How to develop your AI intuition

Prince Canuma
Sep 2, 2018 · 5 min read

In this article, I’m going to help you develop your inner compass to navigate this uncharted field.

Chill, I got your back.

Having the sensation that you are doing the right thing or that you are on the right track at least, can reduce the distance between your goal and the work you put towards achieving that goal (a project you are developing for example).

It is frustrating to get stuck, no one likes it, but worst than that is going through the learning process and still when it comes to applying what you learned you have no idea or intuition on how to get started or overcome a roadblock.

“Fear not !” I say, I’m not perfect but I have failed enough to say that failure is your best friend when it comes to diving on a new field, learning, developing or solve a problem and etc. Steve Harvey said :

“Failure is a great teacher, and I think when you make mistakes and you recover from them and you treat them as valuable learning experiences, then you’ve got something to share.” — Steve Harvey

Not only it will teach you but it will also impact on your internal compass, it will create this intuition that will guide you to the other side where success is just there waiting for you. But it doesn’t end there, as Steve Harvey says :

“ A person has to remember that the road to success is always under construction. You have to get that through your head. That it is not easy becoming successful.” — Steve Harvey

You have to embrace failure!

If you want to succeed in the any field not only AI you’ve got to get the basics right. The most successful people, work on the core basic skills more than everyone else.

GET YOUR MATH RIGHT

See what I did here ? Math — Matrix.

If you really want to be a badass in the field of AI, you’ve got to first, get your math right. Meaning Linear Algebra and Calculus. You don’t have to master them, because this are vast fields and it's hard even for people with fancy PhD degrees to master them. YOU ONLY HAVE TO KNOW THE BASICS, such that it is enough for you to understand the content related to AI.

It’s crucial you understand this topics.

I have had the luck of being introduced to this topics on my 1st semester of CSE, even though I didn’t stop there, I bought a book of Linear Algebra and I keep my notes from the 1st semester so that when needed, I can revise a topic or method that is running away from me. But that might not be your case that’s why I have also looked for non-conventional methods and found a ton of Online resources to learn from namely:

With your math right you got yourself an advantage, and it will save you a lot of time down the road, because you can use as many black-box methods that abstract the real work, but reality is that you need to get dirty sometimes and you better be ready to get dirty or else you will be caught with your pants down and you don’t want that, do you ?

What do I mean by “getting dirty” ?

“You can run, but you can’t hide.”

I see you. Thought no more maths, didn’t you ?

AI is a fast changing field, it is evolving at a mind-blowing speed. The problems we face in the real world are very complex and some of them need custom systems, meaning you will need to create a custom AI system, and to do that you will need the knowledge of the math behind those fancy high level functions, to actually change it or bend it to your own will .

Marvel fans, prepare to be amazed with what you can do with this technology.

Tony’s closest

Let’s say we want to create a system that can detect and label Iron Man suits in a image/video. A system like this has different variables to take into account, from the details that help identify or distinguish different suits (features/input data) and the corresponding label class (name of the suit/output).

“Machine learning algorithms all seek to learn a mapping from inputs to outputs” — Jason Brownlee, Ph.D.

This algorithms are functions that predicts the output/label (y) given the input /image (X) and it is called Hypothesis h(x).

Jarvis what kind of problem are we facing now ?”

So, from requisites above mentioned for our system, we can see right away that this is a classification problem.

“Classification is a technique where we categorize data into a given number of classes. The main goal of a classification problem is to identify the category/class to which a new data will fall under.”— Rohit Garg

For our system we can use Logistic Regression.

Definition: Logistic regression is a machine learning algorithm for classification. In this algorithm, the probabilities describing the possible outcomes of a single trial are modelled using a logistic function.

Using this function our hypothesis will look like this:

This hypothesis maps all predictions greater than 0.5 as 1 and bellow 0.5 as 0 (0.5 ≤ h(x) ≤ 1).

Hulkbuster — input/image(x)

h(x) will give us the probability that our output/label(y) is 1 (100%). For example h(x)= 0.9, this gives us a 90% probability that it is Hulkbuster(y) from image above, and a probability that it is Mark VI(y) of 10% because it has a small resemblance to Mark VI. The sum of the two probabilities is 100%.

This is just an example of where and how, having basic math knowledge could come in handy.


Conclusion

With more understanding of the basic math for AI, you will have an amazing intuition and you will create amazing things with technology.

“Remember great things take time to grow so be resilient and stay hungry.”

If you liked this post please give me a round of applause👏🏽 👏🏽,it will contribute and mean a lot to me, giving more strength to get you the best AI news.

Prince Canuma

Written by

Computer Engineering Student, Web Dev. & AI/ML dev

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade