Search papers, labs, and topics across Lattice.
100 papers published across 8 labs.
More than half of algorithm mentions in NLP papers are for direct use, signaling a significant shift in how researchers engage with algorithms over time.
Non-parametric identification of causal mechanisms from steady-state observations could revolutionize data analysis in fields constrained by experimental limitations.
Factorizable Normalizing Flows enable efficient modeling of complex parameter-dependent densities without the combinatorial explosion of sampling joint configurations.
Learning the structure of open quantum systems is now feasible with a simple, efficient algorithm that overcomes unique challenges posed by Lindbladians.
ENC-ODE redefines biomarker prediction in neurodegenerative diseases by accurately modeling continuous dynamics from sparse clinical events, outperforming conventional methods.
Non-parametric identification of causal mechanisms from steady-state observations could revolutionize data analysis in fields constrained by experimental limitations.
Factorizable Normalizing Flows enable efficient modeling of complex parameter-dependent densities without the combinatorial explosion of sampling joint configurations.
Learning the structure of open quantum systems is now feasible with a simple, efficient algorithm that overcomes unique challenges posed by Lindbladians.
ENC-ODE redefines biomarker prediction in neurodegenerative diseases by accurately modeling continuous dynamics from sparse clinical events, outperforming conventional methods.
A novel polynomial relaxation reveals a one-to-one correspondence between local minima in Ising problems and their relaxed counterparts, enabling efficient gradient-based optimization.
A groundbreaking certification method ensures that language models can generate reliable physical designs without the risk of forgery, achieving zero false certifications in adversarial conditions.
Achieving accurate reconstruction of scalar potentials in the false vacuum regime reveals new insights into strongly coupled quantum systems through advanced machine learning techniques.
McMg slashes the computational burden of solving complex Helmholtz equations by leveraging learned phase-space corrections, making it a game-changer for high-frequency wave problems.
Explicit belief states in BayesEvolve lead to significantly improved sample efficiency in scientific discovery, challenging the reliance on mere memory archives.
Mistral outperforms existing methods with a striking 90.5% accuracy in topic classification, revealing a new standard for entity extraction in grant proposals.
Clarus transforms the landscape of scientific collaboration by enabling autonomous agents to work together in a structured, traceable, and resource-aware manner.
BioBERT outshines other models in detecting dosing errors, achieving a notable ROC-AUC of 0.794, paving the way for safer clinical trials.
Language models undergo a crystallization-like process during alignment, transitioning from high entropy to a concentrated distribution that reveals fundamental limits of alignment.
More than half of algorithm mentions in NLP papers are for direct use, signaling a significant shift in how researchers engage with algorithms over time.
Cybersecurity research needs a dedicated AI framework that evolves with the threat landscape, rather than relying on static models that can't keep pace with machine-speed attacks.
Achieving accurate material decomposition in sparse-view DECT could revolutionize medical imaging by enabling safer, lower-radiation scans without sacrificing detail.
SkelEM achieves high-fidelity detail restoration in volume microscopy in under five steps, outperforming existing self-supervised methods.
Current T2I models fall short in scientific reasoning, but fine-tuning on the new SciIR dataset boosts performance by over 20%.
Transforming correlated baths to de-correlated ones could significantly cut computational costs in simulating open quantum systems.
Fragment construction, not solution, drives the algorithmic bias in predicting tritium binding energies in molten salts, revealing a critical insight for future quantum-classical approaches.
Water release during metal oxide reduction is controlled by the evolution of nanopore structures, fundamentally altering our understanding of redox kinetics.
Proton-mediated dynamics in water clusters reveal that size significantly alters dissociation behavior, with sharp increases in H-ejection activity as cluster size grows.
Integrating clinical risk factors into deep learning models can dramatically enhance the accuracy of coronary artery stenosis grading, outperforming traditional methods.
Simpler molecular optimization methods can outperform advanced techniques in tackling the NMO Benchmark, revealing new insights for nanotechnology research.
Fine-tuning a 7B parameter model with a new Arabic-Russian corpus boosts translation performance, bridging a critical gap in scientific communication.
BPBO achieves substantial reductions in brickwork pattern size without sacrificing the blindness property of UBQC, challenging the limits of quantum computation optimization.
LLMs can autonomously generate efficient checkpoint/restart code for complex scientific applications, rivaling human expertise in resilience engineering.
StreamGuard can reduce the impact of failures in real-time data streams by up to 6x with minimal overhead, revolutionizing resilience in scientific computing.
Achieving a fourfold reduction in computational complexity for quantum simulations could revolutionize drug discovery efficiency in the industry.
A graphene barrier forms on Au nanoparticles during laser heating, significantly altering their mass dynamics in a vacuum.
TCR-SRIM reveals that relying on predicted structures can obscure critical interaction patterns, challenging the efficacy of popular structure prediction models in TCR-epitope learning.
ReactionAtlas uncovers nearly 47,000 reactions from just eight seed molecules, revolutionizing our understanding of carbohydrate chemistry and its implications for the origins of life.
Causal AI reveals that optimizing support-to-catalyst ratios can dramatically enhance electrocatalyst performance, challenging traditional material design approaches.
A unified platform for molecular machine learning that supports 100 elements and incorporates uncertainty quantification could democratize access to advanced chemical property predictions.
Flawless pyrite terraces can catalyze the self-assembly of L-Cysteine, reshaping our understanding of prebiotic molecular synthesis.
The openCOSMO-RS-Phi model achieves high accuracy in predicting thermodynamic properties while being fully open-source, democratizing access to advanced EoS tools.
Autocatalytic replication can dramatically speed up the search for reactive targets, but it comes with trade-offs that could limit efficiency.
The third frequency moment sum rule offers a new, reliable way to estimate kinetic energy in quantum systems, challenging traditional methods that may misrepresent short-wavelength behavior.
Optimal tuning is not just beneficial but essential for accurate predictions of electronic fundamental gaps at finite temperatures, fundamentally changing how we approach hybrid functional theory.
Spontaneous symmetry breaking in ZnPc aggregates reveals emergent chirality that could transform our approach to chiroptical and quantum materials.
Unlocking the visual data in materials science literature could revolutionize how researchers access and utilize experimental knowledge encoded in figures.
LLM4MOF reveals that language-model agents can autonomously and interpretable design complex materials, outperforming conventional search methods at a fraction of the cost.
Amplifying quantum pseudorandom states to t-copy security without extra assumptions opens new avenues for quantum cryptography.
TEPID-ADAPT-VQE achieves chemical accuracy in excited-state simulations while using only one hyperparameter, streamlining the optimization process for quantum chemistry.
Manganese-functionalized GelMA hydrogels could revolutionize precision oncology by enabling real-time MRI-guided immunotherapy with enhanced therapeutic efficacy.
Achieving an order-of-magnitude improvement in force direction prediction, CliffordSTF redefines the capabilities of interatomic potentials.
EFT enables LLMs to evolve solutions across diverse optimization tasks, achieving over 10% performance gains and state-of-the-art results in challenging mathematical problems.
A new AI tool can catch 34% more mathematical errors in scientific papers, transforming the peer review landscape.
ENS achieves up to 10x better accuracy than traditional methods in turbulent flow scenarios by directly utilizing the PDE residual as input for iterative error correction.
Learning a truncated Gaussian can be done in optimal time and sample complexity without the usual computational overhead of gradient descent.
Head Fisher alignment can be efficiently estimated in LLMs, revealing critical insights into task similarity that traditional metrics miss.
RecallRisk-BERT outperforms traditional models by integrating text and tabular data, achieving a remarkable accuracy of 0.963 in predicting recall severity.
Achieving 93% accuracy in insect authentication despite batch variations, BISN sets a new standard for robustness in near-infrared spectroscopy.
Transformers can generate complex triangulations that are vital for advancing our understanding of Calabi-Yau manifolds in string theory.
BOBa's innovative bandit-guided approach cuts down computational costs while boosting screening efficiency in the vast chemical space of drug discovery.
Multi-path forecasting reveals that early dynamics can predict divergent future behaviors, enabling more precise interventions in biopharmaceutical production.
L-DPS achieves robust inverse solutions while slashing inference costs, outperforming traditional methods in challenging sparse and noisy environments.
The Extra Trees model achieved an impressive 96.92% accuracy in detecting cirrhosis, highlighting the untapped potential of machine learning in liver disease diagnostics.
Simulation-based inference can achieve rapid and accurate Bayesian calibration of epidemiological models, outperforming traditional MCMC methods in both speed and efficiency.
Time-domain and nonlinear HRV indices are not only stable but also reveal significant gender differences, challenging existing assumptions about their clinical application.
AHOIS not only autonomously discovers new scientific hypotheses but also self-corrects through Socratic questioning, revolutionizing how we approach experimentation in high-dimensional systems.
Continuous modeling of ventricular motion can enhance heart failure risk prediction, outperforming traditional cardiac markers.
An optimal number of thought leaders can enhance team impact, but too many may stifle innovative ideas.
Women in science face a tilted playing field where institutional prestige amplifies success unevenly, favoring men even at lower-ranked institutions.
Achieving logarithmic parallel depth, MergeLLL significantly outperforms classical lattice reduction methods in both efficiency and quality of reduced bases.
Achieving mean deviations of only 0.44 arcminutes from established ephemerides, solarsystem redefines lightweight astronomical calculations.
GenMF achieves a breakthrough in 3D asset fabrication by preserving visual details while reducing stress concentrations, making monochromatic prints more reliable and visually appealing.
Second-order perturbative corrections to LR-SCI yield near-FCI accuracy for static molecular properties, revolutionizing the approach to complex quantum systems.
Achieving chemical accuracy with a search-free ansatz, this method outperforms traditional quantum architecture searches in estimating molecular ground-state energies.
Encapsulating photosystems within a virus-based nanocontainer can dramatically boost electron transfer efficiency, paving the way for greener chemical synthesis.
GRAINS achieves up to 47.8x speedup in genome graph analysis by processing data directly within storage, revolutionizing efficiency in genomic research.
Transformer models not only excel in classifying bacterial Raman spectra but also outperform traditional methods without requiring any preprocessing.
Efficiently training Hamiltonian Neural Networks can now be achieved even under noisy conditions, preserving energy conservation without excessive computational overhead.
Identifiability bounds reveal the precise conditions under which distinct ODEs can be distinguished from solution data, transforming our understanding of equation recovery in scientific machine learning.
Quantum graph neural networks can now achieve scalability and expressivity guarantees, paving the way for practical applications in complex relational data tasks.
The fTNN achieves unprecedented accuracy in solving fractional PDEs, outperforming traditional methods even in challenging scenarios with boundary singularities.
ArBG achieves a remarkable 60% reduction in zero-shot energy error for peptide systems, challenging the dominance of flow-based sampling methods.
Mixing rates improve by 1.5-3 times with the new SA-PAL method, showcasing a dramatic leap in sampling efficiency for complex systems.
KANs show promise in aerodynamic prediction but struggle with stability and hyperparameter sensitivity compared to MLPs and GNNs.
A novel meta-optimization framework that evolves evaluation criteria can dramatically accelerate scientific discovery, achieving a 67x speedup in algorithm performance.
LLMs excel at factual recall but falter on quantitative reasoning and conceptual tasks, revealing critical gaps in their domain-specific capabilities.
Achieving over 51% peak efficiency in a compact, wideband Doherty power amplifier through innovative deep learning techniques could revolutionize power amplifier design.
LLMs are reshaping the complexity of scientific language, evidenced by increased word turnover and altered relationships between style and complexity metrics.
Formulaic expression desensitization not only boosts dataset scale but also enhances model generalization in extracting key sentences from scientific literature.
NeuMatEx outperforms PBR techniques by extracting complex neural materials with unprecedented visual fidelity and precision from multi-view images.
A novel metamaterial enables real-time, ultrasensitive detection of molecular transformations by combining SEIRA and SERS, revolutionizing chemical fingerprinting.
Achieving noise-floor accuracy in self-limiting saturation curve modeling with just seven measurements could revolutionize experimental design in fields like pharmacology and materials science.
HI-NQS achieves chemical accuracy on strongly correlated systems with a scaling advantage over traditional methods, all without quantum computing resources.
Grounding EEG interpretations in hardware capabilities can drastically reduce unsupported conclusions and improve the reliability of scientific software.
Heavy-tailed estimators in Variational Monte Carlo can undermine convergence, but a new clipping method shows promise for robust optimization in electronic structure calculations.
Medical records and cardiac biomarkers alone outperform vascular graphs in predicting pulmonary embolism risk, raising questions about the value of complex vascular representations.
OncoSynth slashes treatment effect estimation errors by up to 66% in oncology, transforming how synthetic data can inform precision medicine.
Misclassifying medical conditions can have severe consequences, and the Ordinal Cross-Entropy framework significantly reduces prediction errors by respecting the ordinal nature of disease severity.
Achieving precise EUV mask designs through a physics-informed neural operator could revolutionize semiconductor manufacturing processes.
Hyperbolic neural quantum states outperform their Euclidean counterparts at critical points, revealing a new frontier in many-body quantum physics modeling.
Blasto-Net achieves remarkable accuracy in blastocyst analysis, outperforming traditional methods while providing interpretable results that can directly influence IVF outcomes.
Achieving 96.06% accuracy in laser welding penetration prediction with only 200 labeled images could revolutionize quality assurance in industrial applications.
AI-assisted workflows can cut down experiment reproduction efforts by up to six times, but struggle with complex analysis tasks requiring human oversight.
Moderately difficult research in NLP achieves greater academic impact, revealing a critical balance for researchers to target.
A dynamic load balancer can reduce node idle time to nearly a millisecond in complex UQ workflows, revolutionizing how we approach scheduling in high-performance computing.
