Skip to Main Content U.S. Department of Energy
Deep Science Agile Initiative

News and Highlights

2019

Deep Learning 2019

Deep learning for power systems. Qiuhua Huang has been invited to attend the INFORMS 2019 Annual Meeting where he will be presenting "Power System Control and Decision-making Via Deep Reinforcement Learning." The conference will take place in Seattle, WA, October 20-23, 2019. Over 12,000 members strong, INFORMS is the international association for professionals working in operations research and analysis.
September 2019

Grid

AI Ups Response Time when the Grid Goes Down. To help grid operators improve response time and contain problems before they escalate, a team of researchers at PNNL is using a type of artificial intelligence called deep reinforcement learning. Read more.
August 2019

Cryptocurrency

How Cryptocurrency Discussions Spread. Every day, thousands of messages on Reddit [and elsewhere] contain discussions of cryptocurrencies. Some lead to increased interest in a cryptocurrency -- not just a discussion of that cryptocurrency, but its actual price/value. Clearly, not all cryptocurrencies are equivalent -- and analyzing a set of them can reveal how bad actors might exploit those differences. Read more.
June 2019

Supercomputing

PNNL Researchers to Present at ICS-2019. Multiple PNNL researchers have had papers accepted to the upcoming Association for Computing Machinery International Conference on Supercomputing to be held in late June in Phoenix, AZ. ICS-2019 is the premier international forum for the presentation of research results in HPC systems. Only 45 of the 193 submitted papers were accepted. Read more.
May 2019

ChemNet

ChemNet Bridges the "Slim-Data Gap". Scientists have developed a deep neural network that sidesteps a problem that has bedeviled efforts to apply artificial intelligence to tackle complex chemistry a shortage of precisely labeled chemical data. Read more.
March 2019

Convolutional Neural Networks

Applying Convolutional Neural Networks (CNNs) to Particle Physics. PNNL researchers recently presented what is thought to be the first use of applying deep CNNs to neutrino physics with modern compute packages on the National Laboratory complex's Leadership Class supercomputer known as Summit. Read more.
March 2019

Flu Research

#flu Research Highlighted in Popular Science. Research by Svitlana Volkova, Ellyn Ayton, Katherine Porterfield, and Court Corley was recently highlighted in Popular Science magazine. The team applied deep learning techniques to analyze more than 170 million tweets over three years. They found that their model could accurately forecast flu-like illnesses at a local level. Read more.
March 2019

Breast Cancer Detection

Deep Learning Targets Breast Cancer Detection. An effort sponsored by the Deep Science for Scientific Discovery agile investment is bringing together PNNL and Fred Hutchinson Cancer Research Center to study data derived from cancer research. Read more.
March 2019

Reuters

Reuters Video Highlights PNNL's Deep Learning. Thanks to a new video by Reuters News Agency, the world is now getting a rare glimpse inside PNNL's underground laboratory—and its role in advancing artificial intelligence. Read more.
March 2019

Richland Scientists

Richland Scientists are Teaching Computers to learn—and Advancing Scientific Frontiers. A news column by Laboratory Director Steven Ashby highlights PNNL's research in machine learning. "Richland scientists are teaching computers to learn&mbsp;and advancing scientific frontiers," was published on January 28, 2019 in the Tri-City Herald. Read the full text here..
March 2019

Automating Expertise

Basic Research Needs for Scientific Machine Learning: Core Technologies for Artificial Intelligence. DOE's Advance Scientific Computing Research (ASCR) brochure on Basic Research Needs for Scientific Machine Learning: Core Technologies for Artificial Intelligence —informed, in part, by PNNL—are now available. Read more.
March 2019

Physics Collaboratoin with PhiLMs

Applying deep learning to physics in new collaboration. PNNL has been tapped to lead a $10 million, four-year effort to uncover hidden physics using new developments in deep learning, a computing technique that harnesses the power of machine learning and big data. Read more.
February 2019

Automating Expertise

ML for Cloud Identification. "Lidar Cloud Detection with Fully Convolutional Networks" will be presented at the upcoming IEEE Winter Conference on Applications of Computer Vision (WACV 2019). Authored by Erol Cromwell and Donna Flynn, the paper focuses on a novel approach for segmenting laser radar (lidar) imagery into geometric time- height cloud locations with a fully convolutional network. To learn more, read the recent press release or view the video.
January 2019

2018

Fish Detection at CSCI 18

Fish Detection at CSCI 18. "Underwater Fish Detection using Deep Learning for Water Power Applications," was presented at the 5th Annual Conference on Computational Science & Computational Intelligence (CSCI 18). The paper describes the research team's testing of YOLO (you only look once) on detecting fish in challenging underwater environments near tidal turbines and other water power devices. The full paper is available on Research Gate. The research also yielded some interesting fish videos, which can be viewed on GitHub.
December 2018

Two Posters Presented at Black in AI

Two Posters Presented at Black in AI. PNNL researchers presented two posters at the second annual Black in AI event, recently held in Montreal, Canada. Eric Corbett and Meg Pirrung presented "Interactive Machine Learning Heuristics," which intruded ten heuristics for machine learning. Omar Clinton and Nathan Hodas presented "Transfer Reinforcement Learning Through Demonstration," which covered the team's research assessing the effects of supervised and unsupervised pre-training algorithms on Sonic the Hedgehog's performance on unseen levels in response to OpenAI's recent challenge.
December 2018

Invited Speaker at NeurIPS

Invited Speaker at NeurIPS. Garrett Goh presented, "Multiple-objective Reinforcement Learning for Inverse Design and Identification" at the Machine Learning for Molecules and Materials workshop at, NeruIPS, the 32nd conference on Neural Information Processing Systems. The paper, coauthored by Haoran Wei (Univ. of Delaware) and Mariefel Olarte, focused on the team's research developing heuristics for curriculum-learning based multiple-objective reinforcement learning, and applying it to the context of chemical identification.
December 2018

Posters at WiML

Posters at WiML. PNNL researchers presented five posters at the Women in Machine Learning (WiML) Workshop on December 3, collocated with Neural Information Processing Systems meeting, NeruIPS.

  • "SHARKZOR: Human in the Loop ML" - Meg Pirrung, Nathan Hodas, Courtney Corley, Lawrence Phillips, and Nancy O'Brien
  • "Generalizability of Temporal Convolutional Neural Networks for Inter-subject EEG Classification" - Leslie Blaha, Gerges Dib, Kayla Duskin, Johnathan Cree, and Jonathan Suter
  • "Neural Networks Enhanced with Social Signals to Improve Forecasting of Cryptocurrency Price" - Maria Glenski, Emily Saldanha, Graham Mueller (Leidos), Ning Yu (Liedos), and Svitlana Volkova
  • "Encoder-Decoder Architectures for Modeling News Content with Different Levels of Credibility" - Aparna Garimella and Svitlana Volkova
  • "Node-Aware Attention Model to Forecast User Interactions in Dynamic Social Graphs" - Prasha Shrestha, Suraj Maharjan, Dustin Arendt, and Svitlana Volkova
December 2018

NVIDIA Guest Speakers at SC18

NVIDIA Guest Speakers at SC18. Court Corley and Nathan Hodas were invited by NVIDIA to speak at SC18 as part of their hosted speaker series.

  • Dr. Hodas' presentation, Leveraging Big Data to do Deep Learning on Small Data, can be viewed here.
  • Dr. Corley's presentation, Advancing Scientific Frontiers through Deep Learning, can be viewed here
December 2018

Supercomputing 2018

Supercomputing 2018. Ang Li's poster, "Binarized ImageNet Inference in 29us," was recently on display at SC18. Posters selected to Supercomputing yearly are subject to rigorous reviews, typically involving careful management of conflicts and several reviews. The SC18 posters program received 165 submissions to the Research Posters and accepted 93 submissions.
November 2018

Automating Expertise

Automating Expertise. ChemNet employs a new training approach that learns expert knowledge from large, initially unlabeled databases and outperforms current state-of-the-art supervised learning methods. Watch the video. Learn more.
September 2018

Neural Network Research Presented at Global Conference

Neural Network Research Presented at Global Conference. At the recent GiMLi 2018: Geometry in Machine Learning, Craig Bakker presented a poster on training methods for neural networks. GiMLi 2018 is co-located with the 35th International Conference on Machine Learning in Stockholm, Sweden, a premiere gathering for researchers in machine learning.
August 2018

Staff Present at Prestigious Artificial Intelligence Conference

Staff Present at Prestigious Artificial Intelligence Conference. Lawrence Phillips, Garrett Goh, and Nathan Hodas presented "Explanatory Masks for Neural Network Interpretability" recently at the International Joint Conference on Artificial Intelligence and the European Conference on Artificial Intelligence. A premiere gathering of researchers in artificial intelligence, IJCAI-ECAI-18 was held July 13 - 19 in Stockholm, Sweden.
August 2018

Chemception Poised to Change How Chemists See Molecules

Chemception Poised to Change How Chemists See Molecules. In a recent conversation with Chemistry World, Nathan Baker discussed Chemception, a deep convolutional neural network framework developed to predict a molecule's chemical properties from its structure. Read more.
August 2018

Four PNNL researchers participate in MATDAT18

Four PNNL researchers participate in MATDAT18, the Material and Data Science Hackathon. The three-day event is supported by the National Science Foundation. Bharat Medasani's research problem was accepted, one of only 21. Read more.
July 2018

Deep Learning Could Help Detect Nuclear Events Worldwide

Deep Learning Could Help Detect Nuclear Events Worldwide. PNNL scientists Emily Mace and Jesse Ward teamed to explore the promise of deep learning to help interpret signals from radioactive decay events, which could indicate underground nuclear testing. Mace presented their work at the 11th MARC conference — Methods and Applications of Radioanalytical Chemistry — in April 2018 in Hawaii. Read more.
July 2018

XXX

Curbing misinformation propagation on social media. Svitlana Volkova and Dustin Arendt, with collaborators, presented "Can You Verifi This? Studying Uncertainty and Decision-Making about Misinformation Using Visual Analytics" at ICSWM. Read more.
July 2018

Working to inform data choices

Working to inform data choices. Nathan Hodas participated in the National Library of Medicine of the National Institutes of Health (NIH) Data Science Drivers Workshop. Read the research proceedings.
July 2018

Training neural networks to classify low-background data

Training neural networks to classify low-background data. Emily Mace presented "Use of Neural Networks to Analyze Pulse Shape Data in Low-Background Detectors" at the Methods & Applications of Radioanalytical Chemistry (MARC) conference. Read more.
July 2018

Fuzzing research cited

Fuzzing research cited. As part of the Artificial Intelligence and Global Security Initiative, the Center for New America Security has published a series of reports related to the implications of the artificial intelligence revolution as it relates to global security. Read more.
July 2018

Teaching AI to Identify Clouds

Teaching AI to Identify Clouds. Co-PIs Donna Flynn and Erol Cromwell's (PCSD) work on developing an artificial intelligence (AI) system was recently featured in the Wall Street Journal, which reaches over 2.2 million readers. The system is being designed to distinguish clouds from other atmospheric constituents in lidar imagery. Read more.
May 2018

Svitlana Volkova

#Flu. Research highlighted in Scientific American. The April 2018 Scientific American, features work by NSD's Svitlana Volkova (Computing and Analytics Division) and team on predicting influenza outbreaks using social media. Read more.
April 2018

Aaron Tuor

Best Paper Finalist. "Protein Mutation Stability Ternary Classification using Neural Networks and Rigidity Analysis," was recognized as a Best Paper Award Finalist at BICOB 2018. The paper was authored by NSD's Aaron Tuor, with collaborators from Western Washington University. Read more.
March 2018

National Academies Roundtable

National Academies Roundtable. Nathan Hodas was an invited panelist at the National Academies of Sciences meeting: Artificial Intelligence and Machine Learning. Read more.
February 2018

2017

Neural Information Processing Systems

NIPS. Researchers from the deep-learning group at PNNL hosted three presentations and a demo at NIPS, the annual Neural Information Processing Systems meeting on December 3-8, 2017. Read more.
December 2017

Women in Machine Learning workshop

WiML. Deep science researchers presented four posters at WiML, the annual Women in Machine Learning workshop held December 4-7, 2017. Read more.
December 2017

Supercomputing 2017

Supercomputing 2017. Court Corley was an invited speaker at the NVIDIA's SC17 booth, November 14, 2017. Read more.
November 2017

PNW Partnerships

PNW Partnership for Data Intensive Biomedical Science. The Deep Learning for Scientific Discovery Agile Investment team hosted the inaugural meeting of the PNW Partnership for Data Intensive Biomedical Science September 27-28, 2017. Read more.
September 2017

Second Annual Faculty Summit

Second Annual Faculty Summit. More than 40 researchers and educators gathered at PNNL June 14-15, 2017 for the second annual Computing@PNNL Faculty Summit, Machine Learning and Human Computer Interaction for Science and Security. Read more.
June 2017

Deep Learning