10 skills you should be Learning before Machine Learning

“Stay here. I’ll be back”.

These are probably the most famous words ever told by a machine.

If you are a movie buff, you’ll know the above reference.

This dialogue is from the movie Terminator 2.

As told by Arnold Schwarzenegger.

Terminator 2 machine

He plays T-800.

The greatest machine ever built 😊.

This movie was released back in 1991.

You see, man has been obsessed with machines for a long time now.

Even before 1991 of course.

Especially in movies (Hollywood), machines, either real remote controlled ones or the ones created by CGI effects are quite common.

Machine Learning has existed for some time.

Be it movies or real life.

And it has gained human attention in almost all the business domains.

Globally.

Currently, we have not yet reached the movie-level machine capabilities.

But we are on our way.

In this post, I’ll cover first and briefly,

-what is Machine Learning

-what does it contain

I won’t go deeper than that as it is a vast subject.

What I’ll mainly cover are the skills you should not only be aware of but also be good at before you start a course on Machine Learning.

The purpose is to help you properly decide whether you are actually ready for this or not.

Cause I’ve seen many people who have paid for an online course on this.

And then simply dropped off in between.

And why?

As far as Ive seen, there are 2 reasons for this.

1. Too many new concepts

2. You need to thoroughly understand basics

Which makes Machine Learning more difficult than initially perceived.

I will also provide a list of sites to learn Machine Learning for FREE.

It would be a great way for you to check if you understand what you are getting into.

Go through these courses mainly to capture concepts and the underlying technical data rerquirements.

So let’s begin?

Whether you are aware or not, the world is slowly getting encroached by machines.

Many of the work that you do or come across is carried out by machines.

Google search

Google search machine learning

Netflix/Hotstar movie recommendations

Netflix

Alexa voice assistant

Amazon Alexa

What exactly is Machine Learning?

The definition is

Machine learning is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed.

Let’s take a Machine Learning example

Say, there are three coins.

Each with different weights

Coin A = 10 grams

Coin B = 20 grams

Coin C = 30 grams

Three Coins with weights

You have a big bundle of coins.

A mixture of all three.

Coins bundle

You want a machine to segregate this bundle.

What do you do?

You feed the machines relevant data.

As input.

Where you inform that,

If a Coin weighs 10 grams, it should be Coin A.

If it weights 20 grams, then it is Coin B.

Similary for Coin C.

The machine will now check the weight every time.

And split the bundle accurately.

Into 3 sets.

As per the instructions YOU gave it.

Now, here’s the next level.

What if there is a coin that weighs 40 grams, or 15 grams?

Or any weight that is not the weight of Coin A, B or C?

What will the machine do then?

Well, either it will stop working or ignore that coin.

Or do something else as instructed.

So you see, here, the scope of the machine productivity, although good, is limited.

In order to use the machine every time, you need to give instructions to it for all the scenarios that it will encounter during the work process.

In other words, you provide the input AND output.

The work in between is carried out by the machine.

This process saves time no doubt.

But what if the machine can do the work without any human input?

In the above coin example, what if the machine can identify the fourth coin and classify it even without any instructions that you provide?

In order to do that, the machine has to LEARN BY ITSELF.

So, in simple terms, Machine Learning is a process where a machine/system learns or evolves on the work it does ON ITS OWN.

It needs to make decisions from data on its own.

Without being programmed.

How does a Machine learn on its own?

Using the data it has and the experience it gains while processing.

Widely speaking, there are 3 broad categories of Machine Learning.

1. Supervised

2. Unsupervised

3. Reinforced

What is Supervised Machine Learning?

Here, the machine already knows what the target output is.

Like the above coin example.

The weight becomes the feature.

And title will be the label.

Once the data is fed, the machine will learn which feature corresponds to which label.

Which is Supervised ML.

In Supervised ML, you need to train and test your model before you deploy it.

The machine will use the labelled data to learn.

On its own.

What is Unsupervised Machine Learning?

Here there is no labelled data for the machine to learn.

Data clusters usually come under Unsupervised ML.

Since the data is not labelled, there is no way to measure  its accuracy.

So if the data is not labelled, how is the machine learning?

With the data that is fed, the machine will create some sort of clusters based on certain data features.

Example:

Say we have a list of weights and heights of people but without their gender.

Height and weight

The machine will learn the structure of the data.

Even though it is not labelled.

With this structure, the machine will cluster the data.

Into groups.

More specifically the gender groups.

Thus, the machine here uses a Cluster Algorithm.

Coming to the third Machine Learning type,

What is Reinforced Machine Learning?

This type of learning works on the basis of feedback loop.

Say, there is an image list of cars, bikes, cycles, planes etc.

If a machine identifies any of these incorrectly, like for an image of a car, it displays the output as a plane.

You then feed the right data to the machine.

So that henceforth, when the machine encounters an image of a car, it will label it correctly as a car.

Currently, of all the three types, Supervised Machine Learning is used the most.

It is relatively easy to understand and simpler to implement.

That was a brief intro to Machine Learning.

It was. Believe me.

Now, coming to the main topic of the skills you need to start on Machine Learning.

Ok, you would have come across a lot of Youtube Ads, Google Ads on Machine Learning courses, right?

Also, on job portals and Linkedin, you would have come across data or charts that show the scope of Machine Learning in the future.

Machine learning lead form

The salary and the hike you get if you know Machine Learning.

It makes you want to sign up for the course at the earliest.

It is mind-blowing right?

It is yes.

So what’s the hurdle here?

The question in hand is:

Can you directly learn Machine Learning on your own?

And what do I mean by that?

Simply, if you can directly sign-up for an online course and start learning?

Sure you can.

As with any other course, it depends on the individual.

There are many that would tell you the same.

Specially, the sales people from the academy that provides the course.

It is unlikely that they will tell you all the things you need to know.

It is only when the course begins that you realise the actual intricacies involved.

But here’s the thing that you DON’T see in the digital network.

Machine Learning has many components.

A few are:

Deep Learning

Neural Networks

Image processing

Algorithms

Clustering & Segmentation

Time series

Visualization

Chatbots

Decision Thinking

Deploying

Internet of Things

Speech Analysis

As I said, these are just a few.

So here’s the most important statement:

In order to learn Machine Learning, you’ll need to know the Basics.

But basics of what?

I’ll get to that.

But regarding the topics, I listed above, you see how incongruent Machine Learning looks?

What I want to convey is that Machine Learning course is more advanced than one might think.

If you directly join a course without doing a basic preparation, you will most likely end up losing money on the course.

By leaving it mid-way.

So, the main topic of the post:

Below are the basics that I mentioned earlier that you should first work on.

1. Programming language

This is a no-brainer. You need to be efficient in languages like Python or R.

Although these are promoted as easy-to-learn languages, it still will take you months before you actually become proficient.

2. Probability and Statistics

This is another hectic huge area that you’ll need to learn or rather re-visit.

Not only should you learn the very basics (Mean, Median, Mode, etc) but also advanced concepts like Naïve Bayes, Markov Models, p-values among others.

3. Algebra

Remember all the high-school concepts?

You should.

I’m talking about variables, coefficients, linear equations, histograms, etc.

4. Calculus

Derivatives, Chain rules, Slopes etc are a few concepts that come under this.

5. Computing

Most of the time, you will work with large data sets.

For which you will need multiple machines.

For this, you should know Apache Hadoop and other cloud computing methods.

6. Unix tools

You need to be great at Unix tools like cat, grep, sort, sed, etc. as most of the machines will be Linux-based.

7. Algorithms and Libraries

There are many Machine Learning algorithms available through libraries.

Like: Tensorflow, Theano Spark, etc.

These involve models like vector machine, decision tree among others.

8. System design

The output of a Machine Learning algorithm will still be a small piece in a bigger picture.

You’ll need to understand how these will be connected to the output that you got.

And learn to connect with them.

9. Data modelling

This refers to the ability to arrive at the structure of a given data set.

You will also need to evaluate regularly to check for the best data model that fits.

10. Lastly, but not the least, you will need to read.

A lot.

List out the websites and channels that you find useful.

Follow them regularly. Buy books on the same.

Read. Learn. Repeat.

Phew!

Was that a lot? Do you feel like the below?

Multi-tasking

Well if not, here’s some good news:

These are just the prerequisites.

Once you know all these, you will be challenged on other advanced Machine Learning concepts as and when you go through it.

As I mentioned at the beginning,

Below are a few sites where you can start with Machine Learning for FREE.

Coursera

Udacity

Google

DataCamp


Conclusion

Again, I stress upon the fact that, by detailing the above, my intention was not to scare but to educate you.

Machine Learning, by all means is lucrative field world-wide.

But this stream requires an extensive level of knowledge across various fields.

As the saying goes, look before you leap Smiling emoji.

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *