General > General Technical Chat

Doctoral studies - perspective

<< < (3/7) > >>

CatalinaWOW:
Even though some of the advice here seems contradictory, from my point of view it is all correct.  This decision is very dependent on you and where you want to go.  There are a few careers where a PhD is a required check box.  Academia and some large corporations and some government jobs.  Beyond that it is wide open.

It has been my observation (I have worked with, managed or observed literally a few hundred PhDs and other professional people over my career) that very good people do well financially regardless of degree and narrow or weak people don't do well, also almost regardless of degree.  Success at getting a PhD is only loosely correlated with being well suited for a particular career.  Collecting a PhD involves several factors.   Obviously being good at academics.  But also putting up with BS, luck in advisors and academic committee members, persistence, creativity, writing skills, academic politics and a long list of other things.  Success at post PhD career depends on how well the mix of skills you brought to the table applies to the rest of your life.  Perhaps fortunately, perhaps not, all of those skills help some places and times.  Only you can determine your satisfaction with that result.

Tomorokoshi:

--- Quote from: Someone on June 18, 2022, 08:20:41 am ---
--- Quote from: jeremy on June 18, 2022, 06:24:16 am ---In Australia the most competitive scholarships still earn less than minimum wage.
--- End quote ---
At the same time, there are a small number of highly competitive scholarships paying well above the minimum wage:
https://scholarships.unimelb.edu.au/awards/john-monash-scholarship
before adding extra scholarships/prizes, and in Australia scholarships are non-taxed so you can work in other jobs or have other income and make some bank. Doing a part time postgraduate project while working is a great option for those with employers who will support it.

--- End quote ---

I know a person who did just that. It takes a good combination of employer, advisor, and student. The project was applicable to work. Perhaps one day per week was dedicated to the project.

Rinnake:
Thanks to all for your help. At the moment, we are finally thinking about whether we have not studied a less promising field (power electronics and control of regulated motor drives), when cybernetics, control and information systems, artificial intelligence, embedded systems, etc. are coming to the fore today ... We are beginning to think that those who would doing doctoral studies in cybernetics would be much more in demand nowadays.

rstofer:
AI in general and ML in particular are starting to fly!  The techniques are being applied to just about everything and one requirement is a strong background in Linear Algebra and, if the model is to be applied to something, there needs to be some kind of interface.  It's one thing to derive the model, it's quite another to actually drive an automobile.

There have always be suboptimal schemes for controlling production lines. Somehow, we're willing to accept a small discard rate.  Maybe ML can play into this and somehow provide guidance to the Programmable Logic Controllers (PLCs).  I have no idea how this works.

I would spend a bunch of time at NVIDIA learning how to use the Jetson Nano as a stepping stone to larger systems.
 NVIDIA is SERIOUS about AI.

https://developer.nvidia.com/embedded-computing

So is Google:

https://coral.ai/docs/dev-board/get-started

And so is PyImageSearch (for money, of course)

https://pyimagesearch.com/2020/03/25/how-to-configure-your-nvidia-jetson-nano-for-computer-vision-and-deep-learning/

Computerphile has put out some videos that are easy to watch:



I don't know where the stumbling blocks come up but that's where the PhD comes in.  Figure out a faster algorithm and the world will flood you with money.

I love MATLAB and they have some nice tools for ML.  At extra additional cost, of course.  I buy the Home license for MATLAB plus several add-ins so it runs a couple of hundred $ per year.  I like their digit recognition example because it shows a convolutional neural network as just a bunch of predefined blocks stacked together.  There's a lot of magic in the blocks but I don't need to know that.  I just chain them together and magic happens.  No, I don't truly understand what I am doing but I'm really good at 'cut and paste'.

Or Python with additional libraries or, in the case of NVIDIA (at least), C++ and Fortran compilers are available. 

The number of CUDA cores on the graphics card becomes an important consideration.  I bought an HP Laptop just because it has an NVIDIA RTX 3070 graphics chip with 5,888 CUDA cores (and it was affordable...).  That's a lot of parallelism!

Good luck!


Benta:

--- Quote from: Rinnake on June 26, 2022, 03:11:41 pm ---Thanks to all for your help. At the moment, we are finally thinking about whether we have not studied a less promising field (power electronics and control of regulated motor drives), when cybernetics, control and information systems, artificial intelligence, embedded systems, etc. are coming to the fore today ... We are beginning to think that those who would doing doctoral studies in cybernetics would be much more in demand nowadays.

--- End quote ---

Yes, and that's where everyone is applying today. Those are the "buzzwords" in fashion. You'll be one of the herd.
I remember from 30...40 years ago the buzz about "everything's going digital, so analog engineers will no longer be needed".
B*llsh*t. Good analog engineers are the best paid people today, because not many good ones are out there. The same can be said of RF engineers, PE engineers, sensor technology etc. All those fields that operate on the interface to the real world.
Digital engineers and progammers are a dime a dozen and can be hired cheaply in India or other low-wage countries.

The bar is much lower for AI etc. candidates. They just need a PC, which is why everyone is flocking to it.
Analog, RF, power, sensors need a million-dollar lab to just get started. If your university has that, use it to the hilt. You'll come out on top.
And VERY IMPORTANT: you need a Mentor. Can be a professor, but also from industry.

Very personal opinion, but from the heart.
Good Luck.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod