Category: TensorFlow Machine Learning

How a machine appears to be learning?

A machine is a general term that are consist of many components. It has memory component to store or record observable distinguishing combination of letters for natural language processing, combination of numbers for computation, combination of shape for object detection and classification, combination of pixels for vision, combination of waveform for voice recognition, combination of sequence of causes and the associated known output for making probabilistic and deterministic decision, and so on.

The machine has many input components such as camera, microphone, keyboard, mouse, touch screen to trigger the processing of many possible combination. Finally it has output component using voice speaker, video screen, mechanical actuator, and whatever human interface to communicate the output to human or machine language to talk to another machine that will use the output as another input.

As you can see from many components of the machine that needs to be properly integrated to appear to be learning THERE ARE MANY HUMAN BRAINS needed to operate, maintain, update, troubleshoot any machine learning. Conclusion, machine is not yet learning by itself. People are responsible for training the machine to learn whatever people want them to learn to further augment the observable human deficiency. Human have limited memory capacity so human invented computer database to increase his memory. Human have limited computation capacity so human invented computer algorithm to increase his computing speed, human have limited association and anomaly detection capacity so human invented artificial intelligence to increase his pattern recognition and anomaly detection…So on. Human will continue to solve his observable deficiency. Source: By Sam Ortega, founder of IN-V-BAT-AI

How a machine appears to be learning? A machine is a general term that are consist of many components. It has memory component to store or record observable distinguishing combination of letters for natural language processing, combination of numbers for computation, combination of shape for object detection and classification, combination of pixels for vision, combination of waveform for voice recognition, combination of sequence of causes and the associated known output for making probabilistic and deterministic decision, and so on. The machine has many input components such as camera, microphone, keyboard, mouse, touch screen to trigger the processing of many possible combination. Finally it has output component using voice speaker, video screen, mechanical actuator, and whatever human interface to communicate the output to human or machine language to talk to another machine that will use the output as another input. As you can see from many components of the machine that needs to be properly integrated to appear to be learning THERE ARE MANY HUMAN BRAINS needed to operate, maintain, update, troubleshoot any machine learning. Conclusion, machine is not yet learning by itself. People are responsible for training the machine to learn whatever people want them to learn to further augment the observable human deficiency. Human have limited memory capacity so human invented computer database to increase his memory. Human have limited computation capacity so human invented computer algorithm to increase his computing speed, human have limited association and anomaly detection capacity so human invented artificial intelligence to increase his pattern recognition and anomaly detection…So on. Human will continue to solve his observable deficiency. Source: By Sam Ortega, founder of IN-V-BAT-AI

Since the dimensions could be in thousands, millions, billions or even trillion of parameters we need the machine to help us learn the pattern. Again the basic principle is the same, the machine or computer did not learn by itself, humans give the training inputs to discover the pattern or relationship. The generalized pattern discovered by machine learning can no longer be represented by known mathematical equations that human can see.

Linear regression is a good visualization to explain machine learning. The machine or computer actually did not learn by itself to discover the generalized line equation in slope form. In this example you gave the computer six (6) training data points to approximate the best line equation.