What is Machine Learning?
Machine Learning is a field of Computer Science that uses statical technologies to give computer systems the ability to 'Learn' with data, without being explicitly programmed.
That means, "ML is all about Learning from Data"
Explicit Programming means, writing codes for each scenario, to handle that situation.
In machine learning, instead of writing explicit code for each scenario, we train models to learn patterns from data, allowing them to make predictions or decisions for unseen situations.
So, We give input and output, but don't write any code for each and every case. ML Algorithms automatically handle them.
An simple example can use:
Summation Function:
In explicit programming, to add 2 numbers, we write specific code that works only for that case. This code won’t work for adding 5 or N numbers without modification.
In contrast, with ML, we can provide an Excel file where each row contains different numbers and their sum. As the ML algorithm trains on this dataset, it learns the pattern of addition. In the future, when given 2, 10, or N numbers, it can perform the addition based on the learned pattern, without needing specific code for each scenario.
Where we are using ML?
- Email Spam Classifier:
In explicit programming, I wrote multiple if-else conditions, such as: “If a keyword appears 3 or more times, it will be flagged as spam.” For example, if the word “Huge” is used 3 times, it’s marked as spam.
Now, imagine an advertising company realize there’s an algorithm like this to detect their spam. So instead of repeating “Huge” 3 times, they use synonyms like “Huge,” “Massive,” and “Big.” In this case, the original rule wouldn’t work. What would be the solution? Should I again change my previous algorithms? How many time I will able to do that?
In ML, the model learns from the data provided and automatically creates algorithms based on that data. If the data changes, the algorithm adjusts accordingly. There’s no need to manually change the algorithm, it will update itself as needed based on the new data.
- Image Classification:
In explicit programming for image classification, we would need to manually write rules to identify features of a dog, like its shape, size, fur color, or tail. These rules would only work for specific images and would not generalize well to all dog breeds. If we encountered new breeds or variations, we would need to add new rules for each one.
In ML, instead of writing specific rules, we provide the model with a large dataset of dog images labeled by breed. The model then learns patterns from the data, such as the common characteristics of different breeds, and uses that learned knowledge to classify new dog images, even if it hasn't seen those exact breeds before. The algorithm automatically adapts to variations in the data.
also, there are thousand of uses of ML. You might wonder,
why wasn’t machine learning as popular before 2010?
- Limited storage capacity made it difficult to store large amounts of data due to the shortage of hard drives.
- There wasn’t enough available data to effectively train machine learning models.
- Hardware limitations, such as less powerful GPUs and processors, restricted the ability to run complex algorithms efficiently.
Nowadays, we are generating millions of data points every day. Using this vast amount of data, ML models are now becoming more accurate, efficient, and capable of solving complex problems. They can learn patterns, make predictions, and automate tasks across various fields such as healthcare, finance, and technology, improving decision-making and driving innovation.
Thank you for taking the time to read through this.
Top comments (0)