Machine Learning: The key to human learning
Machine learning is a technique that allows computers to learn and gain intelligence through experience. Currently, it is being widely used by Google to learn more about its users to provide them with more accurate information. In order to expand their machine learning capabilities, Google took over Deepmind Technologies, a British Artificial Intelligence company, in 2014. Deepmind has conducted research on using artificial intelligence to play various Atari Games like Space invaders. At the initial stage of the research, the computer was losing badly as it had no strategy except for the basic rules of the game. However, the computer continuously played the game for a period of 24 hours. Finally, through the strategies it has learnt over the training period, it was able to tackle the game easily.
This incredible technique used by computers has inspired me to think how we could adapt this technique to enhance the way humans learn something. On a superficial level, just like how machine learning requires the computer to repeat the process and learn from it, we humans too could continuously practice and learn from each repetition. As the saying goes, ‘practice makes perfect’, we humans would eventually master it but over a span of months rather than a day.
However, when I delved further into this vast field of machine learning, I learnt about the various types of machine learning. The following part of my article explores how these types could be adapted to suit our learning needs:
Supervised Learning is where the system is fed with training examples of input and output. By studying these examples the system formulates a function to produce outputs for new inputs. This method would prove to be very useful for studying well-structured content such as test material. We could go through a set of sample questions or problems with solutions and analyse the pattern of various ways to approaching and solving them. A general method could be formulated to tackle new questions with ease.
Unsupervised Learning, on the other hand, involves the system inferring a model from a set of unlabeled inputs. The system identifies key features in the data. Along the same lines, humans could learn a new language. For instance, we could watch numerous videos of the language and find a pattern of how its sentence structure works and even the meaning of some words. This is a top-down approach to learning a new language instead of the bottom-up approach where we start with basic words with their translations which could a take much longer time.
Last, but not least, Reinforcement Learning. It involves a feedback system where the computer initiates and action and the environment returns the relevant feedback. This technique can especially be useful in learning soft skills such as public speaking. You could start speaking in front of a group of friends who could give you feedback which could be used to improve different aspects of public speaking. Then, you could advance to speaking in front of a growing number of audience. You could use the feedback constructively and reinforce your style of presentation and material with the criticism provided to make your speech better.
Though there are many other machine learning algorithms, I found these three basic algorithms being able to provide us with some motivation to fine tune our way of learning. If a computer can win a game of Go against a world champion, why can’t we?
P.S. This is my first Medium article. Please comment on how I can improve further. Thank you!