Andrew Ng’s Machine Learning Simplified — Part 1
A simplified version of the renowned Machine Learning course.
Today we start with Week 1 Part 1 : Introduction👇
The good old machine learning definition -
A machine is said to learn from experience E if it’s performance on task T as measured by performance measure P improves with E .
Ehhh too wordy? Let’s break it down.
Take the example of Email spam classification which is the most common way all of us experience ML in our daily lives.
Here, the task is to classify mails (T)
The performance of the algorithm would me measured by the percent of correctly classified mails (P)
And the experience is watching the user manually classify (E)
So if this E makes the algorithm better as indicated by P, we say:
Now let’s look at the types of Machine Learning Algorithms. They can be broadly classified as
- Supervised Learning
2. Unsupervised Learning
3. Semi-supervised Learning
4. Reinforcement Learning
Let’s discuss the first two in detail :
1. Supervised Learning
Here, we let the machine learn with supervision at each step. Providing the correct answer be in label or value for each of the sample and then letting the machine figure out.
Similar to Classwork where teacher solves all the problems on the board herself.
Supervised Learning can be further broken down to two types -
(i) Classification — dividing samples into categories, discrete output. Ex: pass,fail
(ii) Regression — producing continuous output like a number. Ex: percentage
Next, 2. Unsupervised Learning
Now as the name suggests, we give no supervision here. The machine is just fed with data and is required to find patterns and clusters that are unknown before. Similar to the homework and assignments where the students figure the solution out themselves.
Unsupervised learning finds application in a lot of industries for forming clusters, making customer segmentations, grouping articles together, finding cohesive groups in social media and separating audios among many.
For a thread version of the blog : https://twitter.com/Aakriiti_Sharma/status/1466368130603646981?s=20
