Hacker News new | ask | show | jobs
Suggestions about Starting a Phd in Machine Learning – Yay or Hell No?
2 points by 0x8416 3314 days ago
2 comments

I've done a PhD. You will start all bright and eager, determined to write the best thesis in history, and excited to learn everything about the field and advance the state of knowledge.

You will finish by being relieved to get shot of the bloody thing, desperate to move on and work on something interesting for a change.

If you love the field enough to cope with that, and still to want to dive in and work on the same sort of thing for 3, 5, or even 8 years, then go for it. Otherwise think very hard.

Why would you do this?

I completed my Msc in Data Communications and Computer Networks (worked on SDN for my thesis) in 2015 and i have been working ever since. Considering going back to school for Phd to continue my work on SDN but reading and seeing all the fuss about ML i am having a rethink abt the SDN and what the future might hold
Changing topic area for a PhD is hard work, but usually of significant value, because you have insights into each from the other. As such, the work provides a return. But it's hard work.

Will the PhD be of net positive value? In some fields, to have a PhD is of net negative value (although perhaps only initially) because potential employers will believe you have only book-learning experience, and are of no practical use.

Do you want to do a PhD? Why? It's hard work, extremely frustrating, and usually somewhat marginal. If you work on truly mainstream material there is always a chance someone will publish before you and potentially prevent you from getting the PhD.

So if you really want to do it, then go for it, but start by understanding your reasons and motivations. Certainly ML is a rich field, full of as-yet-undiscovered techniques and algorithms, and undoubtedly useful. But remember, a PhD is not the only way to get into a field - another path might be better. Don't be excessively tempted for the initially easy option.

Thanks so much for taking out your time to reply. I agree with you a 100% in all you have pointed out.

I had plan to either focus on Cisco Certifications (got a few certs already) and build up my career from there or go for a Phd in SDN or ML.

However i have decided to just learn ML (coursera) on weekends or whatever and focus on certifications while working.

What do you think?

I really can't say. I wouldn't want to think I'd put you off doing a PhD if that's really the best thing for you, but I can't judge if it would be. Only you can judge that, and all I'm trying to do is give you more information about what it is like. I'm really pleased I did a PhD, but it really isn't for everyone, it really is hard, and it really might not be of value to you.

But it might be. Weigh the possibilities. By all means start to apply the discipline you would need to start some self-study and see where it goes.

And whatever you decide, good luck!

You are not putting me off because i have been thinking about it hence the reason i decided to post it.

I will be applying to a University in Australia, New Zealand or Canada. Also it has to be on scholarship which i read is quite competitive.

Thank you for all the information, i will be looking at all options closely. Thank you again.

What's the end goal?
Either build a startup around ML in Africa or get a job on completion of my studies. So wanted to hear opinions from people in the field or doing this already.
If you get a full scholarship then probably, that's usually the case though. The PhD will give you prestige and credibility if you start your own business, and will open some doors when you are looking for a job (but also close others). It will also set you back a few years from your goals, and the amount of money you will get with a PhD won't be more than with a MS. Best perk of having a PhD is working in academia, which can be a great life style: much lower stress, decent compensation, lots of vacation time and recognition.
Oh yeah it has to be on full scholarship if accepted, great else i might have to just continue working and focus on Cisco Certifications.

Working in academia seems like fun but later in life not at this age :). Thanks