Our kaggle competiton paper is here, to facilitate automatic plant classification utilizing digitized plant specimens at @NYBGHerbarium and five other institutions. We are preparing for a new kaggle competition for 2022, harnessing images from 50 institutions around the world.
New Research: The Herbarium 2021 Half–Earth Challenge Dataset and Machine Learning Competition: Herbarium sheets present a unique view of the world's botanical history, evolution, and biodiversity. This makes them an all–important… https://t.co/mYQmvWh0qu #PlantScience#PlantSci
Visualizing the attention layers shows what the machine focuses on when discerning a class (bronzed cowbird) from others.
Method: "TS-CAM: Token Semantic Coupled Attention Map for Weakly SupervisedObject Localization" (https://t.co/XPtY5WhnXx)
Data: https://t.co/hDcVawtbqe
Highlight of this week: From a class project for Fundamentals of Machine Learning class. Letters 'a' and 'b' segregate nicely on the PC axes. Plus note the PC1 eigenvector! It clearly shows that the shape differences between a and b are the major source of variation.
Interested in pursuing a PhD in ecosystem science, or know someone who is? Come work with us https://t.co/0kRt42eXC4 We're merging ecosystem theory, big data and Earth system models, and we need your help!
App. deadline @ESPM_Berkeley: Dec 1st. Email for more info. Please RT :)
Being a mass media fellow this summer was an absolutely amazing and transformative experience. If you're interested in making a career out of science writing, this is the perfect place to start!
Current status: 25 reviewer invites sent 😩 & 1 acceptance!
Reviewing papers is unpaid/undervalued & part of a broken publishing system, but editors will need to find someone to review YOUR paper...do us a favor and review someone else's paper...
If you stay in academia (or most careers) there will always be the guilt of “I could have done better.”
I channel this into drive to do even better next time, and I let go of the guilt by recognizing that putting more time into the last goal would take away from future goals.
At the heart of good time management is the realization that every "yes" is a "no" to something else. Before you commit time to do something, ask yourself what you will have to not do as a result. Then ask yourself whether that is a trade worth making.
Please RT: @WSL_research is seeking a #postdoc for the project “Scaling of Diversity in Swiss Forests.” The position is open until filled, and @ian_mcfadden_ can vouch that Zürich is a wonderful area to live and work in: https://t.co/CMCcWR2n5N
What happens after censusing all stems >1cm DBH on large-scale plots? Data cleaning and meaning making! ForestGEO is offering 2 paid internships based in its DC HQ to contribute to and learn more about data management. Applications DUE August 30. https://t.co/OIU9k5yx2B
Indiana will see a dramatic increase in dangerously hot days over the next few decades, even if the Paris Agreement's goals are met. If these goals aren't met, summer conditions will become progressively more dangerous. Are we protecting the vulnerable? https://t.co/KQLaU5KRWK
https://t.co/cmuUj4bhyU.
Smith et al., 2019.
Ground-based #LiDAR surveys conducted monthly to quantify seasonal variation of LAI vertical profile, for 4 years in #Amazon_#TropicalForest.
https://t.co/8kRjH35mIA
Vertical leaf structure in tropical forests.
-Temp. and precip. explains 50% of #ANPP variation among lowland tropical rain forests.
-Vertical leaf area profile explains another 38% of the variation.
Research by @kc_cushman and @kellner_lab