@pkoo562@zrcjessica From the retrospective, was it wise to base the experiment around deep ensembles initially? It makes intuitive sense, I just wonder if ensemble parsimony is showing its age. Perhaps student models and ensemble models converge in performance but for entirely different reasons.
@YudiXie235 Excited to learn about your journey and progress. I am interested in the origins of Bayesian modeling and its divergence from frequentist approaches, where the dichotomy makes sense. The latter part of your post seems too behavioral to be impartial.
@selcukorkmaz In addition to differences to be aware of when reporting either SE or SD, work that uses one measure or the other can suggest important details about model assumptions, experiment conditions, etc. For example, if SE is used but not SD, it could be there are Bayesian alternatives.
A helpful perspective on the societal value of empirical pursuit. STEM disciplines and public funded research needs more support than ever, especially going in to this next administration.
The $3B Human Genome Project was started under a Republican administration and later acknowledged by the Obama admin to have returned $141 to the economy for every $1 spent. Science has a higher ROI than any other category of government spending.
@OdedRechavi Publish negative results however possible, whether a lab protocol, preprint, or manuscript. Further, why was it negative, could it be bias or latent factors, and does it warrant a new letter, method, or review? Do others in your field agree or have similar unreported findings?
@MishaTeplitskiy This is a norm that needs to change. Replication of science is science, and further presents its own set of challenges over original or entirely novel work. Groundbreaking insight has its place, but we should not neglect teaching/training/collaboration opportunities so much.
Well, OK, I guess it all depends on your worldview; you can see your place in the world as something to defend from those who might want to take some or all of it from youβor you can see it as a grand adventure, a place to make dreams come to fruition for yourself & others.
(1/3)
Job search insights: #employers should recommend and re-evaluate good #applicants for alt open positions, not simply serve decline emails. #Recruitment cost can be drastic, rerouting benefits applicants & employers, and it would improve org perception. https://t.co/at8AAYpzl2