Human learning and machine learning

One day, my junior asked me, how to improve programming skill. Then I told her, only practicing a lot and a lot. Then, I told her about how can we learn like machine learning.

Just like machine learning (such as Artificial Neural Network), we need a lot of training samples. More samples will produce better result. So, as we learn, we also need to have a lot of samples. Then, the learning process need to be repeated. After enough learning, a person can surely categorise the new sample. Because, even a machine can do so. For example, an expert is able to appraise the fake painting and the real painting. This is because he or she has many experiences.

This makes me aware that, there are 3 points about learning. Firstly, the talent, that is the innate perception of a person. Similarly, for the machine learning, the innate perception is the learning algorithm, for ANN, it will be the network topology. Thus, different person has different talent; different learning algorithm can solve different type of problem.

Secondly, the training samples of machine learning. For the human, it is the experiences. The more experiences will produce better learning.

Thirdly, the learning output of the machine learning. For ANN, it will be the weights (connections) output. Similarly, the learning output is the knowledge. One can get the knowledge of the other person, but cannot get the experience, because he or she doesn’t have the training samples.

Therefore, when you have a talent, and apply your talent in a correct field. The experiences you gathered will become the knowledge of your own. Though you can pass your knowledge to someone else, unless he or she has same talent as yours, he or she cannot understand the knowledge. If someone has the same talent as yours, unless he or she experiences what you experienced, he or she cannot produce the knowledge like yours.

However, human learns better than machine, because human can learn how to learn.