Machine Learning (ML)
So there has been a lot of talk on Machine Learning (ML) for the past 6 decades. One man in particular, Geoffrey Everest Hinton, a cognitive psychologist, and computer scientist has been on the forefront. His notable work has been on artificial neural networks which have been a building block to what we now know as Artificial Intelligence (AI). So, what really is ML? This is the study of computer algorithms that improve automatically through experience. Arthur Samuel (1959) described ML as a field of study that gives computers the ability to learn without being explicitly programmed. In the early 1950s, Arthur Samuel developed a checkers game and allowed the computer to play against himself and itself, with time the computer came to learn the best board positions to take and bad board positions not to take when playing the checkers game board. The computer finally became better than Arthur Samuel in the game of checkers. This is because a computer could play 20,000 hours and more of checkers game in a short time while for a human this is an impossible feat to achieve. There have been various use cases of ML since that time. Most probably in your day to day activities, you have interacted with ML without knowing. ML is broad, here are a couple of things you need to know as there are various learning algorithms that ML uses;
- Supervised learning algorithm which involves the use of training data to get the desired outputs from one or more inputs. The training data is put through a mathematical model, this algorithm if it improves over time and gives the desired outputs from the training data, this suggests that the computer has learned. A good example will be to take various photos of mangoes and mark it as the desired output and then take photos of mixed fruits ensuring that mangoes are there too and make this as an input data. If the algorithm identifies the mangoes in the photos of the mixed fruit, then the computer has learned if not, you keep trying until it gets it correctly, use cases have been applied on facial recognition software.
- Reinforcement learning involves the area of machine learning where the machine in a certain environment is provided data and awarded if it gets the data correctly. Look at it like training a dog, if you tell the dog to sit down and it understands you, you give it a cookie, if not you repeat until it gets it and then you reward it.
- Unsupervised learning algorithm is where data with only inputs is taken and the machine gives it structure depending on the commonalities the algorithm finds.