3
$\begingroup$

I am interested to know, if someone wants to be an AI expert, what should he/she know, as we can see this is a vast field today!

For example, if someone works on machine vision, should he/she know voice recognition or data mining?

In other words, should someone know everything from image processing to machine vision, if he/she wants to be an expert in that field or are there some specific subfields even in the vision section?

$\endgroup$
3
$\begingroup$

That depends on how you define the word "expert".

An expert in any field is a person who has an intuition that points him to the right direction in that field .

Take Feynman for example , He may not know about all of the physics of his time , but given enough time he could understand any research paper and can work on it without hesitation .

In the present situation there are very few "experts" in AI , but if you ask for one , my answer would be Geoffrey Hinton. Reasons:- Unlike most of the other researchers he doesn't come up with a new algorithm or a model by solely relying on mathematical proofs . If you observe most of his papers totally lack a formal mathematical argument. But his observations are almost always right . Dropout , Capsule nets , Greedy training are some of his innovations which are hardly mathematical but completely transformed the field. He is not from a formal Engineering/math/physics background unlike most researchers but entered AI because he wanted to know how the brain works.

Having said that there are many experts in subfields of AI , like yann lecun in convolutional networks , etc.

$\endgroup$
2
$\begingroup$

Congratulations, that is a very demanding question and difficult to answer. But I try my best, because the topic is interesting. At first, it is important to specify the person's skills. There are two possible perspectives on machine vision. The first one is called “Computer vision engineer”, which means that the subject has to do with C++ coding, setting up deeplearning hardware and implementing mathematical formulas. The second perspective is called “Machine vision artist” and emphasizes the background in photography, painting and gesture recognition in computer games.

Apart from the personal skills somebody brings in into the job, there are also two options for the subject itself. Computer vision can be discussed under graphical aspects, which is the traditional one and is realized with OpenCV and similar tools. Here is the idea to convert an image which is already there into machine readable data. The other option is, to consider computer vision in a linguistic context. That means, the raw image is not important, instead it is about language models, semantic networks and domain vocabulary and only in a second step, the model is grounded into perceived OpenCV data.

At the end, I would suggest to formulate the job description as follows: Target person is an artist with a background in computergames. And he has the obligation to work closely with a computer linguist to build the new generation of computer vision software.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.