mind vs machine

In 1792, Mary Walstone Craft questioned her work "Protection of women's rights". "What are the benefits of human beings in savage?" She replied. "In reason and virtue, human beings can acquire some knowledge." No one thinks that men and women are intellectually equal or that they have excellent intelligence in barbaric creation today But they are thinking about human beings and machines. We feel that we are always better than animals because "conclusion, judgment, or reasoning and inference from the premises" (Random Bookstore Dictionary).

Two years ago, our co-founder Jeff Hawkins and Donna Dubinski wrote a blog article "What is Machine Intelligence and Machine Learning and Deep Learning and Artificial Intelligence (AI)?" Numenta's advanced evaluation is that there are three major categories in this field, we call it classical AI, simple neural network. As shown in the table below, the biological neural network

As detailed in the article on Machine Learning 101 and Linear Regression, problems to be solved by machine learning can be categorized mainly into supervised machine learning and unsupervised machine learning. A supervised learner learns from tagged data, such as data on house characteristics. This includes house prices for housing price forecasts. In other words, the monitored machine learns to learn marked data points and predict future tags.

It is difficult to understand how everything works with terms in the jungle of terms like supervision and no teacher, or regression and classification. However, machine learning under these complex levels has some basic ideas to deepen the understanding of the whole picture from deeper, more mechanized understanding. Machine learning is focused on finding rules (thinking about functions) that use parameters (also called weights w) to predict the output y of the input x. When "training" the algorithm, define x, y, and w as the minimum cost function. In general, the cost function is defined as a change in the difference between the predicted value and the actual value. This function defines our "error" for prediction - which explains why it is called a cost function. High cost! Since x and y are known before learning, we try to find out which w finds the lowest point of the cost function.