IBM defines machine learning as a branch of artificial intelligence and computing that focuses on using data and algorithms to mimic the way people learn and gradually improve their accuracy.
However, we mostly see examples of machines learning and emulating only the most superficial aspects of the subject.
Social media, for all its shortcomings, is adept at identifying images that instantly reveal the underlying truth of a situation, and this tweet shows us just how far machine learning still has to go.
This picture of puzzle pieces randomly arranged in piles of the same color to create a recognizable imitation of the desired result will surely come to mind the next time a Tesla crashes into a police car parked on the hard shoulder because of it machine learning can’t piece together the puzzle pieces of a car parked on the hard shoulder where no cars normally are.
UC Berkeley identifies three parts of machine learning algorithms:
A Decision Process: Generally, machine learning algorithms are used to make a prediction or classification. Based on some input data, which may or may not be labeled, your algorithm creates an estimate of a pattern in the data.
An Error Function: An error function is used to evaluate the prediction of the model. If examples are known, an error function can perform a comparison to assess the accuracy of the model.
A Model Optimization Process: When the model can better fit the data points in the training set, the weights are adjusted to reduce the discrepancy between the known example and the model estimate. The algorithm repeats this evaluation and optimizes the process by automatically updating the weightings until a precision threshold is reached.
Supervised learning: The dataset used has been pre-marked and classified by the users so that the algorithm can see how accurate its performance is.
Unsupervised learning: The raw data set used is not tagged and an algorithm identifies patterns and relationships within the data without the help of users.
Semisupervised learning: The data set contains structured and unstructured data that lead the algorithm to independent conclusions on its way. By combining the two types of data in a training dataset, machine learning algorithms can learn how to label unlabeled data.
Reinforcement learning: The dataset uses a “reward / punishment” system that provides feedback to the algorithm to learn from its own experience through trial and error.
Source link
