Search papers, labs, and topics across Lattice.
100 papers published across 2 labs.
Agreement among perturbed inputs can mislead accuracy assessments, as fine-tuning on consensus can paradoxically reduce performance.
Standard Young tableaux can reveal nuclear-spin symmetry without complex projections, streamlining molecular spectroscopy analysis.
An exact closed-form update for orthogonality-constrained optimization could revolutionize efficiency in matrix-aware methods like Muon.
Surv-IPTB outperforms traditional survival analysis methods by accurately estimating treatment benefits in complex, nonlinear patient data.
FALM-PINN reduces relative \(L^2\) errors by up to 100 times compared to current benchmarks in solving high-frequency nonlinear PDEs.
An exact closed-form update for orthogonality-constrained optimization could revolutionize efficiency in matrix-aware methods like Muon.
Surv-IPTB outperforms traditional survival analysis methods by accurately estimating treatment benefits in complex, nonlinear patient data.
FALM-PINN reduces relative \(L^2\) errors by up to 100 times compared to current benchmarks in solving high-frequency nonlinear PDEs.
Predictive modeling can slash nanodrug development time and costs by accurately forecasting nanoparticle properties with minimal empirical data.
SEAM reveals that even accurate local predictions can mask significant global inconsistencies in scientific explanations, challenging traditional validation approaches.
Agreement among perturbed inputs can mislead accuracy assessments, as fine-tuning on consensus can paradoxically reduce performance.
Zero-loss exactness in optimal transport dynamics could revolutionize how we approach matching problems in high-dimensional spaces.
Current LLMs only partially grasp epitope information, revealing significant gaps in their utility for antibody drug discovery.
Ant-Q slashes circuit loading overhead to near zero, unlocking the potential for deeper quantum circuits and faster experimental throughput.
Overlapping electronic states in YbO lead to unexpected vibrational irregularities that challenge conventional understanding of molecular spectra.
Small molecular systems can deliver accurate shear viscosity results at a fraction of the computational cost, transforming simulation strategies in fluid dynamics.
Surface electric fields can dictate the formation of key interstellar ions, bridging astrophysical chemistry and surface catalysis.
Nonlinear modal synthesis can now be computed with reduced costs while retaining high fidelity in frequency control, revolutionizing simulations of musical instrument dynamics.
Standard Young tableaux can reveal nuclear-spin symmetry without complex projections, streamlining molecular spectroscopy analysis.
The first algorithm for constrained density functional theory with rigorous convergence guarantees achieves high accuracy and efficiency without compromising robustness.
RxnCLF outperforms traditional models by leveraging a unified reaction graph, revealing a new paradigm for reaction yield prediction that enhances interpretability and generalization.
Graph propagation over a temporal knowledge graph can predict clinical advancement with unprecedented accuracy, especially for cases lacking direct evidence.
MetaboLLM-GIN not only outperforms conventional models in predicting stress hyperglycemia but also transforms biochemical knowledge into interpretable metabolite graphs.
Advanced forecasting models can cut anomaly detection errors by up to 42%, transforming how we secure research networks against real threats.
A simple regularity condition can unlock powerful validation techniques for conditional mean embeddings across diverse mathematical applications.
Kastor reduces forecasting error by nearly 43% while enhancing the physical fidelity of simulations, outperforming traditional methods in both accuracy and efficiency.
Achieving 5.75-fold higher fine-tuning throughput, BioM-JEPA redefines how we approach representation learning in single-cell transcriptomics.
Operator-residual feedback slashes the rate of misleading score-only decisions from nearly 40% to under 2%, ensuring that autonomous agents make choices grounded in physical reality.
Achieving unprecedented accuracy in divertor heat-flux reconstruction, SafeDivertor outperforms traditional methods by directly utilizing plasma-state signals during discharge.
Quantum one-way functions could redefine the landscape of cryptographic security, offering new primitives that withstand adversarial attacks better than their classical counterparts.
Pretraining on relevant scientific images can significantly boost quality assessment performance, outperforming larger datasets.
Linking electronic momentum changes directly to nuclear dynamics could revolutionize our understanding of electron-phonon interactions in solid-state physics.
Quantum circuits can now capture the complex energy landscapes of proteins, revealing ensemble-level insights that traditional methods miss.
Tracking Reeb space sheets reveals persistent structures in time-varying bivariate fields, unlocking new insights into dynamic data relationships.
MT-GNN achieves a groundbreaking 2.29% improvement in predicting brain morphology, setting a new standard for accuracy in clinical imaging.
Extreme response estimation in structural optimization just got a major upgrade with the GSS-SPCE framework, slashing computational costs while enhancing accuracy.
LLMs can cut kinetic model discovery iterations by up to 79% while maintaining high predictive accuracy, revolutionizing the efficiency of chemical engineering research.
Fine-tuning Earth-based weather models can unlock accurate Martian weather predictions within just 10 training epochs, transforming our approach to planetary atmospheric forecasting.
The assumption that SCMS converges to a static ridge is fundamentally flawed, revealing a need for a new understanding of density ridge extraction.
Active learning can cut candidate evaluation space in half while retaining nearly all valuable design options in materials optimization.
Myopic planners can fail spectacularly when faced with the need to acquire capabilities for future experiments, leading to unbounded approximation ratios in goal-directed discovery.
CL-PINN outperforms traditional methods by balancing accuracy and efficiency, making it a game-changer for solving parameterized PDEs without observational data.
Achieving expressivity beyond the 1-WL limit, this method transforms 3D neuronal data into cyclic metric graphs, enhancing graph learning without additional training.
RASPL outperforms traditional methods by leveraging formula preservation, achieving superior robustness in soil-loss predictions under degraded conditions.
Language models are far more capable in scientific coding than previously thought, with corrected evaluations revealing accuracy improvements of up to 92%.
ORACLE achieves up to 104.4x faster circuit design while meeting nearly all target specifications, revolutionizing multi-objective optimization in analog circuit design.
EviGraph boosts the reliability of autonomous research agents by ensuring every claim is grounded in a validated evidence chain, leading to a 40% increase in claim support.
PSPACE-completeness of the reachability problem in 3-VAS reveals a crucial boundary in computational complexity that was previously unresolved.
Achieving superior anatomical coherence and visual realism in 3D MRI synthesis, VoxStruct3D redefines the landscape of volumetric generation.
Stark spectroscopy in water reveals how solvent interactions can dramatically alter molecular dipole responses, challenging previous assumptions about molecular behavior in polar environments.
Minor tweaks in rotational energy can drastically alter ionization dynamics and product distributions in $\mathrm{CF_2I_2}$, revealing a new layer of control in molecular fragmentation processes.
Traditional estimators for binding affinities can be off by nearly 1 kcal/mol, revealing a critical flaw in how bound states are defined in molecular simulations.
Achieving reaction energy predictions with unprecedented accuracy using a symmetry-guided quantum algorithm could revolutionize our approach to simulating complex chemical reactions.
Shifting the focus from regulating AI use in research to fundamentally rethinking the infrastructure of scholarly communication could redefine trust in academic publishing.
Achieving 87.71% negotiation accuracy, this architecture revolutionizes how scientific workloads are managed across diverse computational environments.
SparseDitto achieves up to 146.61x speedup for sparse matrix operations by dynamically customizing GPU kernels based on input patterns.
Ground-state diatomic molecular anions exhibit unexpected crossings with excited states that could significantly boost electron attachment in ultracold experiments.
CheMLFlow slashes the orchestration overhead in scientific machine learning, enabling researchers to focus on their core contributions without getting bogged down in workflow complexities.
Rethinking failure metrics in scientific computing could drastically enhance resource efficiency in Exascale systems.
Spectral fingerprints can differentiate between unique molecular structures with identical 2D connectivity, revolutionizing how we assess chemical similarity.
SC-UCG reveals that accurate modeling of phase transitions in complex biomolecular systems can be achieved without predefined collective variables, challenging traditional coarse-graining methods.
Classical methods can outperform quantum sampling in configuration interaction, challenging the presumed superiority of quantum approaches in pre-fault-tolerant quantum chemistry.
Infinite clusters are impossible in finite time for large enough cluster sizes, but a surprising blowup occurs for more negative rates.
A novel variational approach allows for accurate polaron simulations at unprecedented computational efficiency, bridging the gap between model systems and real materials.
LMO-based wave function expansions can reduce quantum gate requirements for Hamiltonian simulations from polynomial to polylogarithmic growth, offering a game-changing efficiency boost for large-scale quantum simulations.
Journals with larger author teams see a boost in citation impact, but AI engagement only drives Impact Factor in specific years.
A novel neural network architecture captures complex contraction dynamics of engineered muscle tissues, achieving high fidelity in parameterization even with limited labeled data.
AM generates smooth trajectories efficiently, sidestepping costly preprocessing and simulation, while achieving competitive results against existing methods.
CheMatE achieves competitive performance by seamlessly integrating chemical structure with natural language understanding, challenging the notion that domain-specific models must sacrifice generality.
CIR-ACTIVA reveals that causal forecasting can outperform traditional methods, allowing for precise multi-horizon predictions in financial markets without retraining.
Structured offline learning can dramatically enhance decision-making in oncology clinical trials, with agents outperforming traditional tools by a significant margin.
Uncertainty quantification in NIROMs can now be achieved with a framework that guarantees reliable predictions even in challenging extrapolation regimes.
ED-DiT achieves a remarkable reduction in prediction error, showcasing the power of physics-guided pretraining in molecular representation learning.
LLMs are transforming PDE workflows, but their effectiveness is hampered by data scarcity and the challenge of applying simulations to real-world scenarios.
Most LLMs fail to design high-quality experiments, revealing a critical gap in AI's role in scientific research.
Semantic equivalence measures can drastically change how we quantify annotator agreement in biomedical text, revealing hidden biases in existing methods.
ANCHOR-RE boosts biomedical relation extraction performance by integrating neuro-symbolic reasoning, achieving precision gains without the need for model fine-tuning.
LLMs exhibit a troubling tendency to generate analogies from a narrow set of domains, limiting their creative potential and cross-domain connections.
Statistical measures that claim to identify language in undeciphered scripts may merely reflect organizational patterns, not linguistic content.
EGMC encryption schemes, once thought secure, can be broken in under 10 minutes, exposing critical vulnerabilities in their design.
Unsupervised domain adaptation reveals a surprising link between uterine contractility and oxygenation patterns, challenging traditional views on gynecological tissue dynamics.
Achieving 93.02% accuracy in molecular structure recognition, MinerU.Chem outperforms existing systems, unlocking new possibilities for AI-driven chemistry research.
SHM-POMDP achieves 2.5x higher information gain in geologic exploration while maintaining principled uncertainty quantification.
Surface potential can become nonmonotonic with charge density, leading to unexpected first-order transitions in electric double layers.
TNASS achieves superior accuracy in multi-scale modeling by automating active space selection, outperforming traditional methods in both ground state energy and dipole moment predictions.
NanoMorph-3D achieves unprecedented reconstruction fidelity and speed by seamlessly integrating physics-driven modeling with advanced attention mechanisms, transforming how we analyze nanomaterials.
Mechanism recovery in AI-generated hypotheses falters in early reasoning stages, revealing vulnerabilities in scientific validity that could mislead experimental planning.
AI scientists excel at idea generation but falter in filtering and prioritizing innovations, revealing a critical gap in their capabilities.
A novel paired recipient-based evaluation reveals that survival prediction models can accurately forecast post-transplant years gained, challenging traditional metrics in kidney transplant outcomes.
EPIK transforms how we incorporate prior knowledge in Bayesian verification, leading to more accurate assessments of software reliability and performance.
Simulating molecular dynamics on noisy quantum devices is now feasible, with results converging to exact solutions in harmonic systems.
FedCARE achieves up to 12.5% better predictive accuracy by enabling personalised model adaptations in federated healthcare settings without compromising data privacy.
Hybrid retrieval methods can achieve perfect recall in scientific question answering, but domain mismatch can undermine precision when using general-purpose rerankers.
Achieving consistent unit commitment solutions across multiple quantum processors could revolutionize how we tackle complex optimization problems in energy systems.
Duschinsky rotation can dramatically shift the spectral signatures in two-dimensional resonance Raman spectroscopy, revealing hidden molecular dynamics.
Nontrivial trace-product switchings of the Gold function are confined to just three even dimensions, challenging assumptions about their prevalence in higher dimensions.
PILOT redefines biomedical entity linking by achieving state-of-the-art results while maintaining efficiency, overcoming the complexities of ambiguous mentions and diverse annotation standards.
Language models struggle with lab-relevant tasks, but onepot-Bench 0 reveals critical gaps in their decision-making abilities that could impact real-world applications.
Reparameterizing optimization problems on simplices leads to a more efficient and accurate method for functional data registration, outperforming traditional approaches.
Unambiguous DNFs can achieve a certificate complexity separation that leads to groundbreaking improvements in communication complexity results, including the Clique versus Independent Set problem.
Fragmented experimental data can be transformed into actionable insights for plastic upcycling, achieving unprecedented accuracy without biased imputation.
Integrating physics with data-driven methods could revolutionize the robustness and interpretability of cardiovascular digital twins.
Explicit spectral representation in Raman spectroscopy can lead to a 19.6% improvement in predictive accuracy over traditional methods.
Adaptive sampling can dramatically enhance fidelity estimation in bosonic quantum states, achieving reproducible results in minutes while being resilient to phase space transformations.
Identifying nonproperness sets can drastically improve the efficiency of real root classification in likelihood equations, revealing hidden structures in statistical models.
