Nicola Romanò @nicolaromano@qoto.org

@koen_hufkens Not my field so can't suggest a specific journal, but I'd definitely start by putting it on BiorXiv

Following up from last week's post, here is the continuation of our journey onto neural networks! In this post I discuss the basics of neural network training. 🤖

https://www.nicolaromano.net/data-thoughts/training-neural-networks/

Keep watching this space as more posts are in the pipeline! Next up: make your first neural network in Python!

Are you interested in the topic? Let me know what you would like to hear next

#neuralnetworks #ai #artificialintelligence #machinelearning #training #statistics #python

@notoriousiptg Interesting and maybe not surprising (?).

@requiem It seems to me like you are looking at it from the wrong point of view. I am sure you have found many other human-introduced bugs as well, and have also found good code written with the aid of Copilot. The problem is not Copilot (which works well in many cases), it is people misusing it and trusting the code without actually checking (or not being able to check) that it works.

@ecological_fallacy @albertcardona @eLife Honestly, I don't care too much for the short assessment... however I think we also have to be realistic and realise that grant and job applications panels won't stop looking at publications any time soon. And they are more likely to read 100 paragraphs than 100 papers.

I have been considering publishing my next article in #elife
I was not very convinced by their new #publishing method, but the more I think about it, the more I like it.

What convinced me is that I think of the way I review papers myself. I won't ever reject a paper unless there is something majorly wrong e.g. from an ethical point of view. Instead, I would rather spend time and give constructive and realistic feedback to improve the study.

This is because of two reasons:

1. If the study idea/methodology etc, is good but maybe is missing some key experiment, I think that the authors must have put a lot of effort, time and money into producing this. I have been through the "your work is not fancy enough for our prestigious journal" crap enough times that I will not engage in that. Ever. There is no reason your paper should not publish negative results if the study is well done.
Also, people's jobs and mental health depend on that, which is way more important.
Also, there are plenty of papers in "fancy journals" that are just piles of bs, so I really won't buy into shiny names (I have just spent an entire day trying to run code from several papers published in high-IF journals to no avail...).

2. If the study is poor, it is easy to say: "This is cr*p, straight reject". This just means the authors will submit elsewhere, hoping the next reviewer won't be bothered reading the paper in depth and will let it through. Even worse, this plays into the hands of #predatory journals. I would rather say this can be accepted after all of these major revisions.
The authors get useful feedback on how to improve their study; they might choose not to act on it, but at least I have made my part.

I would be interested in hearing other views on this.

#publishing #peerreview #science