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 11645 publications
Preview abstract
Modern transformers have been scaled up in size, data, training-time compute, and more recently test-time compute based on Chain-of-Thought (CoT), but the consequences of trading off these parameters are unknown. We construct explicit tradeoffs between two of these aspects: the number of layers in a transformer and the number of CoT tokens. Our first result is general: with mild assumptions and a single-token prompt, for any integer k, a transformer using soft attention and SwiGLU MLP layers with L layers and n CoT tokens can be converted to a transformer with at most ckL layers and ⌈n/k⌉ CoT tokens, and vice versa, although with an increase to the size of other parameters. We then a multi-token prompt path following task, finding that additional preprocessing from the multi-token prompt changes the depth-CoT relationship in this case. We construct an attention-only network with L layers and 2^(L−2) heads that can traverse 2^(L−1) − 1 edges along a path with a single CoT token, demonstrating at least an exponential relationship between increases in depth and reductions in the number of CoT tokens needed to traverse the entire path. This relationship for the path following task holds empirically as well, as demonstrated by training networks with up to 5 layers, with the tradeoff between depth and the maximum number of edge traversals in a single CoT token exactly matching our theoretical constructions.
View details
A 3D Scene Graphs Survey: Open Challenges and Future Directions
Dennis Rotondi
Francesco Argenziano
Sebastian Koch
Nathan Hughes
Martin Büchner
Johanna Wald
Lukas Schmid
Daniele Nardi
Abhinav Valada
Liam Paul
Luca Carlone
Kai Arras
Annual Review of Control, Robotics, and Autonomous Systems (ARCRAS), 10 (2027) (to appear)
Preview abstract
3D Scene Graphs (3DSGs) have emerged as a powerful representation for spatial AI by combining geometric grounding with semantic and relational abstractions of the environment. Their expressiveness has made them relevant to a broad range of problems in robotics and computer vision, including mapping, task and motion planning, scene understanding, and many others. However, the field remains fragmented: different communities adopt distinct formulations, construction pipelines, and evaluation protocols, making it difficult to compare methods, identify common assumptions, and assess remaining challenges for robust real- world deployment. This survey provides a unified and critical review of 3DSGs, with particular emphasis on open challenges and future directions. We first formalize 3DSGs under a common definition and analyze the principal modeling choices that characterize existing formulations, including node and edge attributes, hierarchical structure, dynamic scene representations, and affordance-aware extensions. We then review how 3DSGs are constructed from raw sensory observations, covering both learning-oriented and construction-oriented systems. Finally, we examine downstream applications and evaluation strategies, from intrinsic graph quality to task-level performance. To support the community, we also provide a dedicated website that organizes and extends the surveyed works.
View details
When Android Apps Request Location with a Reason: A Developer Study and User Behavior Experiment
May Smith
Aman Jain
Proceedings on Privacy Enhancing Technologies (2027)
Preview abstract
Recent reports have highlighted how mobile apps share user location data with third parties, risking user privacy and platform trust. Although location data is highly sensitive, when users grant apps location access, they may not know the full extent to which it is used. We study how requiring Android apps to show a reason for location access could impact developers, users, and the platform.
We surveyed 323 Android app developers and found most supported such a requirement. The majority said it would have a positive impact on user privacy, trust for apps, and trust for Android, where impact on user trust for Android correlated most strongly with support. Many developers also said the intervention would increase the number of users granting location access. Yet their open-ended comments also revealed consistent concerns, such as apps providing dishonest reasons and platform verification.
To study the impact on user behavior, we conducted a randomized controlled experiment with 2579 US Android users. We tested how users' decisions to grant location access were impacted by app type, whether reasons were included in the requests, and the content of the reasons, including monetization. We did not find the reasons impacted users' decisions; decisions were instead driven by app type and demographics. Yet we did find the reasons could have a positive impact on user perception for the platform when the reasons did not include using data for ads. Our findings provide insights into developers' willingness to implement privacy-enhancing changes, and expose limits to improving user privacy by simply adding information to user interfaces.
View details
Securing Multi-Agent Systems: An Empirical Analysis of Security Prompt Hardening and Residual Risks
Mithil Roshan
Shashank Nagesh
(2026)
Preview abstract
The rapid adoption of agentic systems powered by
large language models (LLMs) introduces significant
security challenges distinct from plain conversational
models, particularly concerning prompt injection and
tool misuse due to their dynamic personas and real-
world tool interactions. This paper investigates the
effectiveness of hardened security prompting in a
task-oriented multi-agent framework, using a coding
assistant as a representative case study. We com-
pare a baseline ”unhardened” agent against a ”hard-
ened” version equipped with explicit security guide-
lines applied across all sub-agents. Our evaluation
across 150+ single-turn and 32 multi-turn attack sce-
narios demonstrates that prompt hardening dramat-
ically improves resilience. With a simple, approxi-
mately 500-token security hardener, single-turn fail-
ure rates dropped from 19.48% to 2.60%, while multi-
turn failure rates decreased from 75.00% to 46.88%.
Furthermore, we show that successfully bypassing the
hardened agent requires significantly more adversar-
ial effort and a greater number of chat turns. How-
ever, the analysis also reveals a critical shift in vul-
nerability taxonomy: as direct attacks fail, adver-
saries exploit the agent’s core functionality via ”Func-
tional Wrappers” (Intent Obfuscation), highlighting
a residual risk that necessitates a shift in the defen-
sive paradigm from static filters to dynamic runtime
state and intent analysis.
View details
Performance analysis of updated Sleep Tracking algorithms across Google and Fitbit wearable devices
Arno Charton
Linda Lei
Siddhant Swaroop
Marius Guerard
Michael Dixon
Logan Niehaus
Shao-Po Ma
Ross Wilkinson
Ryan Gillard
Conor Heneghan
Pramod Rudrapatna
Mark Malhotra
Shwetak Patel
Google, Google, 1600 Amphitheatre Parkway
Mountain View, CA 94043 (2026) (to appear)
Preview abstract
Background: The general public has increasingly adopted consumer wearables for sleep tracking over the past 15 years, but reports on performance versus gold standards such as polysomnogram (PSG), high quality sleep diaries and at-home portable EEG systems still show potential for improved performance. Two aspects in particular are worthy of consideration: (a) improved recognition of sleep sessions (times when a person is in bed and has attempted to sleep), and (b) improved accuracy on recognizing sleep stages relative to an accepted standard such as PSG.
Aims: This study aimed to: 1) provide an update on the methodology and performance of a system for correctly recognizing valid sleep sessions, and 2) detail an updated description of how sleep stages are calculated using accelerometer and inter-beat intervals
Methods: Novel machine learning algorithms were developed to recognize sleep sessions and sleep stages using accelerometer sensors and inter-beat intervals derived from the watch or tracker photoplethysmogram. Algorithms were developed on over 3000 nights of human-scored free-living sleep sessions from a representative population of 122 subjects, and then tested on an independent validation set of 47 users. Within sleep sessions, an algorithm was developed to recognize periods when the user was attempting to sleep (Time-Attempting-To-Sleep = TATS). For sleep stage estimation, an algorithm was trained on human expert-scored polysomnograms, and then tested on 50 withheld subject nights for its ability to recognize Wake, Light (N1/N2), Deep (N3) and REM sleep relative to expert scored labels.
Results: For sleep session estimation, the algorithm had at least 95% overlap on TATS with human consensus scoring for 94% of nights from healthy sleepers. For sleep stage estimation, comparing with the current Fitbit algorithm, Cohen’s kappa for four-class determination of sleep stage increased from an average of 0.56 (std 0.13) to 0.63 (std 0.12), and average accuracy increased from 71% (std 0.10) to 77% (std 0.078)
Conclusion: A set of new algorithms has been developed and tested on Fitbit and Pixel Watches and is capable of providing robust and accurate measurement of sleep in free-living environments.
View details
Preview abstract
High-volume enterprise service organizations face a persistent challenge in transitioning from reactive support models to proactive, preventative ones. This paper introduces the Agentic Trend-to-Knowledge (ATK) methodology, a novel, autonomous framework designed to address this gap. The ATK methodology employs an AI agent that operates in a recurring, closed loop. It first uses a two-stage process for the autonomous thematic analysis of recent support cases to identify the most significant recurring issue. It then leverages Retrieval-Augmented Generation (RAG) to source relevant institutional knowledge. A key innovation is the agent's adaptive, bimodal response: if relevant knowledge is found, it drafts a proactive communication for human review; if a knowledge gap is detected, it autonomously creates a content creation task for the appropriate team. This transforms the agent from an automation tool into a proactive process owner that creates a virtuous cycle of continuous improvement for both case deflection and knowledge base quality. By automating the entire workflow from insight to action, the ATK framework provides a concrete methodology for shifting from a "human-in-the-loop" to a more strategic "human-on-the-loop" operational paradigm.
View details
Preview abstract
When managing complex, unpredictable (non-deterministic) AI agents using simple, fixed control systems (like finite state machines), operational failures and accountability issues often arise. This document introduces a probabilistic governance and telemetry framework to resolve these problems. Instead of following a rigid sequence of steps, this framework defines a multi-dimensional operational boundary, a 'behavioral volume', and assigns the agent a goal. This allows the agent to use its own reasoning to achieve the goal while remaining within the defined boundaries. A separate telemetry layer monitors the agent's actions by calculating metrics, such as alignment scores and drift velocity, to measure how much the agent deviates from its intended behavior. This system provides a method for guiding, monitoring, and securing autonomous agents, effectively managing the performance and security of an unpredictable AI workforce in complex environments.
View details
GroundSet: A Cadastral-Grounded Dataset for Spatial Understanding with Vector Data
Roger Ferrod
Maël Lecene
Krishna Sapkota
Vered Silverman
Sylvain Lobry
Computer Vision -- ECCV 2026, Springer, Cham, pp. 93-108
Preview abstract
Precise spatial understanding in Earth Observation is essential for translating raw aerial imagery into actionable insights for critical applications like urban planning, environmental monitoring and disaster management. However, Multimodal Large Language Models exhibit critical deficiencies in fine-grained spatial understanding within Remote Sensing, primarily due to a reliance on limited or repurposed legacy datasets.
To bridge this gap, we introduce a large-scale dataset grounded in verifiable cadastral vector data, comprising 3.8 million annotated objects across 510k high-resolution images with 135 granular semantic categories. We validate this resource through a comprehensive instruction-tuning benchmark spanning seven spatial grounding tasks.
Our evaluation establishes a robust baseline using a standard LLaVA architecture. We show that while current RS-specialized and commercial models (e.g., Gemini) struggle in zero-shot settings, high-fidelity supervision effectively bridges this gap, enabling standard architectures to master fine-grained spatial grounding without complex architectural modifications.
Data, pretrained model and code are available at: https://huggingface.co/datasets/RogerFerrod/GroundSet
View details
Towards conversational artificial intelligence for disease management
Khaled Saab
David Stutz
Yong Cheng
Kavita Kulkarni
Sara Mahdavi
Joelle Barral
James Manyika
Ryutaro Tanno
Adam Rodman
Nature (2026)
Preview abstract
Although large language models have shown promise in diagnostic dialogue, their capabilities for effective management reasoning, including disease progression, therapeutic response and safe medication prescription, have remained underexplored. We have advanced the previously demonstrated diagnostic capabilities of the Articulate Medical Intelligence Explorer (AMIE) using a new large-language-model-based agentic system optimized for multivisit clinical management and dialogue. To ground the reasoning of AMIE in authoritative clinical knowledge, we leveraged the long-context capabilities of Gemini, combining in-context retrieval with structured reasoning to align its output with up-to-date clinical practice guidelines and drug formularies. In a randomized, blinded virtual Objective Structured Clinical Examination study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multivisit case scenarios designed to reflect the guidance of the UK National Institute for Health and Care Excellence and BMJ Best Practice guidelines. AMIE was non-inferior to PCPs in management reasoning, as assessed by specialists, and scored better both with respect to preciseness of treatment and investigation, and in terms of its alignment with and grounding in clinical guidelines. To benchmark medication reasoning, we developed RxQA, a multiple-choice question benchmark that was derived from two national drug formularies (from the USA and UK) and validated by board-certified pharmacists. Although AMIE and PCPs both benefited from the ability to access external drug information, AMIE outperformed PCPs on higher-difficulty questions. Although further research will be needed before real-world translation of AMIE, its strong performance across evaluations marks a significant step towards use of conversational artificial intelligence as a tool in disease management.
View details
Identification of camera trap images by artificial intelligence and human experts produces similar multi-species occupancy models
Daniel Thornton
Travis King
Lucy Perera-Romero
Alissa Anderson
Rony Garcia-Anleu
Scott Fitkin
Carly Vynne
Journal of Applied Ecology (2026)
Preview abstract
The use of camera traps in ecology and conservation has expanded rapidly, but the time spent to accurately identify species in camera trap images remains a fundamental challenge that limits project scope and impact. Although artificial intelligence (AI) is often used to speed up image processing, a human review step is still standard practice to arrive at final species identifications. A potentially transformative next step is thus to remove humans entirely from the analysis chain and still produce accurate statistical models for subsequent inference across a wide diversity of species and regions.
Here, we compare the output of Bayesian multi-species occupancy models derived from a complete AI workflow (no human review of images) with a general species classifier (SpeciesNet) to those from an expert (human) workflow, using large-scale camera datasets from three study areas and two distinct and diverse mid-large mammal assemblages. We apply several pre- and post-processing steps to the AI workflow to improve model agreement. We perform a comprehensive model comparison, assessing agreement in identified species–environment relationships and rates of occupancy and detection, and similarity of spatial projections of occupancy.
We found that for most species of mammal, AI based models were remarkably similar to expert-based models, with some variability based on post-processing decisions. This agreement was robust, holding across multiple metrics of comparison (e.g. parameter estimates, precision, occupancy and detection rates), multiple study sites and at species and community levels. Similarity in model output occurred even in the presence of misclassification errors, suggesting our approach was resilient to some level of false negatives and positives. Substantial divergence in model output and subsequent inference, while rare, was most prevalent for rarely detected species.
Synthesis and applications: The use of a global AI classifier to identify species and reproducible pre- and post-processing decisions makes our approach broadly applicable and particularly beneficial for national and international monitoring programs that collect large amounts of photo data on threatened, at risk, or management sensitive species and wildlife communities. A fully automated workflow will allow such programs to progress more rapidly from photo collection to analysis, inference and decision-making.
View details
Digital Shadow AI Risk Theoretical Framework (DART): A Framework for Managing Data Disclosure and Privacy Risks of AI tools at Work
Master's Thesis (2026)
Preview abstract
The accelerated integration of generative AI technologies and agentic AI tools, particularly those like ChatGPT, into workplace settings has introduced complex challenges concerning data governance, regulatory compliance, and organizational privacy (GDPR 2016; CCPA/CPRA). This study introduces the Digital Shadow AI Risk Theoretical Framework (DART)—a novel theoretical framework designed to systematically identify, classify, and address the latent risks arising from the widespread, and often unregulated, use of AI systems in professional environments (NIST, 2023; OECD AI Policy Observatory, 2023). DART introduces six original, interrelated constructs developed in this study: Unintentional Disclosure Risk, Trust-Dependence Paradox, Data Sovereignty Conflict, Knowledge Dilution Phenomenon, Ethical Black Box Problem, and Organizational Feedback Loops. Each construct reflects a unique dimension of risk that emerges as organizations increasingly rely on AI-driven tools for knowledge work and decision-making.
The framework is empirically tested through a mixed-methods research design involving hypothesis testing and statistical analysis of behavioral data gathered from cross-sectional surveys of industry professionals. Two cross-industry surveys (Survey-1: 416 responses, 374 analyzed; Survey-2: 203 responses, 179 analyzed) and CB-SEM tests supported seven of eight hypotheses; H4 (sovereignty) was not significant; H7 (knowledge dilution) was confirmed in replication. The findings highlight critical gaps in employee training, policy awareness, and risk mitigation strategies—underscoring the urgent need for updated governance frameworks, comprehensive AI-use policies, and targeted educational interventions. This paper contributes to emerging scholarship by offering a robust model for understanding and mitigating digital risks in AI-enabled workplaces, providing practical implications for compliance officers, risk managers, and organizational leaders aiming to harness the benefits of generative AI responsibly and securely. The novelty of DART lies in its explicit theorization of workplace-level behavioral risks—especially Shadow AI, which unlike Shadow IT externalizes organizational knowledge into adaptive systems—thereby offering a unified framework that bridges fragmented literatures and grounds them in empirical evidence.
View details
Gaze Target Estimation Anywhere with Concepts
Xu Cao
Houze Yang
Vipin Gunda
Inki Kim
Jim Rehg
Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (2026)
Preview abstract
Estimating human gaze targets in-the-wild is a formidable challenge. Existing computer vision algorithms rely on brittle, multi-stage pipelines that require explicit inputs like head bounding boxes and human pose, causing initial detection errors to cascade and lead to system failure. To overcome this, we introduce the \textbf{Promptable Gaze Target Estimation (PGE)} task, a new end-to-end, concept-driven paradigm. PGE conditions gaze prediction on flexible user text or visual prompts (e.g., "the boy in the red shirt" or "person in point [0.52, 0.48]") to identify a specific subject's target, which eliminates the rigid dependency on intermediate localization cues. We develop a scalable data engine to generate \textbf{Gaze-Co}, a dataset and benchmark of 120K high-quality, prompt-annotated image pairs. We also propose \textbf{AnyGaze}, the first model designed for PGE. AnyGaze uses a Transformer-based detector to fuse features from frozen encoders and simultaneously solves subject localization, in/out-of-frame presence, and gaze target heatmap estimation. AnyGaze achieves state-of-the-art performance on standard gaze target estimation benchmarks, setting a strong baseline for this new problem even on a difficult out-of-domain, real-world clinical dataset. We will open-source the AnyGaze model and the Gaze-Co benchmark.
View details
Accurate ground state energy estimation with noise and imperfect state preparation
Tom O'Brien
Alicja Dutkiewicz
Stefano Polla
ArXiv (2026)
Preview abstract
We introduce a classical estimator for the post-processing of quantum phase estimation (QPE) data generated either by quantum-Fourier-transform-based or quantum-signal-processing-based methods. These methods sample from a distribution supported on $[0, 2\pi]$, depending on the location of the target phase within this window. We focus on the case where the target phase lies within a smaller promise region, where no other phases are present, which is typical of e.g. ground state energy estimation of gapped quantum systems. Our estimation technique is based on filtering the signal within the promise region and recovering the phase through a moment-projection estimator. We show that our methods are robust in the presence of both additional phases outside the promise region and global depolarizing noise. We observe in the noiseless case our estimator achieves an exponential improvement over naive mean estimation in certain parameter regimes. In the presence of global depolarizing noise, no robust state of the art is known, and our estimator achieves a bias exponentially small in the circuit depth $T$ at fixed circuit fidelity $F$, and a variance proportional to $T^{-2}$, improving by a factor of $T^4$ over the naive approach. To mitigate realistic circuit-level noise, we combine our method with the explicit unbiasing scheme described in Ref.~\cite{dutkiewiczError2025}. As an illustrative example, we implement these estimators on a small-scale simulation of the Ising model. We observe that the moment-projection estimator retains some residual bias removable by the explicit unbiasing scheme, but this was significantly smaller than the variance in the system for the parameter range tested. The robustness of the moment-projection estimator in the presence of both multiple eigenvalues and noise makes phase estimation with limited depth practical for early fault tolerant quantum experiments.
View details
Mobility-Embedded POIs: Learning What a Place Is and How It’s Used from Human Movement
Shushman Choudhury
Shang-Ling Hsu
Cyrus Shahabi
Forty-third International Conference on Machine Learning (2026)
Preview abstract
Recent progress in geospatial foundation models (GeoFMs) has highlighted the
importance of learning general-purpose representations for real-world locations,
particularly Points of Interest (POIs) where human activity concentrates. Yet, ex-
isting POI representations remain largely static, drawing from textual metadata
(e.g., category labels, descriptions) and spatial attributes (e.g., coordinates, neigh-
borhood context), all of which describe what a place is, but not how it is actu-
ally used. We argue that human mobility provides a complementary and dynamic
signal, capturing real-world visitation patterns that reveal how places function in
practice. To this end, we introduce Mobility Embedded POIs (ME-POIs), a
pretraining framework that learns POI representations directly from sequences of
human visits. Each visit is encoded as a contextualized embedding that captures
the POI’s static attributes as well as its temporal and sequential context, includ-
ing when the visit occurs and which visits surround it. These visit embeddings
are aligned with learnable POI embeddings via a contrastive objective, grounding
POI representations in their real-world usage patterns. To address the long tail of
sparsely visited POIs, we transfer visitation distributions from data-rich anchors
to sparse locations, leveraging multi-scale spatial proximity to capture local and
regional patterns, and functional similarity to enable transfer across semantically
related POIs. We demonstrate the utility of ME-POIs for a set of automated map
enrichment tasks, critical in geospatial intelligence. We show empirically that
by embedding visitation dynamics, ME-POIs outperform text- and location-only
baselines, proving that mobility-informed embeddings provide a stronger founda-
tion for modeling place function and change.
View details
IBBench-Light: A Paired Evaluation of Task-Conditioned Responses to External Directives
Mark Zhou
Zhaoyi Li
Janet Sung
Iris Qian
Hang Xiao
2026
Preview abstract
in many large language model (LLM) applications, a serialized prompt contains both a user request and external records such as webpages, email, memory, or tool outputs. The same embedded directive may need to be applied for one task and treated as data for another. IBBench-Light tests this contrast with 12 semantic tasks rendered through four source wrappers and three embedding forms. The construction yields 144 matched records and 288 prompts per model. Four 4-bit open-weight instruction models produce 1,152 archived single-run greedy responses. Under the exact output contract, paired exact-contract accuracy (PECA) ranges from 1.4% to 67.4%; Qwen3-4B and Mistral-7B have the two highest point estimates on this four-model panel. Task-cluster resampling describes variation across the 12 semantic bases, while standalone-target scoring separates some target-selection failures from response-form errors. A post-hoc leading-target rule changes Phi-4-mini’s paired score substantially; the archived run lacks the stop-reason metadata needed to resolve whether generation termination caused these continuations. The findings describe synthetic, single-turn tasks under one user-role serialization. Adaptive attacks, role-level hierarchy, and tool-mediated effects require separate experiments
View details
