Project investigator, Dr. Hassan Najafi, received a grant from Cisco Research in the amount of $112,651 to carry out research on "Time-Based Stochastic Processing for Edge AI" during 9/1/2021 – 8/31/2022, as the PI, with a Co-PI from UC, Irvine. The research activities are closely related to the PREFER project task led by Dr. Najafi, called Stochastic Computing.
Louisiana EPSCoR, Board of Regents featured the PREFER project in its recent Newsletter article, entitled "Precise Weather Forecasting via Intelligent and Rapid Harnessing of Big Data." The article related the project themes and their expected outcomes to benefit the economy and human welfare of the nation in general, and of participating jurisdictions in particular, as can be found at https://web.laregents.org/wp-content/uploads/2021/07/vol13no02.pdf.
Dr. Nian-Feng Tzeng delivered an invited presentation on the PREFER project and its solutions for precise regional flood forecasting at IEEE 7th World Forum on Internet of Things (WF-IoT 2021), held in New Orleans. The presentation was made in a hybrid form during Session SPES10.1 on Data Acquisition for Flood Forecasting, with most attenees being remote (see the photo below).
Dr. Xu Yuan was invited to present a talk on "Precise Hydrological Predictions via Neural Networks" at IEEE 7th World Forum on Internet of Things (WF-IoT 2021), held in New Orleans. His presentation was among the three in Session SPES 10.2 on Data Analysis and Forecasting, with active discussions at the end of the session, as shown by the photo below.
Summer AI/ML tutorial session for project RUE participants held in the hybrid manner, with 6 engaged face-to-face on the UL campus and 10 attending remotely (see the photos below). The tutorial is lectured by Xu Yuan, project co-PI, twice per week during June and July, assisted by Purushottam Sigdel and Yihe Zhang.
Newest micronet station deployed by two PREFER undergraduate students, Jaye Hockett and Ryan Thomas (shown in the photo), in Western Kentucky University (WKU) 800-acre Agricultural Farm to examine many facets of surface thermal inversions for inversion mitigation and also to gather meteorological readings as inputs to PRERER weather prediction modelets under development. It is among 5 inversion-equipped stations (3 mesonet and 2 micronet stations), which belong to the Kentucky state-wide Mesonet.
Research funding of $200K from the Kentucky Transportation Cabinet awarded to PREFER researcher, Eric Rappin, as a co-investigator, working on the project of "Impact of Real-Time Weather Information" during 7/1/2021 and 12/31/2022. Investigating the correlation of real-time weather variables and automobile crash locations & severity, the project aims to produce recommendations for real-time message signs to alert potential weather related safety risks.
Our first annual PREFER Advisory and Evaluation Board (AEB) meeting was held on May 14, with all four AEB members participating. This 3.5-hour meeting permited key PREFER team participants to (a) highlight the whole project scope and involved research tasks and summarize overall team accomplishmentes, and (b) let task leads outline major research activities and outcomes over the past 8.5 months plus planned work topics and technical solutions for the coming year. In the last meeting session, the AEB members offered their candid and valuable comments and suggestions on project activities and future research topics, pointing out needed team effort on further collaboration among tasks and on more ties of technical solutions to the project goal of better weather forecasting regionally. Possible outreach to local high school students for their involvement in project activities could be attempted as well. A total of 26 team members participated in the meeting, besides the four AEB members. Seven task leads made their presentations on Mesonet Surface Observations (Task 1), In-Memory Processing (Task 3), Error Detection and Recovery (Task 4), Stochastic Computing (Task 5), GPGPU Support (Task 6), enhanced weather data gathering for Meteo Modelets (Task 7), and Focused Meteorological Problems (Task 8).
The recorded video of this meeting can be found here.
PREFER team meeting took place for the PI/PD to review the project achievement summary and for project task leads to present their respective research activities and progress, in preparation for the upcoming advisory and evaluation board meeting on May 14. A total of 25 team members participated in the meeting, which featured the following task lead presentations: Xu Yuan on Mesonet Surface Observations (Task 1), Nianfeng Tzeng on In-Memory Processing (Task 3), Lu Peng on Error Detection and Recovery (Task 4), Hassan Najafi on Stochastic Computing (Task 5), Li Chen on GPGPU Support (Task 6), Stuart Foster on Meteo Modelets (Task 7), Sytske Kimball on Focused Meteorological Problems (Task 8), Xu Yuan and Nianfeng Tzeng on Undergraduate Research Experience (Task 10).
PREFER senior investigator, Dr. Mathieu Kourouma, won an equipment enhancement grant funded by Louisiana BoR on "Stimulating Teaching, Learning, and Instructional Alignment in Computer Science Department at SUBR" (with Award Number: 082ENH-21) for $199,864 during July 2021 - June 2022. PREFER investigator, Dr. Lynette Jackson, co-led this funded project.
PREFER senior investigator, Dr. Lu Peng, is awarded by Louisiana BoR with an equipment enhancement grant on "Accelerated Machine Learning, Big Data, and Graphics Education and Research Equipment" (with Award Number: 041ENH-21) for $119,900 during July 2021 - June 2022. He is one of five awardees for the grant.
A research grant is awarded by Louisiana BoR under its Proof of Concept/Prototype Category to PREFER senior investigator, Dr. Li Chen, for her work on "Fast and Efficient Scheduler for Deep Learning in GPU Cluster" with the funding amount of $39,990 during June 2021 - May 2022.
Twenty-two team participants joined this Zoom meeting, including 11 graduate and undergraduate members, who were introduced to the team for the first time. Besides a brief summary on team research activities made by the PI, the meeting included two technical presentations for information sharing, on (1) weather modelets and their initial results and (2) GPU support for bigdata analytics, delivered respectively by Ph.D. student participants: Yihe Zhang and Pisacha Srinuan. Discussions followed after technical presentations.
PREFER senior investigator, Dr. Lu Peng, receives NIH funding for the project of "Detection and Automatic Privacy-Protected Contact Tracing System Designed for COVID-19" (with Project Number: 1U01AA029348-01), serving as the project leader to work jointly with four other co-PIs. The project addresses a new paradigm for rapid and direct pathogen detection and identification, with automatic data transition and event tracing in a blockchain-encrypted manner. It starts on December 21, 2020 and ends on November 30, 2022, for a total funding level of $888,643.
All team participants attend the 2-day virtual meeting hosted by NSF's EPSCoR PDs in part or in full. They met cognizant NSF directors, learnt project reporting and evaluation expectations, and were informed budgetary issues. Short presentations by all funded T2 projects permitted our team members to know the scopes of all other projects for possible future collaboration. Programmatic experience shared by PIs from earlier years' funded T2 projects offered valuable guidance for our project going forward. Details about team management, governance, coordination, and possible conflict/problem resolution were also clarified during the 2nd day's breakout session. The team initiated the development of a strategic and milestone plan as well in the 2nd meeting day.
Most team participants joined this virtual meeting to sort out details about workforce development regarding undergraduate engagement and summer research support. The meeting also updated team members with data output gathering service for all project activities and inform about the NSF's project kick-off meeting to take place in the following week.
A presentation, entitled "The Future of Weather Forecasting," was given by Eric Rappin at the Commonwealth Computing Summit at the University of Kentucky on 13 Oct. 2020, briefly outlining how PREFER is aimed at advancing next-generation forecasts through a set of simple machine learning models (dubbed 1-D modelets). It discussed the initial PREFER approach on training ML models to correlate meteorological quantities based on years of Kentucky Mesonet observations (e.g., how does wind direction impact humidity). Further investigation would be on developing 2-D modelets to account for spatial correlations.
All 14 team participants joined the meeting via Zoom for 1.5 hours to review technical tasks under the funded project, to affirm their respective task involvement, and finalize the team execution strategy that contains the activity level, the task level, the whole team level. Each task has one lead, who is responsible for assigning and coordinating research activities included in the task. A task-level meeting is expected to hold monthly or as needed, whereas a team-level meeting is expected three times annually.
University of Louisiana at Lafayette, 104 E University Ave, Lafayette, LA 70503-2014 (337)482-5811