A Basic Intro to CNNs in Machine Learning
For beginners interested and curious about ML.
What is machine learning?
In machine learning, humans try to train the computer to learn in a similar way we do, such as identifying that an image is a cat by looking at different images of different cats. We learn that an image is a cat by seeing that it has two pointy ears, a tail, and four legs (most of the time). This is similar to how machines identify an image as a cat — by being shown hundreds, thousands, and even millions of cat images, eventually the machine will learn and identify a cat’s most important and distinctive features that make it a cat. Therefore, when shown an image of a cat, the computer can identify that it is indeed a cat.
What are CNNs?
CNNs stand for Convolutional Neural Networks, and they are one of the most popular models used in machine learning. They can be used to identify and classify between different sounds, different images, etc. For example, if shown numerous cat and dog images, a CNN can eventually learn each animal’s distinctive traits and classify a new example of an image into the correct category.
How do CNNs work?
CNNs contain kernels, which are matrices of numbers that serve as weights. These weights can help determine which part of data is most important. A kernel scans over data, such as an image, for example, and the kernel’s numbers are multiplied with the inputted data (each part of the data can be transformed into different numbers) to extract the most important or relevant features and patterns to produce an output. In classification tasks, that output could be the name of a category of an image, like cat or dog.
Why are CNNs important?
Machine learning utilizing CNNs can apply to many areas of life. For example, in the medical field, CNNs can be trained to classify between different types of skin cancer and serve as an early detection device for patients.