~/aixsci
weekly updates

learn/model families

Convolutional neural network

A convolutional neural network is a learning system built for data laid out on a grid, such as an image, that finds small local patterns and reuses the same pattern detectors everywhere.

What it is

Many kinds of scientific data sit on a grid: the pixels of a photograph, the slices of a three-dimensional scan, a plot of signal strength against time or frequency. A convolutional neural network, often shortened to CNN, works on such grids. It slides a small window across the data and, at each position, applies the same set of weighted sums, called a filter. One filter might respond to an edge, another to a blob, another to a diagonal streak. Because the same filter is used at every position, the network does not have to learn separately what an edge looks like in each corner of the image.

The outputs of those filters form a new grid, and further layers of filters run over that, so later layers respond to combinations of earlier patterns and to larger regions. Between layers the network usually shrinks the grid by keeping only the strongest response in each small patch, a step called pooling, and passes values through a simple rule that discards negatives. Whole blocks of layers are often stacked with shortcut connections that let information skip ahead, which makes very deep networks easier to train. At the end, the network can report a class, a position, or an outline.

Why scientists use it

Scientists reach for a CNN when the thing they want to read off the data is a shape or a texture rather than a number in a table. Deciding whether a smudge in a scan is a particle, or whether a trace is a real signal or an instrumental wobble, is work that people can do by eye but cannot easily write down as rules. A CNN is a way of learning those rules from examples that have already been labelled by hand.

It stands in, then, for the trained eye of an experienced observer, applied at a speed and volume no person could match. Modern instruments produce more images, spectra and scans than anyone can inspect. A CNN also tends to need fewer examples than a network that treats every pixel as unrelated, because reusing filters across positions builds in an assumption that is true of most physical data: a feature means much the same thing wherever it appears.

In the record

Limits

A CNN only learns what its labelled examples contain. Faced with an instrument, sample or sky it never saw in training, it may still report a confident answer, and it gives no warning that it is guessing. Its assumption that local patterns matter most makes it weaker when the meaning of the data depends on long-range relationships. It finds correlations, not causes, and can lock on to incidental features of how the data were collected. It also rarely explains itself, so its outputs usually need checking against independent measurements.

Written by AIxSci: the background from general knowledge, the examples from checked records.

By field

Named models

Every article

73articles
21structural-biology
6materials-chemistry
40astronomy
6climate-weather

aix-00283

Deep learning forecast of El Niño suggests Indian and Atlantic oceans extend warning time

forecastingsimulation surrogate

climate-weather · ai-result · Science Advances · 2026

aix-00284

Neural network turns hourly satellite heat images into Gulf Stream current maps

structure determination

climate-weather · ai-result · Nature Geoscience · 2026

aix-00287

Satellite image dataset extended to four channels for typhoon strength prediction

property prediction

climate-weather · ai-result · arXiv · 2026

aix-00294

Neural network trained on simple equations spots tipping points in real data

classificationforecasting

climate-weather · ai-result · Royal Society Open Science · 2022

aix-00298

Generative AI model forecasts UK rainfall 90 minutes ahead from radar images

forecasting

climate-weather · ai-result · Nature · 2021

aix-00299

Neural network forecasts weather straight from raw satellite and station observations

forecastingstructure determination

climate-weather · ai-result · Nature · 2025

aix-00278

Neural networks sort millions of stars to build a young-star candidate catalogue

classification

astronomy · ai-result · Astronomy and Astrophysics · 2026

aix-00274

Neural network trained on simulated data hunts solar wave trains in radio spectra

detectionclassification

astronomy · ai-result · Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences · 2026

aix-00219

Neural network supplies the sideways flows needed to measure the quiet Sun's magnetic energy

property prediction

astronomy · ai-result

aix-00221

Neural network measures rotation periods for Kepler's main-sequence stars

property prediction

astronomy · ai-result · arXiv · 2026

aix-00223

Neural networks weigh cluster X-ray maps to estimate galaxy cluster masses

property prediction

astronomy · ai-result · Monthly Notices of the Royal Astronomical Society · 2023

aix-00215

Neural network ranks gravitational wave triggers in search for neutron star mergers

detection

astronomy · ai-result · Physical review. D/Physical review. D. · 2025

aix-00245

Neural network trained on survey data retuned to spot transients for a liquid mirror telescope

classification

astronomy · ai-result

aix-00248

Image classifiers flag a gas kink in a young star's disc, pointing to a planet

classification

astronomy · ai-result · The Astrophysical Journal · 2023

aix-00251

Astronomers map the mass of galaxy cluster Abell 2744 using lensed images

classification

astronomy · ai-supporting · Astronomy and Astrophysics · 2022

aix-00254

Neural networks read a fast radio burst's dispersion straight from its spectrum

property prediction

astronomy · ai-result

aix-00209

Neural network turns Kaguya camera images into global maps of lunar surface chemistry

property prediction

astronomy · ai-result · Nature Communications · 2023

aix-00153

Neural networks pick amino acid sequences for short peptides in protein binding sites

candidate generation

structural-biology · ai-result · Bioinformatics · 2022

aix-00152

Neural networks trained to find the cutting site on serpin proteins

segmentation

structural-biology · ai-result · Bioinformatics Advances · 2026

aix-00150

Deep learning sorts crystal structures for single-molecule magnet behaviour

classificationproperty prediction

materials-chemistry · ai-result · IUCrJ · 2024

aix-00167

Neural network scans eleven years of night-sky images for mesospheric wave fronts

detection

astronomy · ai-result · Earth Planets and Space · 2025

aix-00168

Deep learning reads cryo-EM maps alongside AlphaFold3 models to build protein structures

structure determinationdetectionclassification

structural-biology · ai-result

aix-00171

Neural network fills gaps in protein maps from raw diffraction data

structure determination

structural-biology · ai-result · Acta Crystallographica Section D Structural Biology · 2025

aix-00138

AI-labelled protein pairs used to test how well function predictors handle unknown proteins

classification

structural-biology · ai-result · Bioinformatics · 2025

aix-00136

Diffusion model generates synthetic images of the Sun sorted by flare strength

candidate generationclassification

astronomy · ai-result · arXiv · 2024

aix-00134

Meta-learning lets a cosmology emulator adapt to new galaxy depth maps quickly

simulation surrogate

astronomy · ai-result

aix-00132

Five codes measured galaxy shapes in simulated Euclid images, one using neural networks

property predictionsimulation surrogate

astronomy · ai-supporting · Astronomy and Astrophysics · 2022

aix-00181

Protein language model embeddings used to predict succinylation sites in proteins

classification

structural-biology · ai-result · Scientific Reports · 2022

aix-00183

Neural network predicts nanostructure optics to design a two-colour laser collimator

property predictionsimulation surrogate

materials-chemistry · ai-result · Nanophotonics · 2023

aix-00184

Machine learning scores short protein stretches that mark proteins for destruction

classification

structural-biology · ai-result · Briefings in Bioinformatics · 2024

aix-00190

A transformer model segments cryo-electron tomograms from three viewing directions at once

segmentationdetection

structural-biology · ai-result · Knowledge-Based Systems · 2025

aix-00192

Study tests whether AI clean-up of cryo-EM maps harms ligand density

denoising

structural-biology · ai-result · Frontiers in Molecular Biosciences · 2024

aix-00129

Neural networks sort supernovae and estimate their distances from brightness alone

classificationproperty predictiondenoising

astronomy · ai-result · arXiv · 2024

aix-00128

Phage proteins drawn as pictures, then sorted by image-recognition networks

classification

structural-biology · ai-result

aix-00125

Classifying stars by type from a single wide-band telescope image

classificationstructure determination

astronomy · ai-result · Astronomy and Astrophysics · 2025

aix-00123

Neural networks swap places with black hole image simulations, both directions

simulation surrogateproperty prediction

astronomy · ai-result

aix-00121

Machine learning sorts mitochondrial proteins into three compartments from sequence alone

classification

structural-biology · ai-result · Journal of Environmental and Public Health · 2022

aix-00118

Neural networks replace a slow statistical fit to find solar oscillations in Hα images

simulation surrogate

astronomy · ai-result · Astronomy and Astrophysics · 2025

aix-00106

Volunteer labels train a model to find asteroid trails in Hubble archive images

detectionclassification

astronomy · ai-result · Astronomy and Astrophysics · 2022

aix-00104

Neural networks sort 400,000 JWST galaxies to trace when spirals and spheroids appeared

classification

astronomy · ai-result · arXiv · 2025

aix-00100

Astronomers test pre-trained image models for sifting telescope alerts

classificationanomaly detection

astronomy · ai-result · arXiv · 2025

aix-00089

Neural networks trained on atomic surroundings score how mutations change proteins

property predictionstructure determination

structural-biology · ai-result · arXiv · 2024

aix-00086

Neural networks trained on simulated radio bursts sort real bursts by shape

classification

astronomy · ai-result · arXiv · 2025

aix-00082

Neural networks turn simulated cluster mass maps into X-ray and SZ images

simulation surrogate

astronomy · ai-result · RAS Techniques and Instruments · 2025

aix-00079

Changing the training penalty improves how a network spots beta-sheets in blurry protein maps

segmentation

structural-biology · ai-result · Bioinformatics Advances · 2024

aix-00077

Neural networks sift 812,118 quasar spectra for hidden gravitational lenses

classificationproperty prediction

astronomy · ai-result · The Astrophysical Journal · 2026

aix-00076

Neural network strips mains-power hum from LIGO data in about a second

denoising

astronomy · ai-result · arXiv · 2023

aix-00073

Neural networks sort eleven years of Mars plasma data into three regions

classification

astronomy · ai-result · arXiv · 2026

aix-00065

Model predicts how fast an enzyme works on a substrate, with uncertainty attached

property predictionexperimental design

structural-biology · ai-result · Briefings in Bioinformatics · 2025

aix-00056

Deep learning tool predicts how single mutations weaken protein partnerships

property prediction

structural-biology · ai-result · Nucleic Acids Research · 2024

aix-00055

Neural networks sift 315,000 galaxy images for rare polar ring systems

classification

astronomy · ai-result · Astronomy and Astrophysics · 2025

aix-00049

Eight machine learning methods sorted supernova gravitational wave signals by nuclear physics model

classification

astronomy · ai-result · Machine Learning Science and Technology · 2024

aix-00043

Neural network maps cosmic voids, walls and filaments from sparse galaxy tracers

segmentation

astronomy · ai-result · The Astrophysical Journal · 2026

aix-00042

Two unsupervised methods compress fast radio burst spectra into a handful of numbers

denoisinganomaly detection

astronomy · ai-result

aix-00041

Astronomers hunt dust-buried young star clusters in eleven nearby galaxies

classification

astronomy · ai-supporting · The Astronomical Journal · 2025

aix-00039

Machine learning predicts whether a dying massive star explodes, from its density profile

classification

astronomy · ai-result · The Astrophysical Journal Letters · 2022

aix-00036

Four million simulated X-ray patterns used to test 21 crystal-symmetry classifiers

classification

materials-chemistry · ai-result · arXiv · 2024

aix-00032

Neural network turns galaxy shape distortions into maps of invisible cluster mass

structure determination

astronomy · ai-result · The Astrophysical Journal · 2025

aix-00031

Keck spectra pin down distances for six gravitational lenses found by a neural network

classification

astronomy · ai-supporting · The Astrophysical Journal · 2026

aix-00030

Software spotted a nearby supernova and booked a telescope within minutes

classificationexperimental design

astronomy · ai-result · The Astrophysical Journal · 2025

aix-00029

Cryo-EM maps show water shifting in actin filaments as they age

detection

structural-biology · ai-supporting · Nature · 2022

aix-00026

Algorithm picks starting powders and temperatures for making inorganic materials

classificationexperimental design

materials-chemistry · ai-result · Nature Communications · 2023

aix-00021

Neural network reads a million star spectra to measure temperature and chemistry

property prediction

astronomy · ai-result · Monthly Notices of the Royal Astronomical Society · 2022

aix-00020

Neural network sifts Kepler signals and validates 301 new exoplanets

classification

astronomy · ai-result · The Astrophysical Journal · 2022

aix-00019

Neural network sorts six hundred thousand noise blips in LIGO gravitational-wave data

classification

astronomy · ai-result · Classical and Quantum Gravity · 2023

aix-00017

Neural network picks bright exploding stars from sky survey alerts and books telescope time

classificationexperimental design

astronomy · ai-result · The Astrophysical Journal · 2024

aix-00013

Neural network segments 650 battery particles to measure their detachment after cycling

segmentationdetection

materials-chemistry · ai-result · Nature Communications · 2020

aix-00010

Robotic lab with machine learning proposes and runs 353 inorganic synthesis experiments

candidate generationproperty predictionclassificationexperimental design

materials-chemistry · ai-result · Nature · 2023

aix-00008

Neural network locates single protein particles in cryo-electron microscope images

detection

structural-biology · ai-result · Communications Biology · 2019

aix-00005

Cryo-tomography maps the myosin filament inside relaxed mouse heart muscle

structure determinationdetectiondenoising

structural-biology · ai-result · Nature · 2023

aix-00003

Researchers map the sixteen-part Commander complex using crystals, cryo-EM and AlphaFold2

structure determinationdetection

structural-biology · ai-result · Cell · 2023

aix-00002

Deep learning turns raw NMR spectra into protein structures without human intervention

detectionproperty predictionstructure determination

structural-biology · ai-result · Nature Communications · 2022

aix-00001

Researchers measure folding stability for hundreds of thousands of protein variants

structure determinationcandidate generationproperty predictionclassification

structural-biology · ai-supporting · Nature · 2023