1. Introduction
The decision of using a committee for managing R&D funds, rather than letting squabbling bureaucrats go on fighting, was learned from the remarkable success of the US military research into AI and machine learning. The scientists and expert system builders who had pioneered AI wanted to deal with expert systems with hopes for building a future machine that can push the aircraft envelope during dogfights with human experts. After devoting substantial funds into expert systems, it became overwhelmingly clear that while they sold a good solid vision, the technology itself was not mature enough to put into practice. The solution was to place a modest investment into machine learning, where the expected value payoffs were potentially revolutionary techniques for automating various forms of data analysis and making data-analytic systems exhibit adaptability unseen in their software ancestors. The risk was low due to already valuing breakthrough toolbox software for data analysts and having a potential force multiplier in creating future intelligent machines. Machine learning turned out to be a money pit of tantalizing techniques with value added here and now, and a huge success in the context of technical feasibility. Planning an AI revival throughout the 21st century, increasing reliance upon machine learning would be a linchpin keeping experience in the loop as technology autonomously does things that would require human intelligence. But success on this grander goal would also bring unexpected triumphs. In the private sector, new machine learning techniques have led to loosely defined businesses such as predictive analytics and data mining. With the understanding about what these tools are and some of their potential impact, this essay is still about using machine learning to learn specific things in the context of AI and data analysis.
1.1. Definition of Machine Learning
High-performance machine learning is closely related to computational statistics, which focuses on making predictions using computers. Many learning problems are formulated in terms of minimization of some loss function on a training set of examples. This is said to be an optimization problem. Restrictions can be imposed on the possible inputs and/or outputs that can be used in learning. Restriction to a given set of input and output representations is said to be pattern recognition. This is in contrast to unsupervised learning, which learns to find patterns in the input data with no example outputs given.
In the field of computer science, artificial intelligence, a connectionist model closely related to human learning, has been using machine learning methods, paying attention to neural networks. High-performance data programs have been using machine learning methods under data mining for the construction of a model of consumer credit risk. Specifications are all examples of attempting to automate the process of human learning. This is what we define as learning to do a task.
Machine learning is used in many computing tasks where designing and programming explicit algorithms with good performance is difficult or infeasible; example applications include email filtering, detection of network intruders or malicious insiders working towards a data breach, optical character recognition (OCR), search engines, and computer vision. In the medical informatics field, machine learning methods have been used to develop decision support systems for the diagnosis of various health conditions. Machine learning methods are used for adapting to the up-to-date digital libraries and searching for medical journal articles.
Machine learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed. The process of machine learning is similar to that of data mining. Both systems search through data to look for patterns. However, instead of extracting data for human comprehension as in the case of data mining, machine learning uses that data to detect changes in the data and adapt program actions accordingly.
1.2. Importance of Machine Learning Applications
Machine learning is becoming more important with the increasing amount of raw data available. Machine learning applications are used for data mining, medical diagnosis, detection of credit card fraud, stock market analysis, general classification, image and speech recognition, and more. Machine learning applications are now being integrated into interactive data mining, which can be used on any query. In today’s computing world, it is becoming more important to extract information from huge databases since the amount of stored information is growing at a fast rate. In the past, extracting this information was a very difficult task. But now, with the recent advancements in machine learning such as predictive models, it is easier. Predictive models can automatically search through large databases to produce a model by analyzing data and making predictions on what the data will do in the future. The models can then be applied in many types of decision making. Prediction provides the most important component of intelligent decision making: foresight. In today’s world of information and technology, it is this kind of information that is the key to market survival and improved efficiency. An example of a successful application of machine learning is the autonomous vehicle. These vehicles use a large amount of real-time data to make decisions and predict what other vehicles on the road will do next.