Machine Intelligence
Google is at the forefront of innovation in Machine Intelligence, with active research exploring virtually all aspects of machine learning, including deep learning and more classical algorithms. Exploring theory as well as application, much of our work on language, speech, translation, visual processing, ranking and prediction relies on Machine Intelligence. In all of those tasks and many others, we gather large volumes of direct or indirect evidence of relationships of interest, applying learning algorithms to understand and generalize.
Machine Intelligence at Google raises deep scientific and engineering challenges, allowing us to contribute to the broader academic research community through technical talks and publications in major conferences and journals. Contrary to much of current theory and practice, the statistics of the data we observe shifts rapidly, the features of interest change as well, and the volume of data often requires enormous computation capacity. When learning systems are placed at the core of interactive services in a fast changing and sometimes adversarial environment, combinations of techniques including deep learning and statistical models need to be combined with ideas from control and game theory.
Recent Publications
InstructPipe: Generating Visual Blocks Pipelines with Human Instructions and LLMs
Jing Jin
Xiuxiu Yuan
Jun Jiang
Jingtao Zhou
Yiyi Huang
Zheng Xu
Kristen Wright
Jason Mayes
Mark Sherwood
Johnny Lee
Alex Olwal
Ram Iyengar
Na Li
Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI), ACM, pp. 23
Preview abstract
Visual programming has the potential of providing novice programmers with a low-code experience to build customized processing pipelines. Existing systems typically require users to build pipelines from scratch, implying that novice users are expected to set up and link appropriate nodes from a blank workspace. In this paper, we introduce InstructPipe, an AI assistant for prototyping machine learning (ML) pipelines with text instructions. We contribute two large language model (LLM) modules and a code interpreter as part of our framework. The LLM modules generate pseudocode for a target pipeline, and the interpreter renders the pipeline in the node-graph editor for further human-AI collaboration. Both technical and user evaluation (N=16) shows that InstructPipe empowers users to streamline their ML pipeline workflow, reduce their learning curve, and leverage open-ended commands to spark innovative ideas.
View details
SEMQA: Semi-Extractive Multi-Source Question Answering
Haitian Sun
NAACL (2024) (to appear)
Preview abstract
Recently proposed long-form question answering (QA) systems, supported by large language models (LLMs), have shown promising capabilities. Yet, attributing and verifying their generated abstractive answers can be difficult, and automatically evaluating their accuracy remains an ongoing challenge.
In this paper, we introduce a new QA task for answering multi-answer questions by summarizing multiple diverse sources in a semi-extractive fashion. Specifically, Semi-extractive Multi-source QA (SEMQA) requires models to output a comprehensive answer while mixing between factual quoted spans---copied verbatim from given input sources---and non-factual free-text connectors that glue these spans together into a single cohesive passage. This setting bridges the gap between the outputs of well-grounded but constrained extractive QA systems and more fluent but harder to attribute fully abstractive answers. Particularly, it enables a new mode for language models that leverages their advanced language generation capabilities, while also producing fine in-line attributions by-design that are easy to verify, interpret, and evaluate. To study this task, we create the first dataset of this kind with human-written semi-extractive answers to natural and generated questions, and define text-based evaluation metrics. Experimenting with several LLMs in various settings, we find this task to be surprisingly challenging, demonstrating the importance of our work for developing and studying such consolidation capabilities.
View details
Artificial intelligence as a second reader for screening mammography
Etsuji Nakai
Alessandro Scoccia Pappagallo
Hiroki Kayama
Lin Yang
Shawn Xu
Christopher Kelly
Timo Kohlberger
Daniel Golden
Akib Uddin
Radiology Advances, 1(2) (2024)
Preview abstract
Background
Artificial intelligence (AI) has shown promise in mammography interpretation, and its use as a second reader in breast cancer screening may reduce the burden on health care systems.
Purpose
To evaluate the performance differences between routine double read and an AI as a second reader workflow (AISR), where the second reader is replaced with AI.
Materials and Methods
A cohort of patients undergoing routine breast cancer screening at a single center with mammography was retrospectively collected between 2005 and 2021. A model developed on US and UK data was fine-tuned on Japanese data. We subsequently performed a reader study with 10 qualified readers with varied experience (5 reader pairs), comparing routine double read to an AISR workflow.
Results
A “test set” of 4,059 women (mean age, 56 ± 14 years; 157 positive, 3,902 negative) was collected, with 278 (mean age 55 ± 13 years; 90 positive, 188 negative) evaluated for the reader study. We demonstrate an area under the curve =.84 (95% confidence interval [CI], 0.805-0.881) on the test set, with no significant difference to decisions made in clinical practice (P = .32). Compared with routine double reading, in the AISR arm, sensitivity improved by 7.6% (95% CI, 3.80-11.4; P = .00004) and specificity decreased 3.4% (1.42-5.43; P = .0016), with 71% (212/298) of scans no longer requiring input from a second reader. Variation in recall decision between reader pairs improved from a Cohen kappa of κ = .65 (96% CI, 0.61-0.68) to κ = .74 (96% CI, 0.71-0.77) in the AISR arm.
View details
Fairness under Covariate Shift: Improving Fairness-Accuracy tradeoff with few Unlabeled Test Samples
Shreyas Havaldar
Jatin Chauhan
Jay Nandy
The 38th Annual AAAI Conference on Artificial Intelligence (2024)
Preview abstract
Covariate shift in the test data is a common practical phenomena that can significantly downgrade both the accuracy and the fairness performance of the model. Ensuring fairness across different sensitive groups under covariate shift is of paramount importance due to societal implications like criminal justice. We operate in the unsupervised regime where only a small set of unlabeled test samples along with a labeled training set is available. Towards improving fairness under this highly challenging yet realistic scenario, we make three contributions. First is a novel composite weighted entropy based objective for prediction accuracy which is optimized along with a representation matching loss for fairness. We experimentally verify that optimizing with our loss formulation outperforms a number of state-of-the-art baselines in the pareto sense with respect to the fairness-accuracy tradeoff on several standard datasets. Our second contribution is a new setting we term Asymmetric Covariate Shift that, to the best of our knowledge, has not been studied before. Asymmetric covariate shift occurs when distribution of covariates of one group shifts significantly compared to the other groups and this happens when a dominant group is over-represented. While this setting is extremely challenging for current baselines, We show that our proposed method significantly outperforms them. Our third contribution is theoretical, where we show that our weighted entropy term along with prediction loss on the training set approximates test loss under covariate shift. Empirically and through formal sample complexity bounds, we show that this approximation to the unseen test loss does not depend on importance sampling variance which affects many other baselines.
View details
Preview abstract
Historically, much of machine learning research has focused on the performance of the algorithm alone, but recently more attention has been focused on optimizing joint human-algorithm performance. Here, we analyze a specific type of human-algorithm collaboration where the algorithm has access to a set of $n$ items, and presents a subset of size $k$ to the human, who selects a final item from among those $k$. This scenario could model content recommendation, route planning, or any type of labeling task. Because both the human and algorithm have imperfect, noisy information about the true ordering of items, the key question is: which value of $k$ maximizes the probability that the best item will be ultimately selected? For $k=1$, performance is optimized by the algorithm acting alone, and for $k=n$ it is optimized by the human acting alone.
Surprisingly, we show that for multiple of noise models, it is optimal to set $k \in [2, n-1]$ - that is, there are strict benefits to collaborating, even when the human and algorithm have equal accuracy separately. We demonstrate this theoretically for the Mallows model and experimentally for the Random Utilities models of noisy permutations. However, we show this pattern is \emph{reversed} when the human is anchored on the algorithm's presented ordering - the joint system always has strictly worse performance. We extend these results to the case where the human and algorithm differ in their accuracy levels, showing that there always exist regimes where a more accurate agent would strictly benefit from collaborating with a less accurate one, but these regimes are asymmetric between the human and the algorithm's accuracy.
View details
CodeQueries: A Dataset of Semantic Queries over Code
Surya Prakash Sahu
Madhurima Mandal
Shikhar Bharadwaj
Aditya Kanade
Shirish Shevade
Innovations in Software Engineering (ISEC), ACM, Bangalore, India (2024)
Preview abstract
Developers often have questions about semantic aspects of code
they are working on, e.g., “Is there a class whose parent classes
declare a conflicting attribute?”. Answering them requires understanding code semantics such as attributes and inheritance relation
of classes. An answer to such a question should identify code spans
constituting the answer (e.g., the declaration of the subclass) as well
as supporting facts (e.g., the definitions of the conflicting attributes).
The existing work on question-answering over code has considered
yes/no questions or method-level context. We contribute a labeled
dataset, called CodeQueries, of semantic queries over Python code.
Compared to the existing datasets, in CodeQueries, the queries
are about code semantics, the context is file level and the answers
are code spans. We curate the dataset based on queries supported
by a widely-used static analysis tool, CodeQL, and include both
positive and negative examples, and queries requiring single-hop
and multi-hop reasoning.
To assess the value of our dataset, we evaluate baseline neural
approaches. We study a large language model (GPT3.5-Turbo) in
zero-shot and few-shot settings on a subset of CodeQueries. We
also evaluate a BERT style model (CuBERT) with fine-tuning. We
find that these models achieve limited success on CodeQueries.
CodeQueries is thus a challenging dataset to test the ability of
neural models, to understand code semantics, in the extractive
question-answering setting
View details
Some of our teams
×
![]()