Materials and AI glossary

Predicting the strength of a material from its composition and processing conditions.

Using conductivity measured for each composition as the answer to learn from.

Grouping samples whose spectra look alike.

Learning the features of a defect directly from microscope images.

Extracting synthesis conditions from a paper and then checking them against the original text.

Proposing candidate crystal structures that match a target property.

Turning paper abstracts into vectors to find related work.

Predicting the properties of a new composition with a model that has finished training.

Fine-tuning an atomistic model pre-trained across many element systems onto one specific alloy.

Adjusting a model trained on large-scale DFT data using a handful of measured values.

Using the mean atomic radius as a descriptor of a composition.

Predicting the band gap of a material in eV.

Separating sample images into sound and defective.

Matching known alloys closely while the error grows for new compositions.

Keeping related samples in the same group so no evaluation information leaks.

Choosing tree depth using validation data.

Raising the polynomial degree lowers the bias but raises the variance.

Judging whether collecting more data would help performance.

Keeping only the descriptors that matter with Lasso when there are many of them.

Approximating the lattice constant as a first-order expression in the composition ratios.

Predicting hardness from a few hundred alloy measurements.

Predicting yield from tabular process data.

Reporting large predictive uncertainty in regions where few experiments exist.

Learning the non-linear relationship between several process variables and a property.

Predicting properties by treating atoms as nodes and neighbouring pairs as connections.

Used as the backbone for models over composition sequences or over sentences in papers.

Reducing dozens of descriptors to two axes to look at how compositions are distributed.

Grouping measured spectra by type to separate families of samples.

Drawing crystal-structure embeddings as a two-dimensional map to inspect the clusters.

Compressing spectra and using the latent representation to find unusual samples.

Moving through the latent space to produce new candidate compositions.

Starting from noise and generating candidate crystal structures.

Learning a concentration profile that satisfies the diffusion equation.

Finding a relationship between dimensionless numbers from measurements.

Evolving layer counts and thicknesses to reach a target reflectance.

Learning the order and the conditions of a synthesis process against a reward.

Selecting a set of candidates that weighs strength and ductility together.

Ranking candidates by predicted band gap with an error range attached.

Separating the scatter that repeat measurements do not reduce from the uncertainty of an unexplored region.

Reporting a predicted strength together with a 90 percent interval.

Confirming on validation data that 90 percent of observations really fall inside the 90 percent interval.

Seeing whether temperature or hold time governs the yield more.

Keeping only the two of five process variables with a large total index for detailed experiments.

Feeding in the input error distributions to obtain the distribution of the predicted property.

Seeing how far an error in measured thickness grows in the calculated conductivity.

Finding the descriptor whose shuffling raises the error the most.

Seeing why the predicted strength came out high for one composition, contribution by contribution.

Leaving the test data untouched until the result is reported.

Scaling on the whole dataset first lets information from the test data seep in.

Removing the size difference between temperature in degrees and time in hours before fitting.

With defective samples at 2 percent, answering “sound” every time still scores 98 percent accuracy.

Reading RMSE alongside MAE, because RMSE reacts more strongly to large errors.

Favouring recall when missing a defect is the costly outcome.

Using a factorial design to see interactions instead of changing one variable at a time.

Exploring a five-variable process space evenly with twenty points.

Treating a prediction for a composition containing an unseen element as indicative only.

Recording measurement conditions and units as standard fields before publishing.

Checking the computed properties of candidate compositions before choosing what to make.

Defining how material, process and property relate so that data from different labs merges onto the same fields.

Aligning datasets that recorded the same property under different names onto standard concepts.

Linking composition, process, property and literature to trace the processing route that produced a property.

Finding “samples containing this element that went through this process” in one path query.

Comparing your own model against published results on the public split.

Comparing different structures made of the same elements.

Computing the energy of candidate crystals to compare their stability.

A negative value still leaves room for decomposition into other compounds being more favourable.

Looking first at candidates near zero, without treating that as the same thing as being synthesisable.

Used as the figure of merit when comparing semiconductor candidates.

Sorting crystal structures by their symmetry.

Following atomic diffusion over time at different temperatures.

Checking which phase forms first at a given alloy composition.

Comparing a measured pattern with computed ones to decide which phase formed.

Measuring the grain size distribution from an image and relating it to strength.

Gathering experimental and computed data to analyse composition-property relationships.

Using the mean and the spread of the constituent elements’ electronegativity as input.

Connecting neighbours within a cutoff radius as edges to express the structure.

Computing diffusion in a system of thousands of atoms with a trained potential.

Generating candidates that would meet a target permittivity.

Computing the materials whose predictions are least certain and adding them to the training set.

Choosing synthesis temperature and composition while spending as few experiments as possible.

Picking the composition with the largest expected improvement as the next experiment.

Narrowing candidates with a property model instead of running DFT every time.

Cutting ten thousand candidates to a hundred with a predictive model before computing and measuring.

Changing the next composition proposal according to the newest experimental result.

Having AI analyse a robot synthesis result and set the conditions for the next run.

Normalising the ways synthesis temperature is written into standard units to build a dataset.

In an optical classifier using metasurfaces, a diffractive neural network directly processes the input image.

It predicts material properties by feeding both the sequence representation and the 3D structure of a polymer together.

In a porous framework, even with the same composition, different topology can lead to completely different properties.

The defect topology of oxygen vacancies can change charge trapping behavior.

Carrier recombination pathways in photoexcited oxides can be tracked with Nonadiabatic Molecular Dynamics.

It can be used to compare how much the charge mobility changes when one substituent is changed in a candidate material molecule.

If you find a vulnerability in the research AI platform, you can receive a reward through the bug bounty program.

In nanowire conductance, the higher the topological visibility, the easier it is to identify Majorana features.

The concept of Demand Response is important in data center operation, where large computing resources are adjusted to match power grid conditions.

It can be used in a materials design interface where the results are updated immediately when experimental conditions are changed.

In a materials process, a virtual simulator can be constructed that reflects temperature changes and equipment operation together.

The response policy of an autonomous laboratory agent is designed in accordance with the Model Spec.

By learning the charge density obtained from DFT, it is possible to rapidly estimate the electronic structure properties of materials.

An equivariant graph neural network is used when predicting charge density or forces while preserving the rotational symmetry of a crystal structure.

It learns material representations by aligning crystal structures and text descriptions close to each other in the same space.

In automated process data, sensor outliers are removed to improve the performance of a regression model.

When predicting the structure of a disordered protein, a conformational ensemble is generated instead of a single structure.

The overall trend of a catalytic surface and the differences among individual active sites can be modeled together with Hierarchical Machine Learning.

In nanoparticle catalysts, the reaction differences among different active sites are interpreted as Site Heterogeneity.

From the body text of a paper, the mixing order of the solution and the heat-treatment conditions are extracted to create an experimental protocol database.

The prediction performance of a materials AI model can be re-verified by an external institution to increase reliability.

It is used when predicting continuous properties such as band gap from composition or structural features.

It is used when constructing a computation pipeline that jointly interprets surface catalyst structures and spectroscopic signals.

It can be used to virtually simulate how the microstructure will change when material processing conditions are altered.

It is used when interpreting defect distribution in a microscope image to determine the next analysis step.

It is used when quickly evaluating the diffusion of electrolyte molecules and the solvation structure.

When materials data are limited, evidential deep learning can provide both the predicted value and the uncertainty for candidate compositions.

In materials sensors, SERS is used to detect low concentrations of contaminants or biomarkers.

When training a large model on sensitive materials data, the training server is separated from the external internet.

Real-space Representation is used in wavefunction approximations that directly handle the electron distribution around atoms.

In crystal structure reconstruction, interatomic distances or sample symmetry can be incorporated as a physics constraint.

The three-dimensional atomic arrangement of a nanomaterial can be reconstructed by Electron Tomography and then corrected with AI.

It is used when generating new structures by representing a crystal structure as a combination of motifs instead of at the atomic level.

A dashboard screen draft can be automatically generated from a request to view experimental data.

In indoor air quality measurement, a sensor estimates the concentration of a specific volatile organic compound.

An electronic nose is used when classifying volatile organic compound types from the response pattern of a gas sensor array.

It learns latent representations from raw crystal structure inputs that are advantageous for property prediction.

In an extraction process, the Liquid-Liquid Phase Diagram is used to find the composition range in which separation proceeds well.

Sensitive experimental data and automation equipment are made available only to authorized researchers through access control.

In a qubit array model, the onsite interaction strength can be inversely estimated from data.

In a semiconductor quantum dot array, site-to-site deviations in onsite energy can be estimated.

In charge stability analysis of a quantum dot qubit array, Hubbard model parameters can be estimated.

When AI is introduced into an autonomous laboratory, governance procedures must be designed together with model accuracy.

In a materials literature summarization tool, a jailbreak attempt may be used to elicit prohibited experimental procedures.

An external validation team performs more detailed model safety testing with Employee-like Access.

It is used when a defect detection model trained in laboratory A is calibrated to field data from a different process condition.

In an autonomous laboratory, auditability is ensured by recording which model recommended the experimental conditions and on what basis.

A literature summarization agent must be protected so that it does not fall for hidden instructions on a webpage and draw incorrect conclusions.

In materials inverse design, local electronic descriptors around atoms can be generated together to reflect material properties.

Crystal structures and local electronic descriptors can be jointly generated to design candidate materials.

The surface adsorption energy and reaction barrier are calculated under constant potential conditions.

A new materials recommendation model can reduce overestimation through Independent Evaluation.

In perovskite solar cells, interfacial passivation reduces recombination and improves efficiency and stability.

A Line Graph Neural Network can be used to predict electronic structure by reflecting bond-angle information in a crystal structure.

The change in energy according to the relative sliding position of the bilayer is represented by a PES to examine stacking stability.

Machine-learning-assisted Molecular Dynamics · ML-assisted MD can be applied to calculate ion migration barriers at a solid electrolyte interface.

In crystal structure analysis, it calls a diffraction pattern processing program to find peaks.

It sequentially calls beamline control software and measurement tools to perform the initial experimental procedure.

It explains the cause of property changes through the relationship between structure and composition.

In a composite coating, if particle dispersion is not uniform, the resulting heterogeneous formulation can lead to large property variations.

The repeating units and substituent information in a polymer composition are split into fragments and used as input for property prediction.

In a microscope image, noise is reduced to restore fine structures more clearly.

In electron beam measurements, real-time data analysis can immediately exclude scans of poor quality.

In X-ray scattering experiments, an intelligent detector can immediately detect abnormal signals and adjust the measurement time.

The open-source dependencies of the materials analysis pipeline are checked with Vulnerability Scanning before deployment.

A Small Language Model can be attached to a materials information workflow that repeatedly performs large-scale literature searches.

The candidate magnetic structure is checked to see whether it is kinetically stable in calculation.

The phase stability of an oxide with an interface is compared using atomistic simulation.

A researcher uses the interface of an AI tool to more easily carry out literature searches and data organization.

Pore structure and conductive distribution in aerogels or composites are controlled by additive manufacturing.

The binding characteristics of the passivation molecule can be evaluated by Quantum Modelling to narrow down candidates.

In perovskite solar cells, candidate molecule screening and process optimization can be integrated into a single Closed-loop Framework.

When adapting an existing MLIP with a small amount of materials data, LoRA can be used to lower the training burden.

Multi-head Fine-tuning can be applied so that phonons and thermal and elastic properties are predicted together in the same ML potential.

When there are many candidate composition descriptors, Recursive Feature Elimination can be used to retain only the key variables.

Physics-Guided Active Learning can be designed to prioritize experimental evaluation of composite candidates with large interfacial barriers.

Joint-Embedding Predictive Architecture can be applied to pretraining that predicts the masked parts of 3D atomic structures.

When deploying a model to a laboratory automation system, the stability of the connection with the equipment must also be checked.

Cluster analysis is used on SEM image features to group candidate microstructures that are similar to one another.

The order parameter field of the material microstructure can be predicted to analyze changes in the domain structure.

Adsorption heat is used to evaluate how strongly hydrogen binds in MOF pores.

When comparing MOF hydrogen-storage candidates, adsorption isotherms at different temperatures are calculated.

When evaluating the hydrogen-storage performance of porous MOF, gas adsorption isotherms are calculated.

In a material recommendation system, it checks so that an advertisement for a specific product is not attached to a sensitive conversation context.

A problem of finding the most stable combination among candidate structures can be viewed as Combinatorial Optimization.

When exploring a broad composition space, a Heuristic is used to select promising candidates first.

It is important when designing the system so that phase-field prediction performance is maintained even when the grid size differs.

It can be used to quickly predict the spatial distribution of a phase-field solution.

It quickly evaluates thousands of candidates in a magnetic composition space to identify promising compositions.

By conditioning a catalytic atomistic potential on system metadata, different reaction systems can be distinguished.

When fine-tuning an MLIP for catalysis with new composition data, the accuracy of the previous catalyst family can decrease.

A research assistant tool that performs literature searches, organizes experimental records, and enters data can be regarded as an Agent.

A robotic system that automatically adjusts synthesis conditions corresponds to nanocrystal synthesis Automation.

Polymer dielectric candidates can be screened by dividing them into substructures such as repeat units and functional groups.

When introducing an LLM for summarizing materials data, it is used to check in advance for sensitive information leakage and incorrect citations.

It is important when comparing materials synthesis conditions by putting several long papers and experimental records into the model at once.

After thin-film growth, thickness and defect distribution are checked as quality control metrics.

In 2D materials CVD, temperature and gas flow are adjusted by process control.

Extremely adversarial inputs are given to an autonomous experiment recommendation model to induce incorrect condition suggestions.

When labels are scarce in materials microscopy images, Self-supervised Learning is first used to learn representations.

In microstructure design, interfacial and crystallographic constraints are added to the loss function to guide feasible structures.

In microstructure inverse design, latent space sampling generates candidate structures outside the training data.

In rare-earth-doped materials, spectroscopic property is used to predict emission spectra and lifetimes.

In infection experiments, single-cell profiling can be used to compare different responses in each cell.

In spectroscopic property prediction, support vector regression can be used to model nonlinear relationships in small datasets.

AI generated a hypothesis for a new composition candidate based on the literature and experimental data.

A classification model is checked when it sounds plausible in its explanation but in fact misleads the user by bypassing prohibited features.

AI inference containing sensitive research data is performed in a Secure Enclave · Trusted Execution Environment to reduce access risk.

A piezoelectric tensor has different components depending on the crystal orientation, so it must be predicted in tensor form.

When evaluating polymer dielectric candidates, descriptors are generated for each substructure and used for screening.

The change in the polarization state of a Ferroelectric affects charge transport characteristics.

We analyze whether Carrier Dynamics change according to polarization changes through ultrafast simulations.

After photoexcitation, the ultrafast charge transfer in the material is simulated with Nonadiabatic Dynamics.

In solid-liquid interface simulations, an interatomic potential is used to predict energy and forces.

The reverse-design result is checked in the verification stage to confirm that it satisfies the process constraints.

In generative crystal structure prediction, it is revealed as a sharp performance drop when existing prototypes are removed.

In generative crystal structure prediction, it appears when candidate structures are generated by learning a structural vector field.

When analyzing heat transfer at an alumina interface, phonon scattering and transmission are considered together.

If phonons undergo inelastic scattering in interfacial heat transfer, the energy transfer pathway can change.

To quickly calculate the electronic dynamics of a photoexcited crystal, one can learn and use an Effective Hamiltonian.

One can first cluster candidate regions of a phase transition and then use a classifier to distinguish Majorana states.

Separate benchmarks are designed for each task, such as structural relaxation and polymorph ranking of organic molecular crystals.

An equivariance model is used in which, when a molecular structure is rotated, the predicted force vector also rotates in the same way.

In powder neutron diffraction, the magnetic diffuse scattering pattern can be used to back-calculate interaction parameters.

The weight matrix of an LLM can be decomposed by singular value decomposition to perform a low-rank approximation and fine-tuning initialization.

In catalyst screening, mechanistic constraints can be added to the loss function to reduce unrealistic candidates.

Unsupervised deep learning can be applied to reconstruct missing information in limited-angle tomography.

Exploring catalyst candidates in latent space can continuously suggest new composition combinations.

Orthogonal transformation can be used when transforming candidate material representations while preserving the structure of latent space.

Transition State Search is performed to predict the rate of a surface catalytic reaction.

Adsorption, surface diffusion, and bond formation are each treated as one elementary reaction.

In catalytic CO2 hydrogenation, the possible elementary reactions are gathered to construct the reaction network.

In electrocatalysis literature, it groups expressions with similar functions and connects them to the same candidate set.

In Ni-based superalloys, vacancy formation energy is predicted to compare diffusion behavior.

In catalyst activity prediction, Interpretable Machine Learning is used to explain which composition variables and process conditions are important.

In a materials paper, composition, process conditions, and property values are automatically extracted to build a dataset.

In nanocrystalline TEM analysis, strain mapping is used to examine local strain around defects.

In TEM images, it is used to distinguish particles, defects, and interfaces for automated measurement.

In-context Learning can be used to make the system follow the equipment operation sequence by placing example experimental procedures in the prompt.

In an autonomous laboratory, the process of recommending the next synthesis conditions and then updating the model again with the experimental results is Closed-loop Optimization.

When the alloy composition is changed, low phase stability can increase the formation of undesired precipitated phases.

When designing low thermal expansion alloys, the linear thermal expansion coefficient is predicted according to temperature change.

In Invar alloys, the Curie Temperature is an important property index when understanding the correlation between magnetism and thermal expansion.

Large-scale structure exploration is performed with a Machine Learning Potential · ML Potential to find candidate phase equilibria under high pressure.

Global Structure Search is carried out to find unknown compounds under high-pressure conditions.

When classifying 2D magnet candidates, Octahedral Crystal Field information can be used as a descriptor.

In alloy precipitation processes, the Order-Disorder Transition governs the formation of the initial phase.

In battery degradation diagnosis, the Virtual Reference Electrode is used to estimate the anode potential.