You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
I hated “idea papers”. High level ideas without experimental evidence were not real science to me. Then I realized that many empirical results are post-rationalized randomness and overffiting. They are not evidence at all. Now I respect idea papers. They are real science.
Someone is putting an idea out there to be tested in a scientific way. They are willing to be proven wrong. They are taking a risk. Someone fitting their theory to their empirical result after the fact takes no risk. Nobody can prove the theory did not come first.
The former one tries to find causality, which is what we are after. The latter one proves nothing, just correlation.
Argyris Ladder of Inferencepic.twitter.com/n975bgucoz
As well as over fitting, people also disregard data as they apply their belief reenforced bias, "I select the data to observe"
This is so true for RL, the signal to noise ratio gets worse due to scoreboard driven research, too many hyper-parameters, and high compute power.
What do you mean by "post-rationalized randomness and overfitting"? If an experiment serves the sole purpose of proving something it has serve their purpose and it's not overfitting right?
Assume the "idea" that 6-sided dice always give a 6 up when thrown. You choose dice until you find biased ones; throw them until you record a streak of 10x consecutive 6s; then present these sole results to support ("prove") the initial claim. A lot of ML papers work like this.
The place where I come from that is called fraud. had no idea.
My example was obvious, reality less so. I notice sometimes authors themselves don't know that. It's a very heterogenous mix of engineers, data scientists, flavors of PhDs that write papers. Causes are many: lack of training in scientific method or statistics, hidden complexity…
Maybe the ocasional promotion lol
"idea papers" cloud give you a very good questions . Or a very good definition of the problem . That's why many data science and Big data projects fails . The data no talk you need a hypothesis or idea to test .
As long as the idea is falsifiable. So many idea papers aren't.
I think you start writing posts back. Better for you and others.
Idea papers with experimental evidence are the real paper.
Unfortunate conferences don't accept idea papers or that there arent any conferences for that genre. Are there particular circles you find idea papers written/post more?
Eventually every scientist gets to this place if they remain a scientist.
Could you please share one idea paper? I just want to understand the difference...
I agree as long as they don't turn into flag planting.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
