@JarekDuda are you familiar with any ML approaches?

@vzn, I work with ML group for a few years

Hey I am new to ML please give me some advice on where to start

@vzn yes, I have long finished my MSc in GR research but can't find a position to continue the research. Now I see there are academia-industry cooperative PhD training in scattering amplitudes in Berlin which involves learning advanced coding skill in some software company. I don't know if I should apply for it since I have too little coding experience and have no idea what advanced coding is like and whether I would like it.

I have taken the course advanced math method for physics and really like it.

But as for coding, I probably even don't know something basic.

@JarekDuda dont know what you mean by this. diverse ML approaches apply to basically almost any types of data. the only question is how good the fit is. the auto-sklearn pkg is an example that does "hyperparameter optimization" ie can try multple different ML algorithms and find best fits by tuning hyperparameters, so think it could yield something for your problem.

@CaptainBohemian sorry to hear you werent able to go in the direction you preferred (sci research) but it is not an uncommon story these days, have read 2 books recently pointing to it (it spans decades, even nearly ½ century). there is a lot of physicist → coding trend, even have seen it repeatedly in SE Physics chat room denizens as maybe you have, it can be done, think there are worse ways to make money :|

@AdityaBharadwaj hi, how about some intro/ bkg? are you a student? what country? what kind of data do you want to work with? etc

@vzn I am a highschool student(9th grade).I started learning some basic classification algorithms and I know python pretty well.I want to work with neural networks and image recognition.My goal is to come to a position where I can build a self driving car.But I do not know where (for free)I can learn about neural nets selfdriving cars and all that.I tried to learn something on youtube but it is not beginner friendly

@vzn but I wonder what kind of coding to be learnt in the software company during the PhD training. Would there be a company which employs people to do research on scattering amplitudes by coding? I prefer to learn coding as part of doing physics than learning coding for other fields I may not be interested in.

@CaptainBohemian there are a lot of coding-related data science jobs, maybe that would be closer to your tastes. have you seen enumaris talk about his work in Physics chat? also machine learning uses a lot of data science. re your example there may not be a ton of physics-related coding jobs out there, but some do exist... its like a "laboratory engineer" type position...

@AdityaBharadwaj there are lots of good books. recently adapted this code, you could learn a lot from it scikit-learn.org/stable/auto_examples/…

@vzn actually I feel data science sounds boring--it sounds like sorting out experimental data. I have long not done that kind of thing. Actually I probably have never done that kind of thing by computer--in my undergraduate experiment courses, I used paper and pen to sort out experimental data rather than computer. When I taught physics experiments, it's my students who sorted out experimental data by computer; I have never done that.

@vzn I usually ignore talk about computing in h bar because I have no idea of that kind of talking.

@vzn This is the curriculum of the PhD training I talked about. Some math techniques therein look interesting but I have no idea of the computing part therein. I guess the part of industrial training wouldn't contain the math part but focuses on the computing. I wonder why those industrial partners would want to cooperate with academia for the PhD training on scattering amplitude.

Are those industrial partners also engaged in the research of scattering amplitudes? Or they just provide the computing-related training which is required for the research of scattering amplitudes?

@CaptainBohemian hey! broaden your horizons! youll have to start thinking more like a geek. theres a lot you could enjoy in CS if you like physics + science. simply think of computation as something like "animated mathematics"! have you ever looked into how much CS is used on the LHC? a significant part of the $15B total was spent on computational hardware/ analysis etc. a not-well-publicized aspect! etc! what country are you in anyway?