I am fascinated by the progress in machine learning and how it is impacting our lives.
I watched a youtube video, in which Andrew Ng, the leading expert in machine learning, demonstrated a device that is being developed by Baidu research for the blind people. The wearable device is composed of a camera, a computer chip and an earplug. The camera functions as the “eyes” for the blind and it takes pictures of the surroundings and feed the images to the computer chip. The computer is able to read and translate the images into descriptive words and the words are then being converted into speech and then fed into the blind person’s ear in real-time. How brilliant and remarkable this is!
Machine learning has been in use for lots of small convenient things like email spam detection, Netflix movie recommendations, Amazon similar product recommendations, machine recognition of handwriting postcodes. The list goes on and on. Most notably the recent promising progress in self driving cars is hugely attributed to the progress in machine learning especially in its application in the computer vision field.
In this post, I will attempt to give a very gentle introduction to machine learning.
Machine learning is a sub-field of artificial intelligence where special types of computer algorithms are used to learn and draw conclusions from known data. The computer visionary Alan Turing in his famous 1950 paper computer machinery and intelligence talked about a machine that can learn like a child to achieve computer intelligence. This is pretty much where the machine learning concept was born.
The main difference between machine learning and the traditional programming lies in the fact that in traditional programming, the programmer needs to give very explicit and specific rules to tell the computer what to do step by step, while in machine learning, the computer can “learn” from data and experiences instead of hard-coded rules. The more data and experiences/scenarios you feed into the computer, the better the outcome is.
At the core of machine learning is the learning algorithm. There is a saying that if AI is the rocket, then machine learning algorithm is the engine, and data is the fuel.
Machine learning is NOT going to replace the traditional programming methods. It just solves problems that traditional programming methods cannot handle.
In the next post, I will talk about the types of machine learning algorithms.
