Dear friends,
The long wait is almost over. On Valentine's Day, we will launch a citizen science collaboration to share the excitement of our search for life in the universe with the general public. We are thrilled to engage the power of citizen science to help us classify thousands of radio signals. Participants will actively contribute to the search with a desktop, laptop, or smartphone. UCLA graduate student Megan Li, who will lead the data analysis, said: "Humanity’s most profound discovery could be a few clicks away.” To kick off the effort, we will host a launch event on Zoom on February 16, 2023 at 6:00 pm PST (register here). The citizen science collaboration is built on the Zooniverse platform with funding from The Planetary Society and the NASA Citizen Science Seed Funding Program. Launch details and updates will be posted on our social media channels (Facebook, Instagram, and Twitter). Please spread the word!
|
|
|
|
This Valentine's Day, fall in love with science. Join a community of citizen scientists who help UCLA astronomers search for life in the universe.
|
|
|
|
|
|
Earlier this week, Peter Xiangyuan Ma and collaborators published the results of their deep-learning search for technosignatures in Nature Astronomy. I am happy to see increasingly sophisticated uses of machine learning (ML) in the SETI context, and the newly published work is definitely an impressive feat. Their ML pipeline handles both the detection and filtering of signals, whereas previous SETI ML applications, including our own (see October 2021 newsletter), typically focus on a single data processing task. However, entrusting signal detection to ML algorithms may have limitations. A general concern about ML algorithms is that not all signals may be analyzed. Ma used a 90% confidence threshold to identify signals of interest among all detected signals. But what if a genuine technosignature registered at 89% on that scale? Humanity would miss the most important phone call in history. For this reason, I anticipate that SETI searches will continue to use a combination of classic and ML approaches, at least for now. The advantage of classic algorithms is that they generally evaluate all detected signals. Once all the detected signals are identified, ML algorithms can improve and accelerate the identification of radio frequency interference.
Ma's ML algorithm identified eight signals not previously detected by the Breakthrough Listen (BL) pipeline. We are eager to run the data through the UCLA SETI Group pipeline to see if those eight signals are detected. Our pipeline uses different algorithms and detects many more signals with the same telescope, receiver, and detection threshold (see November 2020 newsletter or our recent paper). It will take a few modifications because we normally use the full time resolution of the data, whereas the BL collaboration uses lower time resolution data products (Lebofsky et al., 2019).
The meeting of the American Astronomical Society (AAS) in Seattle earlier this year had a session titled "Technosignatures" at which Megan and I spoke, and several other technosignature talks and posters. It is wonderful to see that this field is gaining traction at the AAS, which may ultimately lead to improved funding for SETI (see March 2022 newsletter). On that note, NASA has posted a job opening for a Senior Scientist for Astrobiology, and this scientist may have influence over future SETI funding by NASA. Hopefully one of the job qualifications is open-mindedness about the search for technosignatures!
I look forward to engaging with many of you on Zooniverse soon.
Warm regards,
Jean-Luc Margot
|
|
|
|
|
