What Does It Mean When AI 'Learns'?
To equip AI with the ability to solve problems independently, it first needs to go through a process called learning.
In this session, we'll explore how AI develops intelligence through learning and examine key concepts involved in this process.
What Does It Mean When AI Learns?
When AI learns, it means creating a model that can solve problems through data.
An AI model is a program that receives input data, processes it, and produces the desired output.
Two key concepts in this process are weight and bias.
Weight
When an AI model processes input data, the weight is the value assigned to each input to indicate its importance.
The higher the weight, the more the input influences the model.
For example, in a model that determines whether an email is spam or not, if a specific word is more critical in deciding spam, the weight of that word will be higher.
Bias
When an AI model processes input data, bias determines how easily the model can produce appropriate results.
The higher the bias, the more aggressively the model processes the input data.
Training AI involves adjusting these weights and biases multiple times to create a model that can process input data correctly.
Data Collection and Preprocessing
The first step in AI learning is collecting data.
To build an AI that distinguishes between dogs and cats, you'll need thousands of pictures of dogs and cats. If you're creating an AI that generates text, you'll need textual data such as news articles and novels.
However, collected data is not always perfect, necessitating a process called preprocessing.
Preprocessing involves cleaning the data and making it easier for the AI model to learn.
Key Preprocessing Steps:
-
Handling Missing Data: Compensating for or removing data that has gaps.
-
Normalizing Data: Standardizing the range of data to make it easier for the model to learn.
-
Data Augmentation: Artificially increasing the amount of training data by rotating images, altering text, etc.
Preprocessed data becomes the prepared material that the AI model can learn from.
Want to learn more?
Join CodeFriends Plus membership or enroll in a course to start your journey.