPR (2024) (to appear)
Preview abstract
Existing 3D scene understanding methods are heavily focused on 3D semantic and instance segmentation. However, identifying objects and their parts only constitutes an
intermediate step towards a more fine-grained goal, which is effectively interacting with the functional interactive elements (e.g., handles, knobs, buttons) in the scene to accomplish diverse tasks. To this end, we introduce SceneFun3D, a large-scale dataset with more than 14.8k highly accurate interaction annotations for 710 high-resolution real-world 3D indoor scenes. We accompany the annotations with motion parameter information, describing how to interact with these elements, and a diverse set of natural language descriptions of tasks that involve manipulating them in the scene context. To showcase the value of our dataset, we introduce three novel tasks, namely functionality segmentation, task-driven affordance grounding and 3D motion estimation, and adapt existing state-of-the-art methods to tackle them. Our experiments show that solving these tasks in real 3D scenes remains challenging despite recent progress in closed-set and open-set 3D scene understanding methods.
View details
Quantifying urban park use in the USA at scale: empirical estimates of realised park usage using smartphone location data
Michael T Young
Swapnil Vispute
Stylianos Serghiou
Akim Kumok
Yash Shah
Kevin J. Lane
Flannery Black-Ingersoll
Paige Brochu
Monica Bharel
Sarah Skenazy
Shailesh Bavadekar
Mansi Kansal
Evgeniy Gabrilovich
Gregory A. Wellenius
Lancet Planetary Health (2024)
Preview abstract
Summary
Background A large body of evidence connects access to greenspace with substantial benefits to physical and mental
health. In urban settings where access to greenspace can be limited, park access and use have been associated with
higher levels of physical activity, improved physical health, and lower levels of markers of mental distress. Despite the
potential health benefits of urban parks, little is known about how park usage varies across locations (between or
within cities) or over time.
Methods We estimated park usage among urban residents (identified as residents of urban census tracts) in
498 US cities from 2019 to 2021 from aggregated and anonymised opted-in smartphone location history data. We
used descriptive statistics to quantify differences in park usage over time, between cities, and across census tracts
within cities, and used generalised linear models to estimate the associations between park usage and census tract
level descriptors.
Findings In spring (March 1 to May 31) 2019, 18·9% of urban residents visited a park at least once per week, with
average use higher in northwest and southwest USA, and lowest in the southeast. Park usage varied substantially
both within and between cities; was unequally distributed across census tract-level markers of race, ethnicity, income,
and social vulnerability; and was only moderately correlated with established markers of census tract greenspace. In
spring 2019, a doubling of walking time to parks was associated with a 10·1% (95% CI 5·6–14·3) lower average
weekly park usage, adjusting for city and social vulnerability index. The median decline in park usage from spring
2019 to spring 2020 was 38·0% (IQR 28·4–46·5), coincident with the onset of physical distancing policies across
much of the country. We estimated that the COVID-19-related decline in park usage was more pronounced for those
living further from a park and those living in areas of higher social vulnerability.
Interpretation These estimates provide novel insights into the patterns and correlates of park use and could enable
new studies of the health benefits of urban greenspace. In addition, the availability of an empirical park usage metric
that varies over time could be a useful tool for assessing the effectiveness of policies intended to increase such
activities.
View details
Preview abstract
Current approaches to Analog Layout Automation
apply ML techniques such as Graph Convolutional Neural
Networks (GCN) to translate netlist to layout. While these ML
approaches have proven to be effective, they lack the powerful
reasoning capabilities, an intuitive human interface, and standard
evaluation benchmarks that have been improving at a rapid de-
velopment pace in Large Language Models (LLMs). The GLayout
framework introduced in this work translates analog layout into
an expressive, technology generic, compact text representation.
Then, an LLM is taught to understand analog layout through
fine-tuning and in-context learning using Retrieval Augmented
Generation (RAG). The LLM is able to successfully layout unseen
circuits based on new information provided in-context. We train
3.8, 7, and 22 Billion parameter quantized LLMs on a dataset
of less than 50 unique circuits, and text documents providing
layout knowledge. The 22B parameter model is tuned in 2 hours
on a single NVIDIA A100 GPU. The open-source evaluation
set is proposed as an automation benchmark for LLM layout
automation tasks, and ranges from 2-transistor circuits to a
∆Σ ADC. The 22B model completes 70% of the tasks in the
evaluation set, and is able to pass DRC and LVS verification on
unseen 4 transistor blocks.
View details
Learning to Rewrite Prompts for Personalized Text Generation
Qiaozhu Mei
Proceedings of the ACM Web Conference 2024
Preview abstract
Facilitated by large language models (LLMs), personalized text generation has become a rapidly growing research direction. Most existing studies focus on designing specialized models for a particular domain, or they require fine-tuning the LLMs to generate personalized text. We consider a typical scenario in which the large language model, which generates personalized output, is frozen and can only be accessed through APIs. Under this constraint, all one can do is to improve the input text (i.e., text prompts) sent to the LLM, a procedure that is usually done manually. In this paper, we propose a novel method to automatically revise prompts for personalized text generation. The proposed method takes the initial prompts generated by a state-of-the-art, multistage framework for personalized generation and rewrites a few critical components that summarize and synthesize the personal context. The prompt rewriter employs a training paradigm that chains together supervised learning (SL) and reinforcement learning (RL), where SL reduces the search space of RL and RL facilitates end-to-end training of the rewriter. Using datasets from three representative domains, we demonstrate that the rewritten prompts outperform both the original prompts and the prompts optimized via supervised learning or reinforcement learning alone. In-depth analysis of the rewritten prompts shows that they are not only human readable, but also able to guide manual revision of prompts when there is limited resource to employ reinforcement learning to train the prompt rewriter, or when it is costly to deploy an automatic prompt rewriter for inference.
View details
Learning from Label Proportions: Bootstrapping Supervised Learners via Belief Propagation
Shreyas Havaldar
The Twelfth International Conference on Learning Representations (ICLR) (2024)
Preview abstract
Learning from Label Proportions (LLP) is a learning problem where only aggregate level labels are available for groups of instances, called bags, during training, and the aim is to get the best performance at the instance-level on the test data. This setting arises in domains like advertising and medicine due to privacy considerations. We propose a novel algorithmic framework for this problem that iteratively performs two main steps. For the first step (Pseudo Labeling) in every iteration, we define a Gibbs distribution over binary instance labels that incorporates a) covariate information through the constraint that instances with similar covariates should have similar labels and b) the bag level aggregated label. We then use Belief Propagation (BP) to marginalize the Gibbs distribution to obtain pseudo labels. In the second step (Embedding Refinement), we use the pseudo labels to provide supervision for a learner that yields a better embedding. Further, we iterate on the two steps again by using the second step's embeddings as new covariates for the next iteration. In the final iteration, a classifier is trained using the pseudo labels. Our algorithm displays strong gains against several SOTA baselines for the LLP Binary Classification problem on various dataset types - Small Tabular, Large Tabular and Images. We achieve these improvements with minimal computational overhead above standard supervised learning due to Belief Propagation, for large bag sizes, even for a million samples.
View details
