Search papers, labs, and topics across Lattice.
27 papers published across 2 labs.
Task success rates for agentic phone use soar from 36.67% to 45.33% through a novel combination of real and mock environments in training.
Federated learning in health research just got easier with FLKit, a structured onboarding toolkit that demystifies the process for diverse teams.
Only the right key can unlock the original image from diffusion models, turning a security risk into a robust feature against unauthorized reconstruction.
AOHP redefines how AI agents interact with operating systems, achieving a 21% boost in task completion and a dramatic cut in execution costs.
Bounded Context violations in DDD projects are shown to significantly hinder maintenance quality, revealing a critical challenge for practitioners.
Task success rates for agentic phone use soar from 36.67% to 45.33% through a novel combination of real and mock environments in training.
Federated learning in health research just got easier with FLKit, a structured onboarding toolkit that demystifies the process for diverse teams.
Only the right key can unlock the original image from diffusion models, turning a security risk into a robust feature against unauthorized reconstruction.
AOHP redefines how AI agents interact with operating systems, achieving a 21% boost in task completion and a dramatic cut in execution costs.
Bounded Context violations in DDD projects are shown to significantly hinder maintenance quality, revealing a critical challenge for practitioners.
The Composition event series has successfully bridged the gap between art and technology, sparking vibrant community engagement and collaboration.
Transitioning from restrictive to permissive licenses can either boost or hinder project activity, depending on the programming language used.
A staggering 39.4% of open source projects may be at risk of license noncompliance due to copy-based reuse, challenging the effectiveness of existing dependency tracking methods.
Controlled randomness in spiking neurons can significantly enhance edge hardware performance while minimizing area and noise sensitivity.
Kiwano revolutionizes speaker verification research by offering a comprehensive, user-friendly toolkit that streamlines experimentation and evaluation.
VeriPort not only backports patches at scale but also uncovers hidden vulnerabilities, correcting over 400 erroneous version reports in the process.
TLMs can unlock sensitive capabilities in open-weight models without risking public exposure, enabling a new paradigm for LLM deployment.
Digital tools are reshaping labor organizing, but they also introduce significant challenges that could undermine union efforts.
Only one out of 37 open-weight AI models meets the necessary evaluation standards to ensure safety and robustness, revealing a critical oversight in current practices.
A novel continuum robot platform achieves reproducibility and ease of control, setting a new standard for research in robotic manipulation.
Digital speech acts could redefine copyright ownership, enabling individuals to retain control over their content without reliance on centralized platforms.
European Portuguese finally gets its own dedicated vision and language model, setting a new standard for inclusivity in multimodal AI.
LLM agents can identify reproducibility problems in 90% of analyzed machine learning papers, leveraging GitHub issues as a novel supervision source.
Current open-source LLMs fall short in effectively classifying complex CTI reports, with the best model only achieving an F1 score of 0.22.
A fixed corpus allows researchers to uncover that nearly a third of top cybersecurity conference papers are arXiv preprints, significantly impacting how we assess publication timelines.
Reproducibility in AI research has surged from 28% to 64% in just a decade, driven by a cultural shift towards open science.
Over 55% of AI models analyzed show compliance risks, underscoring a critical gap in the current compliance landscape.
A new open-source framework transforms how multi-center radiology studies monitor progress and explore data, enabling real-time insights and efficient collaboration.
Reproducibility in vulnerability datasets can now be achieved at scale, with ARVO demonstrating 81% success in reproducing real-world vulnerabilities.
Only 1% of SSD simulator experiments are reproducible, but StorRep offers a solution with six robust artifacts that can transform storage research practices.
MultiMolecule transforms how researchers interact with biomolecular models by linking execution context and standardized checkpoints, enabling seamless adaptation and deployment.
LineageMark ensures that contributor watermarks survive multiple model adaptations, maintaining integrity even under fine-tuning and pruning.
