Finally! Accurate Protein Prediction Courtesy of Artificial Intelligence

September 7, 2021

The dream of predicting a protein shape just from its gene sequence is now a reality thanks to an artificial intelligence (AI) algorithm recently validated by a research collaboration including Berkeley Lab’s Molecular Biophysics & Integrated Bioimaging Division. Predicting protein shapes is a long sought-after breakthrough for structural biologists because it offers a key to understanding protein functions to accelerate treatments for diseases like cancer and COVID-19.

X-Ray Experiments, Machine Learning Could Trim Years Off Battery R&D

April 2, 2021

Berkeley Lab’s COSMIC X-ray instrument contributed to a battery study that used an innovative approach to machine learning to speed up the learning curve about a process that shortens the life of fast-charging lithium batteries.

Supernovae Twins Open Up New Possibilities for Precision Cosmology

May 6, 2021

Using machine learning techniques, Berkeley Lab cosmologists have found a way to double the accuracy of measuring distances to supernova explosions – one of their tried-and-true tools for studying the mysterious dark energy that is making the universe expand faster and faster.

Researchers Hunt for New Particles in Collider Data

Berkeley Lab physicists participate in exploration of three years of data from CERN’s ATLAS detector

February 19, 2021

Researchers are employing machine-learning algorithms to search for new particles in ATLAS detector data collected from 2015-18, without relying on any particle simulations.

Berkeley Lab’s Advanced Monitoring Capabilities Still in Use 10 Years After Fukushima Earthquake and Nuclear Power Plant Disaster

March 11, 2021

Eleven years after Berkeley Lab scientists began monitoring radiation loosed by an earthquake that struck the Fukushima nuclear powerplant in Japan, their monitoring efforts are still going strong. Now researchers are using machine learning to help improve the efficiency of their long-term monitoring efforts.

Berkeley Lab Scientist’s Work Deciphering Early Universe Secrets Garners 2021 Gruber Cosmology Prize

Uroš Seljak developed theoretical predictions and mathematical tools, including machine learning algorithms, for studying the early universe

October 2, 2020

A trio of physicists – including Uroš Seljak of Lawrence Berkeley National Laboratory (Berkeley Lab) – has been awarded the 2021 Gruber Cosmology Prize for their work studying the large-scale structure of the universe as well as the properties of its first instant of existence. Seljak's recent work incorporates machine learning and other advanced computational techniques critical to the success of upcoming cosmology experiments including DESI and CMB-S4

A Day in the Half Life: Machine Learning

February 08, 2021

In addition to making everyday life easier, machine learning (ML) algorithms are beginning to advance scientific and medical research in dramatic ways. In fact, the range of potential applications is so huge that the question has shifted from “Can we use machine learning to solve this?” to “Do we understand these algorithms well enough to use ML for this?” In this podcast, two Berkeley Lab ML experts discuss machine learning and its scientific applications.

Artificial Intelligence Finds More Than 1,200 Gravitational Lensing Candidates

February 2, 2021

A research team with participation by Berkeley Lab physicists has used artificial intelligence to identify more than 1,200 possible gravitational lenses – objects that can be powerful markers for the distribution of dark matter. The count, if all of the candidates turn out to be lenses, would more than double the number of known gravitational lenses.

AI for Efficient Flying Qubits

Project Aims to Up the Efficiency of Converting Optic Signals to Quantum Connections

December 2, 2020

How do quantum computers connect over fiber networks? Right now, not very well. A team of Berkeley Lab scientists led by post-doctoral researcher Mekena Metcalf is using artificial intelligence techniques to boost the efficiency of a key process that enables qubits to "fly" across optical fiber.

A Machine Learning Solution for Designing Optical Materials

(Photo Credit: PlasmaChem)

December 2, 2020

Understanding how matter interacts with light – its optical properties – is critical in a myriad of energy and biomedical technologies, but calculating these properties is computationally intensive. And the inverse problem – designing a structure with desired optical properties – is even harder.

Now Berkeley Lab scientists have developed a machine learning model that can be used for both problems – calculating optical properties of a known structure and, inversely, designing a structure with desired optical properties.

Driver-less Data Acquisition

Self-learning algorithm breaks new ground in data collection and analysis

October 2, 2020

A self-learning algorithm developed by the CAMERA group at Berkeley Lab has enabled researchers at the Institut Laue-Langevin to for the first time run an autonomous data analysis during a neutron scattering experiment.

Machine Learning Takes on Synthetic Biology

September 25, 2020

Berkeley Lab scientists are developing smart algorithms with the potential to reduce bioengineering development cycles from years to months or even weeks. The field of synethic biology is rife with possibilities that reach beyond today's consumer-products applications (from burgers to face creams) and into fields of medicine, agriculture, climate, energy, and materials.


Berkeley Lab's Aggregate Data Standard for AI Modeling

August 31, 2020

Much of today's scientific data comes in a form that machine learning algorithms can’t use effectively. Berkeley Lab's ENDURABLE aims to make complex datasets compiled from a wide variety of sources more useful to scientists using the latest artificial intelligence techniques to make new discoveries.

Clues to COVID-19 Treatments Could Be Hiding in Existing Data

Berkeley Lab Scientists Want to Find Them

July 15, 2020

Berkeley Lab computing and bioinformatics experts are working together to develop a platform that consolidates COVID-19 data sources and uses the unified library to make predictions – about potential drug targets, for example.

Can CT Scans Be Used to Quickly and Accurately Diagnose COVID-19?

June 1, 2020

Berkeley Lab data scientist Daniela Ushizima is exploring whether image recognition algorithms and a data analysis pipeline powered by machine learning can help accurately distinguish COVID-19 abnormalities in CT scans and chest X-rays from other overlapping respiratory illnesses.

Smart Farms of the Future

Making Bioenergy Crops More Environmentally Friendly

June 2, 2020

Berkeley Lab scientists are marrying machine learning with agriculture in field research that also includes molecular biology, biogeochemistry, environmental sensing technologies, and machine learning. This “farm of the future” could revolutionize agriculture and create sustainable farming practices that benefit both the environment and farms.

Mining the Literature for New Insights

Machine Learning Tool Sets Sites on COVID-19

April 28, 2020

A team of Berkeley Lab scientists who normally spend their time researching things like high-performance materials for thermoelectrics or battery cathodes have built a text-mining tool in record time to help the global scientific community synthesize the mountain of scientific literature on COVID-19 being generated every day.

Estimating COVID-19's Seasonal Cycle

May 20, 2020

Is COVID-19 seasonal like the flu, waning in warm summer months then resurgent in the fall and winter? Berkeley Lab has launched a new project to apply machine-learning methods to health and environmental datasets, combining high-res climate models and seasonal forecasts, to tease out the answer.

Seeing the Universe Through New Lenses

May 14, 2020

Berkeley Lab scientists are employing a neural network and an award-winning algorithm to sort through troves of ground-based space images and identify strong gravitational lenses. Created by galaxies and other massive space objects, the lense bend light to reveal distant objects and phenomena along the same path, making them powerful probes of dark matter, dark energy, and other cosmic mysteries.

Applying ‘Decision Science’ to Transportation Behaviors

March 12, 2020

Why do some people adopt electric vehicles, ride-sharing, ride-hailing, or online shopping while others don’t? Berkeley Lab researcher Anna Spurlock spearheads the WholeTraveler Transportation Behavior Study, a three-year project that uses machine learning techniques to try to understand what drives human choices in transportation.

Microbiome Matchmaker

December 2, 2019

A unique neural network architecture called mmvec (microbe-metabolite vectors) is opening new windows into the complex microbial communities in the environment and our bodies. Such "microbiomes" are devilishly difficult to untangle, but doing so could help us understand processes that give rise to many biological and medical mysteries, such as inflammatory bowel disease.

Simulating the Universe in TensorFlow

March 6, 2020

In this blog post, Berkeley Lab's Uroŝ Seljak and his collaborators from the Berkeley Center for Cosmological Physics explain how to simulate a tiny universe in their FlowPM, a machine learning cosmological simulator that's both faster and smarter than conventional approaches.

Artificial Intelligence to Enable Smart Farming

February 28, 2020

Lawrence Berkeley National Lab scientists Ben Brown and Haruko Wainwright have been working with the University of Arkansas and Glennoe Farms on a pilot project called the “AR1K Smart Farm”. The project aims to improve sustainability in agriculture – increasing the yields while reducing chemical inputs and improving soil health - using sensors and artificial intelligence.

Stabilizing Synchrotron Light

November 19, 2019

Machine learning can predict noisy fluctuations in synchrotron light sources and correct them before they occur. The breakthrough solves a decades old issue and will make new, higher powered light sources, such as the Advanced Light Source located at Berkeley Lab, even more useful to scientists.

Etalumis Reverses Simulations to Reveal New Science

November 12, 2019

Deep neural networks are one key to helping scientists "reverse engineer" simulations, feeding experimental datasets back into models to reveal new science. Large Hadron Collider data provided the test case for this new framework, called Etalumis.

Deep Learning Expands Study of Nuclear Waste Remediation

November 9, 2019

A research collaboration including Berkeley Lab demonstrated how physics-informed generative adversarial networks (GANs) can be used in analyzing complex, large-scale science problems. The team modeled subsurface flows in the study of nuclear waste remediation, achieving exaplop/s-throughput in the process.

Sherry Li

Berkeley Lab Researchers Help Caltrans Predict Traffic Tie-Ups

November 4, 2019

Berkeley Lab researchers are using machine learning to help improve Caltrans’ real-time decision making with more accurate traffic modeling. The system is being tested in Los Angeles County's I-210 corridor testbed in conjunction with state and local agencies.

Predicting Watershed Behavior

July 30, 2019

Studying watersheds using learning techniques can lead to better predictions about downstream water supply and quality while offering better information for stewarding these vital resources.

Putting AI Tools in the Hands of Scientists

July 29, 2019

Some 175 researchers attended the DL4Sci School to acquire new tools and skills that advance deep learning studies

Learning to Look for Elusive Inoviruses

July 22, 2019

Using a machine learning approach, a team led by the Joint Genome Institute at Berkeley Lab has opened a new window on the number and diversity of elusive inoviruses. Trawling caches of metagenomic data, the researchers' machine learning method identified 10,000 probable inovirus sequences in six broad families.

A Clearer Understanding of Extreme Weather

July 24, 2019

Berkeley Lab scientists share their cutting edge machine learning and deep learning studies of climate and extreme weather.

Mining for Hidden Knowledge

July 3, 2019

Berkeley Lab study finds that text mining of scientific literature can lead to new discoveries.

To Pump or Not to Pump?

June 3, 2019

Berkeley Lab scientists aim to create a new machine learning tool for smarter groundwater management.

The Nature of Dark Matter

May 14, 2019

Using deep learning a Lab-led team aims to enhance the use of gravitational lensing to study dark matter.

The Future of Scientific Machine Learning

April 15, 2019

Early, mid-career scientists identify applied math challenges, sow seeds of future collaborations.

The Quest to Convert CO₂ into Fuels

April 8, 2019

Materials Project's machine learning approach opens new vistas in photocathode research.

Identifying Suicide Risks in Veterans

April 4, 2019

Berkeley Lab team uses deep learning to help VA address veterans' psychological, medical challenges.

Better Simulations on the Cheap

March 10, 2019

ExaLearn will interpolate details from sparse data to create surrogate models, first for cosmology.

Detecting Global Climate Patterns

February 25, 2019

ClimateNet will use deep learning to shed light on the changing behavior of climate and extreme weather.

Combing Cosmological Data for the Big Picture

January 25, 2019

Berkeley Lab researcher wins Large Synoptic Survey Telescope machine-learning competition.

Tracing the Nature of Neutrinos

December 21, 2018

IceCube research garners best paper award at IEEE machine learning conference.

Identifying Extreme Weather

November 20, 2018

Berkeley Lab-led team shares 2018 Gordon Bell Prize for deep learning climate application.

Optimizing Traffic Models

October 28, 2018

Berkeley Lab team using machine learning in smart and sustainable mobility solutions.

Supporting Sustainable Farming

October 28, 2018

Harnessing the power of machine learning and microbiology

Topology, Physics & Machine Learning Take on Climate Research Data Challenges

September 4, 2018

New data analytics tools could dramatically impact large-scale science data projects

Probing the Fabric of the Cosmos

September 5, 2018

Deep Learning and 3D simulations assist scientists in exploring the physics of the universe.

Learning from Living Cells

August 23, 2018

Infrared beams, machine learning show cell types in a different light

A Groundwater Early Warning System

August 13, 2018

Berkeley Lab researchers devise system to monitor contaminant plumes

Berkeley Lab-BIDS Fellows Share Machine Learning Expertise

August 30, 2018

Partnership Enriches ML4Sci Workshop and California Water Data Hackathon

Pinpointing Earthquake Impacts

June 28, 2018

Using ML techniques, new simulations can break down potential major quake impacts by building location and size.

The Challenges of Big Data and Advanced Computing Solutions

July 12, 2018

Berkeley Lab's Katherine Yelick contributes expert testimony to Congressional Committee hearing.

A ‘Google for Science’

June 19, 2018

Berkeley Lab team automates metadata discovery to search scientific images and data.

New ECP Co-Design Center to Focus on Exascale Machine Learning

July 20, 2018

Berkeley Lab One of Eight National Labs Participating in 'ExaLearn'

Physicists, Machine Learning Experts Team Up to Tackle TrackML

June 11, 2018

ML challenge aims to quickly reconstruct the paths of millions of electrically charged particles created in colliders.

Biofuels Bonanza

May 29, 2018

Predicting microbial biofuel production to speed up bioengineering

Extreme Weather

March 29, 2018

Deep learning at 15 pflops to identify extreme weather at scale

Metagenomic Clustering

March 12, 2018

Making sense of a genomic ‘data deluge’

Teaching Computers to Guide Science

March 6, 2018

Berkeley Lab, UC Berkeley's “Iterative Random Forests” deliver science insights

‘Minimalist Machine Learning’

February 21, 2018

Machine learning algorithm analyzes experimental images from limited training data, speeding up the deployment of learning tools

Targeting More Effective Cancer Drugs

August 16, 2017

Machine learning model predicts protein binding for better drug candidates

Machine Learning Enhances Predictive Modeling of 2D Materials

March 2, 2017

Using machine learning algorithms could reduce the time it takes to accurately predict the physical, chemical, and mechanical properties of nanomaterials from years to months.

Unlocking Mysteries of the Universe

January 30, 2018

Teaching machines to analyze simulations of exotic subatomic ‘soup’

JCAP Study Uses Neural Net to Predict Materials' Optical Properties